Skip to content

"rsync for cloud storage" - Google Drive, S3, Dropbox, Backblaze B2, One Drive, Swift, Hubic, Wasabi, Google Cloud Storage, Azure Blob, Azure Files, Yandex Files

3.0 MB
Raw Blame

% rclone(1) User Manual % Nick Craig-Wood % Nov 06, 2026

NAME

rclone - manage files on cloud storage

SYNOPSIS

Usage:
  rclone [flags]
  rclone [command]

Available commands:
  about       Get quota information from the remote.
  archive     Perform an action on an archive.
  authorize   Remote authorization.
  backend     Run a backend-specific command.
  bisync      Perform bidirectional synchronization between two paths.
  cat         Concatenates any files and sends them to stdout.
  check       Checks the files in the source and destination match.
  checksum    Checks the files in the destination against a SUM file.
  cleanup     Clean up the remote if possible.
  completion  Output completion script for a given shell.
  config      Enter an interactive configuration session.
  convmv      Convert file and directory names in place.
  copy        Copy files from source to dest, skipping identical files.
  copyto      Copy files from source to dest, skipping identical files.
  copyurl     Copy the contents of the URL supplied content to dest:path.
  cryptcheck  Cryptcheck checks the integrity of an encrypted remote.
  cryptdecode Cryptdecode returns unencrypted file names.
  dedupe      Interactively find duplicate filenames and delete/rename them.
  delete      Remove the files in path.
  deletefile  Remove a single file from remote.
  gendocs     Output markdown docs for rclone to the directory supplied.
  gitannex    Speaks with git-annex over stdin/stdout.
  gui         Open the web based GUI.
  hashsum     Produces a hashsum file for all the objects in the path.
  help        Show help for rclone commands, flags and backends.
  index       Write static directory listings into a remote.
  link        Generate public link to file/folder.
  listremotes List all the remotes in the config file and defined in environment variables.
  ls          List the objects in the path with size and path.
  lsd         List all directories/containers/buckets in the path.
  lsf         List directories and objects in remote:path formatted for parsing.
  lsjson      List directories and objects in the path in JSON format.
  lsl         List the objects in path with modification time, size and path.
  md5sum      Produces an md5sum file for all the objects in the path.
  mkdir       Make the path if it doesn't already exist.
  mount       Mount the remote as file system on a mountpoint.
  move        Move files from source to dest.
  moveto      Move file or directory from source to dest.
  ncdu        Explore a remote with a text based user interface.
  nfsmount    Mount the remote as file system on a mountpoint.
  obscure     Obscure password for use in the rclone config file.
  purge       Remove the path and all of its contents.
  rc          Run a command against a running rclone.
  rcat        Copies standard input to file on remote.
  rcd         Run rclone listening to remote control commands only.
  rmdir       Remove the empty directory at path.
  rmdirs      Remove empty directories under the path.
  selfupdate  Update the rclone binary.
  serve       Serve a remote over a protocol.
  settier     Changes storage class/tier of objects in remote.
  sha1sum     Produces an sha1sum file for all the objects in the path.
  size        Prints the total size and number of objects in remote:path.
  sync        Make source and dest identical, modifying destination only.
  test        Run a test command
  touch       Create new file or change file modification time.
  tree        List the contents of the remote in a tree like fashion.
  version     Show the version number.

Use "rclone [command] --help" for more information about a command.
Use "rclone help flags" for to see the global flags.
Use "rclone help backends" for a list of supported services.

Rclone syncs your files to cloud storage

rclone logo

About rclone {#about}

Rclone is a command-line program to manage files on cloud storage. It is a feature-rich alternative to cloud vendors' web storage interfaces. Over 70 cloud storage products support rclone including S3 object stores, business & consumer file storage services, as well as standard transfer protocols.

Rclone has powerful cloud equivalents to the unix commands rsync, cp, mv, mount, ls, ncdu, tree, rm, and cat. Rclone's familiar syntax includes shell pipeline support, and --dry-run protection. It is used at the command line, in scripts or via its API.

Users call rclone "The Swiss army knife of cloud storage", and "Technology indistinguishable from magic".

Rclone really looks after your data. It preserves timestamps and verifies checksums at all times. Transfers over limited bandwidth; intermittent connections, or subject to quota can be restarted, from the last good file transferred. You can check the integrity of your files. Where possible, rclone employs server-side transfers to minimise local bandwidth use and transfers from one provider to another without using local disk.

Virtual backends wrap local and cloud file systems to apply encryption, compression, chunking, hashing and joining.

Rclone mounts any local, cloud or virtual filesystem as a disk on Windows, macOS, linux and FreeBSD, and also serves these over SFTP, HTTP, WebDAV, FTP and DLNA.

Rclone is mature, open-source software originally inspired by rsync and written in Go. The friendly support community is familiar with varied use cases. Official Ubuntu, Debian, Fedora, Brew and Chocolatey repos. include rclone. For the latest version downloading from rclone.org is recommended.

Rclone is widely used on Linux, Windows and Mac. Third-party developers create innovative backup, restore, GUI and business process solutions using the rclone command line or API.

Rclone does the heavy lifting of communicating with cloud storage.

What can rclone do for you? {#what}

Rclone helps you:

  • Backup (and encrypt) files to cloud storage
  • Restore (and decrypt) files from cloud storage
  • Mirror cloud data to other cloud services or locally
  • Migrate data to the cloud, or between cloud storage vendors
  • Mount multiple, encrypted, cached or diverse cloud storage as a disk
  • Analyse and account for data held on cloud storage using lsf, ljson, size, ncdu
  • Union file systems together to present multiple local and/or cloud file systems as one

Features {#features}

  • Transfers
    • MD5, SHA1 hashes are checked at all times for file integrity
    • Timestamps are preserved on files
    • Operations can be restarted at any time
    • Can be to and from network, e.g. two different cloud providers
    • Can use multi-threaded downloads to local disk
  • Copy new or changed files to cloud storage
  • Sync (one way) to make a directory identical
  • Bisync (two way) to keep two directories in sync bidirectionally
  • Move files to cloud storage deleting the local after verification
  • Check hashes and for missing/extra files
  • Mount your cloud storage as a network disk
  • Serve local or remote files over HTTP/WebDav/FTP/SFTP/DLNA
  • Experimental Web based GUI

Supported providers {#providers}

(There are many others, built on standard protocols such as WebDAV or S3, that work out of the box.)

  • 1Fichier
  • Akamai Netstorage
  • Alibaba Cloud (Aliyun) Object Storage System (OSS)
  • Amazon S3
  • Autonomys Auto Drive
  • Bizfly Cloud Simple Storage
  • Backblaze B2
  • Box
  • Ceph
  • China Mobile Ecloud Elastic Object Storage (EOS)
  • Arvan Cloud Object Storage (AOS)
  • Citrix ShareFile
  • Clever Cloud Cellar
  • Cloudflare R2
  • Cloudinary
  • Cubbit DS3
  • DigitalOcean Spaces
  • Digi Storage
  • dosya.dev
  • Dreamhost
  • Drime
  • Dropbox
  • Enterprise File Fabric
  • Exaba
  • Fastly Object Storage
  • Fastmail Files
  • FileLu Cloud Storage
  • FileLu S5 (S3-Compatible Object Storage)
  • Filen
  • Files.com
  • FlashBlade
  • FTP
  • Gofile
  • Google Cloud Storage
  • Google Drive
  • Google Photos
  • HDFS
  • Hetzner Object Storage
  • Hetzner Storage Box
  • HiDrive
  • Hitachi Content Platform (HCP)
  • HTTP
  • Huawei Drive
  • Huawei OBS
  • iCloud Drive
  • ImageKit
  • Internet Archive
  • Internxt
  • Jottacloud
  • IBM COS S3
  • IDrive e2
  • Impossible Cloud
  • Infomaniak Swiss Backup (S3)
  • Infomaniak Swiss Backup (Swift)
  • Intercolo Object Storage
  • IONOS Cloud
  • kDrive
  • Koofr
  • Leviia Object Storage
  • Liara Object Storage
  • Linkbox
  • Linode Object Storage
  • Magalu
  • Mail.ru Cloud
  • Memset Memstore
  • MEGA
  • MEGA S4
  • Memory
  • Microsoft Azure Blob Storage
  • Microsoft Azure Files Storage
  • Microsoft OneDrive
  • Minio
  • Neon Object Storage
  • Nextcloud
  • Blomp Cloud Storage
  • OpenDrive
  • OpenStack Swift
  • Oracle Cloud Storage Swift
  • Oracle Object Storage
  • Outscale
  • OVHcloud Object Storage (Swift)
  • OVHcloud Object Storage (S3-compatible)
  • ownCloud
  • pCloud
  • Petabox
  • PikPak
  • Pixeldrain
  • premiumize.me
  • put.io
  • Proton Drive
  • QingStor
  • Qiniu Cloud Object Storage (Kodo)
  • Quatrix by Maytech
  • Rabata Cloud Storage
  • RackCorp Object Storage
  • Rackspace Cloud Files
  • Rclone Serve S3
  • RelAix.Storage
  • rsync.net
  • Scaleway
  • Scality (RING / ARTESCA)
  • Seafile
  • Seagate Lyve Cloud
  • SeaweedFS
  • Selectel
  • Servercore Object Storage
  • SFTP
  • Shade
  • Sia
  • SMB / CIFS
  • Spectra Logic
  • Storj
  • Synology
  • SugarSync
  • Tencent Cloud Object Storage (COS)
  • Uloz.to
  • US3
  • Wasabi
  • WebDAV
  • Yandex Disk
  • Zadara Object Storage
  • Zero Services (ZERO-Z3)
  • Zoho WorkDrive
  • Zata
  • The local filesystem

Virtual providers

These backends adapt or modify other storage providers:

  • Alias: Rename existing remotes
  • Archive: Read archive files
  • Cache: Cache remotes (DEPRECATED)
  • Chunker: Split large files
  • Combine: Combine multiple remotes into a directory tree
  • Compress: Compress files
  • Crypt: Encrypt files
  • Hasher: Hash files
  • Union: Join multiple remotes to work together

Links

Install

Rclone is a Go program and comes as a single binary file.

Quickstart

See below for some expanded Linux / macOS / Windows instructions.

See the usage docs for how to use rclone, or run rclone -h.

Already installed rclone can be easily updated to the latest version using the rclone selfupdate command.

See the release signing docs for how to verify signatures on the release.

Script installation

To install rclone on Linux/macOS/BSD systems, run:

sudo -v ; curl https://rclone.org/install.sh | sudo bash

For beta installation, run:

sudo -v ; curl https://rclone.org/install.sh | sudo bash -s beta

Note that this script checks the version of rclone installed first and won't re-download if not needed.

Linux installation {#linux}

Precompiled binary {#linux-precompiled}

Fetch and unpack

curl -O https://downloads.rclone.org/rclone-current-linux-amd64.zip
unzip rclone-current-linux-amd64.zip
cd rclone-*-linux-amd64

Copy binary file

sudo cp rclone /usr/bin/
sudo chown root:root /usr/bin/rclone
sudo chmod 755 /usr/bin/rclone

Install manpage

sudo mkdir -p /usr/local/share/man/man1
sudo cp rclone.1 /usr/local/share/man/man1/
sudo mandb

Run rclone config to setup. See rclone config docs for more details.

rclone config

macOS installation {#macos}

Installation with brew {#macos-brew}

brew install rclone

NOTE: This version of rclone will not support mount any more (see #5373). If mounting is wanted on macOS, either install a precompiled binary or enable the relevant option when installing from source.

Note that this is a third party installer not controlled by the rclone developers so it may be out of date. Its current version is as below.

Homebrew package

Installation with MacPorts {#macos-macports}

On macOS, rclone can also be installed via MacPorts:

sudo port install rclone

Note that this is a third party installer not controlled by the rclone developers so it may be out of date. Its current version is as below.

MacPorts port

More information on macports.org.

Precompiled binary, using curl {#macos-precompiled}

To avoid problems with macOS gatekeeper enforcing the binary to be signed and notarized it is enough to download with curl.

Download the latest version of rclone.

cd && curl -O https://downloads.rclone.org/rclone-current-osx-amd64.zip

Unzip the download and cd to the extracted folder.

unzip -a rclone-current-osx-amd64.zip && cd rclone-*-osx-amd64

Move rclone to your $PATH. You will be prompted for your password.

sudo mkdir -p /usr/local/bin
sudo mv rclone /usr/local/bin/

(the mkdir command is safe to run, even if the directory already exists).

Remove the leftover files.

cd .. && rm -rf rclone-*-osx-amd64 rclone-current-osx-amd64.zip

Run rclone config to setup. See rclone config docs for more details.

rclone config

Precompiled binary, using a web browser {#macos-precompiled-web}

When downloading a binary with a web browser, the browser will set the macOS gatekeeper quarantine attribute. Starting from Catalina, when attempting to run rclone, a pop-up will appear saying:

"rclone" cannot be opened because the developer cannot be verified.
macOS cannot verify that this app is free from malware.

The simplest fix is to run

xattr -d com.apple.quarantine rclone

Windows installation {#windows}

Precompiled binary {#windows-precompiled}

Fetch the correct binary for your processor type by clicking on these links. If not sure, use the first link.

Open this file in the Explorer and extract rclone.exe. Rclone is a portable executable so you can place it wherever is convenient.

Open a CMD window (or powershell) and run the binary. Note that rclone does not launch a GUI by default, it runs in the CMD Window.

If you are planning to use the rclone mount feature then you will need to install the third party utility WinFsp also.

Windows package manager (Winget) {#windows-chocolatey}

Winget comes pre-installed with the latest versions of Windows. If not, update the App Installer package from the Microsoft store.

To install rclone

winget install Rclone.Rclone

To uninstall rclone

winget uninstall Rclone.Rclone --force

Chocolatey package manager {#windows-chocolatey}

Make sure you have Choco installed

choco search rclone
choco install rclone

This will install rclone on your Windows machine. If you are planning to use rclone mount then

choco install winfsp

will install that too.

Note that this is a third party installer not controlled by the rclone developers so it may be out of date. Its current version is as below.

Chocolatey package

Scoop package manager {#windows-scoop}

Make sure you have Scoop installed

scoop install rclone

Note that this is a third party installer not controlled by the rclone developers so it may be out of date. Its current version is as below.

Scoop package

Package manager installation {#package-manager}

Many Linux, Windows, macOS and other OS distributions package and distribute rclone.

The distributed versions of rclone are often quite out of date and for this reason we recommend one of the other installation methods if possible.

You can get an idea of how up to date or not your OS distribution's package is here.

Packaging status

Docker installation {#docker}

The rclone developers maintain a docker image for rclone.

These images are built as part of the release process based on a minimal Alpine Linux.

The :latest tag will always point to the latest stable release. You can use the :beta tag to get the latest build from master. You can also use version tags, e.g. :1.49.1, :1.49 or :1.

$ docker pull rclone/rclone:latest
latest: Pulling from rclone/rclone
Digest: sha256:0e0ced72671989bb837fea8e88578b3fc48371aa45d209663683e24cfdaa0e11
...
$ docker run --rm rclone/rclone:latest version
rclone v1.49.1
- os/arch: linux/amd64
- go version: go1.12.9

There are a few command line options to consider when starting an rclone Docker container from the rclone image.

  • You need to mount the host rclone config dir at /config/rclone into the Docker container. Due to the fact that rclone updates tokens inside its config file, and that the update process involves a file rename, you need to mount the whole host rclone config dir, not just the single host rclone config file.

  • You need to mount a host data dir at /data into the Docker container.

  • By default, the rclone binary inside a Docker container runs with UID=0 (root). As a result, all files created in a run will have UID=0. If your config and data files reside on the host with a non-root UID:GID, you need to pass these on the container start command line.

  • If you want to access the RC interface (either via the API or the Web UI), it is required to set the --rc-addr to :5572 in order to connect to it from outside the container. An explanation about why this is necessary can be found in an old pythonspeed.com article.

    • NOTE: Users running this container with the docker network set to host should probably set it to listen to localhost only, with 127.0.0.1:5572 as the value for --rc-addr
  • It is possible to use rclone mount inside a userspace Docker container, and expose the resulting fuse mount to the host. The exact docker run options to do that might vary slightly between hosts. See, e.g. the discussion in this thread.

    You also need to mount the host /etc/passwd and /etc/group for fuse to work inside the container.

Here are some commands tested on an Ubuntu 18.04.3 host:

# config on host at ~/.config/rclone/rclone.conf
# data on host at ~/data

# add a remote interactively
docker run --rm -it \
    --volume ~/.config/rclone:/config/rclone \
    --user $(id -u):$(id -g) \
    rclone/rclone \
    config

# make sure the config is ok by listing the remotes
docker run --rm \
    --volume ~/.config/rclone:/config/rclone \
    --user $(id -u):$(id -g) \
    rclone/rclone \
    listremotes

# perform mount inside Docker container, expose result to host
mkdir -p ~/data/mount
docker run --rm \
    --volume ~/.config/rclone:/config/rclone \
    --volume ~/data:/data:shared \
    --user $(id -u):$(id -g) \
    --volume /etc/passwd:/etc/passwd:ro --volume /etc/group:/etc/group:ro \
    --device /dev/fuse --cap-add SYS_ADMIN --security-opt apparmor:unconfined \
    rclone/rclone \
    mount dropbox:Photos /data/mount &
ls ~/data/mount
kill %1

Snap installation {#snap}

Get it from the Snap Store

Make sure you have Snapd installed

sudo snap install rclone

Due to the strict confinement of Snap, rclone snap cannot access real /home/$USER/.config/rclone directory, default config path is as below.

  • Default config directory:
    • /home/$USER/snap/rclone/current/.config/rclone

Note: Due to the strict confinement of Snap, rclone mount feature is not supported.

If mounting is wanted, either install a precompiled binary or enable the relevant option when installing from source.

Note that this is controlled by community maintainer not the rclone developers so it may be out of date. Its current version is as below.

rclone

Source installation {#source}

Make sure you have git and Go installed. Go version 1.26 or newer is required, the latest release is recommended. You can get it from your package manager, or download it from golang.org/dl. Then you can run the following:

git clone https://github.com/rclone/rclone.git
cd rclone
go build

This will check out the rclone source in subfolder rclone, which you can later modify and send pull requests with. Then it will build the rclone executable in the same folder. As an initial check you can now run ./rclone version (.\rclone version on Windows).

Note that on macOS and Windows the mount command will not be available unless you specify an additional build tag cmount.

go build -tags cmount

This assumes you have a GCC compatible C compiler (GCC or Clang) in your PATH, as it uses cgo. But on Windows, the cgofuse library that the cmount implementation is based on, also supports building without cgo, i.e. by setting environment variable CGO_ENABLED to value 0 (static linking). This is how the official Windows release of rclone is being built, starting with version 1.59. It is still possible to build with cgo on Windows as well, by using the MinGW port of GCC, e.g. by installing it in a MSYS2 distribution (make sure you install it in the classic mingw64 subsystem, the ucrt64 version is not compatible).

Additionally, to build with mount on Windows, you must install the third party utility WinFsp, with the "Developer" feature selected. If building with cgo, you must also set environment variable CPATH pointing to the fuse include directory within the WinFsp installation (normally C:\Program Files (x86)\WinFsp\inc\fuse).

You may add arguments -ldflags -s to omit symbol table and debug information, making the executable file smaller, and -trimpath to remove references to local file system paths. The official rclone releases are built with both of these.

go build -trimpath -ldflags -s -tags cmount

If you want to customize the version string, as reported by the rclone version command, you can set one of the variables fs.Version, fs.VersionTag (to keep default suffix but customize the number), or fs.VersionSuffix (to keep default number but customize the suffix). This can be done from the build command, by adding to the -ldflags argument value as shown below.

go build -trimpath -ldflags "-s -X github.com/rclone/rclone/fs.Version=v9.9.9-test" -tags cmount

On Windows, the official executables also have the version information, as well as a file icon, embedded as binary resources. To get that with your own build you need to run the following command before the build command. It generates a Windows resource system object file, with extension .syso, e.g. resource_windows_amd64.syso, that will be automatically picked up by future build commands.

go run bin/resource_windows.go

The above command will generate a resource file containing version information based on the fs.Version variable in source at the time you run the command, which means if the value of this variable changes you need to re-run the command for it to be reflected in the version information. Also, if you override this version variable in the build command as described above, you need to do that also when generating the resource file, or else it will still use the value from the source.

go run bin/resource_windows.go -version v9.9.9-test

Instead of executing the go build command directly, you can run it via the Makefile. The default target changes the version suffix from "-DEV" to "-beta" followed by additional commit details, embeds version information binary resources on Windows, and copies the resulting rclone executable into your GOPATH bin folder ($(go env GOPATH)/bin, which corresponds to ~/go/bin/rclone by default).

make

To include mount command on macOS and Windows with Makefile build:

make GOTAGS=cmount

There are other make targets that can be used for more advanced builds, such as cross-compiling for all supported os/architectures, and packaging results into release artifacts. See Makefile and cross-compile.go for details.

Another alternative method for source installation is to download the source, build and install rclone - all in one operation, as a regular Go package. The source will be stored it in the Go module cache, and the resulting executable will be in your GOPATH bin folder ($(go env GOPATH)/bin, which corresponds to ~/go/bin/rclone by default).

go install github.com/rclone/rclone@latest

In some situations, rclone executable size might be too big for deployment in very restricted environments when all backends with large SDKs are included. To limit binary size unused backends can be commented out in backends/all/all.go and unused commands in cmd/all/all.go before building with go build or make

Ansible installation {#ansible}

This can be done with Stefan Weichinger's ansible role.

Instructions

  1. git clone https://github.com/stefangweichinger/ansible-rclone.git into your local roles-directory

  2. add the role to the hosts you want rclone installed to:

    - hosts: rclone-hosts
      roles:
        - rclone
    

Portable installation {#portable}

As mentioned above, rclone is single executable (rclone, or rclone.exe on Windows) that you can download as a zip archive and extract into a location of your choosing. When executing different commands, it may create files in different locations, such as a configuration file and various temporary files. By default the locations for these are according to your operating system, e.g. configuration file in your user profile directory and temporary files in the standard temporary directory, but you can customize all of them, e.g. to make a completely self-contained, portable installation.

Run the config paths command to see the locations that rclone will use.

To override them set the corresponding options (as command-line arguments, or as environment variables):

Autostart

After installing and configuring rclone, as described above, you are ready to use rclone as an interactive command line utility. If your goal is to perform periodic operations, such as a regular sync, you will probably want to configure your rclone command in your operating system's scheduler. If you need to expose service-like features, such as remote control, GUI, serve or mount, you will often want an rclone command always running in the background, and configuring it to run in a service infrastructure may be a better option. Below are some alternatives on how to achieve this on different operating systems.

NOTE: Before setting up autorun it is highly recommended that you have tested your command manually from a Command Prompt first.

Autostart on Windows

The most relevant alternatives for autostart on Windows are:

  • Run at user log on using the Startup folder
  • Run at user log on, at system startup or at schedule using Task Scheduler
  • Run at system startup using Windows service

Running in background

Rclone is a console application, so if not starting from an existing Command Prompt, e.g. when starting rclone.exe from a shortcut, it will open a Command Prompt window. When configuring rclone to run from task scheduler and windows service you are able to set it to run hidden in background. From rclone version 1.54 you can also make it run hidden from anywhere by adding option --no-console (it may still flash briefly when the program starts). Since rclone normally writes information and any error messages to the console, you must redirect this to a file to be able to see it. Rclone has a built-in option --log-file for that.

Example command to run a sync in background:

c:\rclone\rclone.exe sync c:\files remote:/files --no-console --log-file c:\rclone\logs\sync_files.txt

User account

As mentioned in the mount documentation, mounted drives created as Administrator are not visible to other accounts, not even the account that was elevated as Administrator. By running the mount command as the built-in SYSTEM user account, it will create drives accessible for everyone on the system. Both scheduled task and Windows service can be used to achieve this.

NOTE: Remember that when rclone runs as the SYSTEM user, the user profile that it sees will not be yours. This means that if you normally run rclone with configuration file in the default location, to be able to use the same configuration when running as the system user you must explicitly tell rclone where to find it with the --config option, or else it will look in the system users profile path (C:\Windows\System32\config\systemprofile). To test your command manually from a Command Prompt, you can run it with the PsExec utility from Microsoft's Sysinternals suite, which takes option -s to execute commands as the SYSTEM user.

Start from Startup folder

To quickly execute an rclone command you can simply create a standard Windows Explorer shortcut for the complete rclone command you want to run. If you store this shortcut in the special "Startup" start-menu folder, Windows will automatically run it at login. To open this folder in Windows Explorer, enter path %APPDATA%\Microsoft\Windows\Start Menu\Programs\Startup, or C:\ProgramData\Microsoft\Windows\Start Menu\Programs\StartUp if you want the command to start for every user that logs in.

This is the easiest approach to autostarting of rclone, but it offers no functionality to set it to run as different user, or to set conditions or actions on certain events. Setting up a scheduled task as described below will often give you better results.

Start from Task Scheduler

Task Scheduler is an administrative tool built into Windows, and it can be used to configure rclone to be started automatically in a highly configurable way, e.g. periodically on a schedule, on user log on, or at system startup. It can run be configured to run as the current user, or for a mount command that needs to be available to all users it can run as the SYSTEM user. For technical information, see Task Scheduler for developers.

Run as service

For running rclone at system startup, you can create a Windows service that executes your rclone command, as an alternative to scheduled task configured to run at startup.

Mount command built-in service integration

For mount commands, rclone has a built-in Windows service integration via the third-party WinFsp library it uses. Registering as a regular Windows service easy, as you just have to execute the built-in PowerShell command New-Service (requires administrative privileges).

Example of a PowerShell command that creates a Windows service for mounting some remote:/files as drive letter X:, for all users (service will be running as the local system account):

New-Service -Name Rclone -BinaryPathName 'c:\rclone\rclone.exe mount remote:/files X: --config c:\rclone\config\rclone.conf --log-file c:\rclone\logs\mount.txt'

The WinFsp service infrastructure supports incorporating services for file system implementations, such as rclone, into its own launcher service, as kind of "child services". This has the additional advantage that it also implements a network provider that integrates into Windows standard methods for managing network drives. This is currently not officially supported by Rclone, but with WinFsp version 2019.3 B2 / v1.5B2 or later it should be possible through path rewriting as described in #3340.

Third-party service integration

To Windows service running any rclone command, the excellent third-party utility NSSM, the "Non-Sucking Service Manager", can be used. It includes some advanced features such as adjusting process priority, defining process environment variables, redirect to file anything written to stdout, and customized response to different exit codes, with a GUI to configure everything from (although it can also be used from command line ).

There are also several other alternatives. To mention one more, WinSW, "Windows Service Wrapper", is worth checking out. It requires .NET Framework, but it is preinstalled on newer versions of Windows, and it also provides alternative standalone distributions which includes necessary runtime (.NET 5). WinSW is a command-line only utility, where you have to manually create an XML file with service configuration. This may be a drawback for some, but it can also be an advantage as it is easy to back up and reuse the configuration settings, without having go through manual steps in a GUI. One thing to note is that by default it does not restart the service on error, one have to explicit enable this in the configuration file (via the "onfailure" parameter).

Autostart on Linux

Start as a service

To always run rclone in background, relevant for mount commands etc, you can use systemd to set up rclone as a system or user service. Running as a system service ensures that it is run at startup even if the user it is running as has no active session. Running rclone as a user service ensures that it only starts after the configured user has logged into the system.

Run periodically from cron

To run a periodic command, such as a copy/sync, you can set up a cron job.

Usage

Rclone is a command line program to manage files on cloud storage. After download and install, continue here to learn how to use it: Initial configuration, what the basic syntax looks like, describes the various subcommands, the various options, and more.

Configure

First, you'll need to configure rclone. As the object storage systems have quite complicated authentication these are kept in a config file. (See the --config entry for how to find the config file and choose its location.)

The easiest way to make the config is to run rclone with the config option:

rclone config

See the following for detailed instructions for

Basic syntax

Rclone syncs a directory tree from one storage system to another.

Its syntax is like this

rclone subcommand [options] <parameters> <parameters...>

A subcommand is an rclone operation required (e.g. sync, copy, ls).

An option is a single letter flag (e.g. -v) or a group of single letter flags (e.g. -Pv) or a long flag (e.g. --progress). No options are required. Options can come after the subcommand or in between parameters too or on the end, but only global options can be used before the subcommand. Anything after a -- option will not be interpreted as an option so if you need to add a parameter which starts with a - then put a -- on its own first, eg

rclone lsf -- -directory-starting-with-dash

A parameter is usually a file path or rclone remote, eg /path/to/file or remote:path/to/file but it can be other things - the subcommand help will tell you what.

Source and destination paths are specified by the name you gave the storage system in the config file then the sub path, e.g. "drive:myfolder" to look at "myfolder" in Google drive.

You can define as many storage paths as you like in the config file.

Please use the --interactive/-i flag while learning rclone to avoid accidental data loss.

Subcommands

rclone uses a system of subcommands. For example

rclone ls remote:path # lists a remote
rclone copy /local/path remote:path # copies /local/path to the remote
rclone sync --interactive /local/path remote:path # syncs /local/path to the remote

rclone config

Enter an interactive configuration session.

Synopsis

Enter an interactive configuration session where you can setup new remotes and manage existing ones. You may also set or remove a password to protect your configuration.

rclone config [flags]

Options

  -h, --help   help for config

See the global flags page for global options not listed here.

See Also

rclone copy

Copy files from source to dest, skipping identical files.

Synopsis

Copy the source to the destination. Does not transfer files that are identical on source and destination, testing by size and modification time or MD5SUM. Doesn't delete files from the destination. If you want to also delete files from destination, to make it match source, use the sync command instead.

Note that it is always the contents of the directory that is synced, not the directory itself. So when source:path is a directory, it's the contents of source:path that are copied, not the directory name and contents.

To copy single files, use the copyto command instead.

If dest:path doesn't exist, it is created and the source:path contents go there.

For example

rclone copy source:sourcepath dest:destpath

Let's say there are two files in sourcepath

sourcepath/one.txt
sourcepath/two.txt

This copies them to

destpath/one.txt
destpath/two.txt

Not to

destpath/sourcepath/one.txt
destpath/sourcepath/two.txt

If you are familiar with rsync, rclone always works as if you had written a trailing / - meaning "copy the contents of this directory". This applies to all commands and whether you are talking about the source or destination.

See the --no-traverse option for controlling whether rclone lists the destination directory or not. Supplying this option when copying a small number of files into a large destination can speed transfers up greatly.

For example, if you have many files in /path/to/src but only a few of them change every day, you can copy all the files which have changed recently very efficiently like this:

rclone copy --max-age 24h --no-traverse /path/to/src remote:

Rclone will sync the modification times of files and directories if the backend supports it. If metadata syncing is required then use the --metadata flag.

Note that the modification time and metadata for the root directory will not be synced. See issue #7652 for more info.

Note: Use the -P/--progress flag to view real-time transfer statistics.

Note: Use the --dry-run or the --interactive/-i flag to test without copying anything.

Logger Flags

The --differ, --missing-on-dst, --missing-on-src, --match and --error flags write paths, one per line, to the file name (or stdout if it is -) supplied. What they write is described in the help below. For example --differ will write all paths which are present on both the source and destination but different.

The --combined flag will write a file (or stdout) which contains all file paths with a symbol and then a space and then the path to tell you what happened to it. These are reminiscent of diff files.

  • = path means path was found in source and destination and was identical
  • - path means path was missing on the source, so only in the destination
  • + path means path was missing on the destination, so only in the source
  • * path means path was present in source and destination but different.
  • ! path means there was an error reading or hashing the source or dest.

The --dest-after flag writes a list file using the same format flags as lsf (including customizable options for hash, modtime, etc.) Conceptually it is similar to rsync's --itemize-changes, but not identical -- it should output an accurate list of what will be on the destination after the command is finished.

When the --no-traverse flag is set, all logs involving files that exist only on the destination will be incomplete or completely missing.

Note that these logger flags have a few limitations, and certain scenarios are not currently supported:

  • --max-duration / CutoffModeHard
  • --compare-dest / --copy-dest
  • server-side moves of an entire dir at once
  • High-level retries, because there would be duplicates (use --retries 1 to disable)
  • Possibly some unusual error scenarios

Note also that each file is logged during execution, as opposed to after, so it is most useful as a predictor of what SHOULD happen to each file (which may or may not match what actually DID).

rclone copy source:path dest:path [flags]

Options

      --absolute                Put a leading / in front of path names
      --combined string         Make a combined report of changes to this file
      --create-empty-src-dirs   Create empty source dirs on destination after copy
      --csv                     Output in CSV format
      --dest-after string       Report all files that exist on the dest post-sync
      --differ string           Report all non-matching files to this file
  -d, --dir-slash               Append a slash to directory names (default true)
      --dirs-only               Only list directories
      --error string            Report all files with errors (hashing or reading) to this file
      --files-only              Only list files (default true)
  -F, --format string           Output format - see lsf help for details (default "p")
      --hash h                  Use this hash when h is used in the format MD5|SHA-1|DropboxHash (default "md5")
  -h, --help                    help for copy
      --match string            Report all matching files to this file
      --missing-on-dst string   Report all files missing from the destination to this file
      --missing-on-src string   Report all files missing from the source to this file
  -s, --separator string        Separator for the items in the format (default ";")
  -t, --timeformat string       Specify a custom time format - see docs for details (default: 2006-01-02 15:04:05)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Copy Options

Flags for anything which can copy a file

      --check-first                                 Do all the checks before starting transfers
  -c, --checksum                                    Check for changes with size & checksum (if available, or fallback to size only)
      --compare-dest stringArray                    Include additional server-side paths during comparison
      --copy-dest stringArray                       Implies --compare-dest but also copies files from paths into destination
      --cutoff-mode HARD|SOFT|CAUTIOUS              Mode to stop transfers when reaching the max transfer limit HARD|SOFT|CAUTIOUS (default HARD)
      --ignore-case-sync                            Ignore case when synchronizing
      --ignore-checksum                             Skip post copy check of checksums
      --ignore-existing                             Skip all files that exist on destination
      --ignore-size                                 Ignore size when skipping use modtime or checksum
  -I, --ignore-times                                Don't skip items that match size and time - transfer all unconditionally
      --immutable                                   Do not modify files, fail if existing files have been modified
      --inplace                                     Download directly to destination file instead of atomic download to temp/rename
  -l, --links                                       Translate symlinks to/from regular files with a '.rclonelink' extension
      --max-backlog int                             Maximum number of objects in sync or check backlog (default 10000)
      --max-duration Duration                       Maximum duration rclone will transfer data for (default 0s)
      --max-transfer SizeSuffix                     Maximum size of data to transfer (default off)
  -M, --metadata                                    If set, preserve metadata when copying objects
      --modify-window Duration                      Max time diff to be considered the same (default 1ns)
      --multi-thread-chunk-size SizeSuffix          Chunk size for multi-thread downloads / uploads, if not set by filesystem (default 64Mi)
      --multi-thread-cutoff SizeSuffix              Use multi-thread downloads for files above this size (default 256Mi)
      --multi-thread-streams int                    Number of streams to use for multi-thread downloads (default 4)
      --multi-thread-write-buffer-size SizeSuffix   In memory buffer size for writing when in multi-thread mode (default 128Ki)
      --name-transform stringArray                  Transform paths during the copy process
      --no-check-dest                               Don't check the destination, copy regardless
      --no-traverse                                 Don't traverse destination file system on copy
      --no-update-dir-modtime                       Don't update directory modification times
      --no-update-modtime                           Don't update destination modtime if files identical
      --order-by string                             Instructions on how to order the transfers, e.g. 'size,descending'
      --partial-suffix string                       Add partial-suffix to temporary file name when --inplace is not used (default ".partial")
      --refresh-times                               Refresh the modtime of remote files
      --server-side-across-configs                  Allow server-side operations (e.g. copy) to work across different configs
      --size-only                                   Skip based on size only, not modtime or checksum
      --streaming-upload-cutoff SizeSuffix          Cutoff for switching to chunked upload if file size is unknown, upload starts after reaching cutoff or when file ends (default 100Ki)
  -u, --update                                      Skip files that are newer on the destination

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone sync

Make source and dest identical, modifying destination only.

Synopsis

Sync the source to the destination, changing the destination only. Doesn't transfer files that are identical on source and destination, testing by size and modification time or MD5SUM. Destination is updated to match source, including deleting files if necessary (except duplicate objects, see below). If you don't want to delete files from destination, use the copy command instead.

Important: Since this can cause data loss, test first with the --dry-run or the --interactive/i flag.

rclone sync --interactive SOURCE remote:DESTINATION

Files in the destination won't be deleted if there were any errors at any point. Duplicate objects (files with the same name, on those providers that support it) are not yet handled. Files that are excluded won't be deleted unless --delete-excluded is used. Symlinks won't be transferred or deleted from local file systems unless --links is used.

It is always the contents of the directory that is synced, not the directory itself. So when source:path is a directory, it's the contents of source:path that are copied, not the directory name and contents. See extended explanation in the copy command if unsure.

If dest:path doesn't exist, it is created and the source:path contents go there.

It is not possible to sync overlapping remotes. However, you may exclude the destination from the sync with a filter rule or by putting an exclude-if-present file inside the destination directory and sync to a destination that is inside the source directory.

Rclone will sync the modification times of files and directories if the backend supports it. If metadata syncing is required then use the --metadata flag.

Note that the modification time and metadata for the root directory will not be synced. See https://github.com/rclone/rclone/issues/7652 for more info.

Note: Use the -P/--progress flag to view real-time transfer statistics

Note: Use the rclone dedupe command to deal with "Duplicate object/directory found in source/destination - ignoring" errors. See this forum post for more info.

Logger Flags

The --differ, --missing-on-dst, --missing-on-src, --match and --error flags write paths, one per line, to the file name (or stdout if it is -) supplied. What they write is described in the help below. For example --differ will write all paths which are present on both the source and destination but different.

The --combined flag will write a file (or stdout) which contains all file paths with a symbol and then a space and then the path to tell you what happened to it. These are reminiscent of diff files.

  • = path means path was found in source and destination and was identical
  • - path means path was missing on the source, so only in the destination
  • + path means path was missing on the destination, so only in the source
  • * path means path was present in source and destination but different.
  • ! path means there was an error reading or hashing the source or dest.

The --dest-after flag writes a list file using the same format flags as lsf (including customizable options for hash, modtime, etc.) Conceptually it is similar to rsync's --itemize-changes, but not identical -- it should output an accurate list of what will be on the destination after the command is finished.

When the --no-traverse flag is set, all logs involving files that exist only on the destination will be incomplete or completely missing.

Note that these logger flags have a few limitations, and certain scenarios are not currently supported:

  • --max-duration / CutoffModeHard
  • --compare-dest / --copy-dest
  • server-side moves of an entire dir at once
  • High-level retries, because there would be duplicates (use --retries 1 to disable)
  • Possibly some unusual error scenarios

Note also that each file is logged during execution, as opposed to after, so it is most useful as a predictor of what SHOULD happen to each file (which may or may not match what actually DID).

rclone sync source:path dest:path [flags]

Options

      --absolute                Put a leading / in front of path names
      --combined string         Make a combined report of changes to this file
      --create-empty-src-dirs   Create empty source dirs on destination after sync
      --csv                     Output in CSV format
      --dest-after string       Report all files that exist on the dest post-sync
      --differ string           Report all non-matching files to this file
  -d, --dir-slash               Append a slash to directory names (default true)
      --dirs-only               Only list directories
      --error string            Report all files with errors (hashing or reading) to this file
      --files-only              Only list files (default true)
  -F, --format string           Output format - see lsf help for details (default "p")
      --hash h                  Use this hash when h is used in the format MD5|SHA-1|DropboxHash (default "md5")
  -h, --help                    help for sync
      --match string            Report all matching files to this file
      --missing-on-dst string   Report all files missing from the destination to this file
      --missing-on-src string   Report all files missing from the source to this file
  -s, --separator string        Separator for the items in the format (default ";")
  -t, --timeformat string       Specify a custom time format - see docs for details (default: 2006-01-02 15:04:05)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Copy Options

Flags for anything which can copy a file

      --check-first                                 Do all the checks before starting transfers
  -c, --checksum                                    Check for changes with size & checksum (if available, or fallback to size only)
      --compare-dest stringArray                    Include additional server-side paths during comparison
      --copy-dest stringArray                       Implies --compare-dest but also copies files from paths into destination
      --cutoff-mode HARD|SOFT|CAUTIOUS              Mode to stop transfers when reaching the max transfer limit HARD|SOFT|CAUTIOUS (default HARD)
      --ignore-case-sync                            Ignore case when synchronizing
      --ignore-checksum                             Skip post copy check of checksums
      --ignore-existing                             Skip all files that exist on destination
      --ignore-size                                 Ignore size when skipping use modtime or checksum
  -I, --ignore-times                                Don't skip items that match size and time - transfer all unconditionally
      --immutable                                   Do not modify files, fail if existing files have been modified
      --inplace                                     Download directly to destination file instead of atomic download to temp/rename
  -l, --links                                       Translate symlinks to/from regular files with a '.rclonelink' extension
      --max-backlog int                             Maximum number of objects in sync or check backlog (default 10000)
      --max-duration Duration                       Maximum duration rclone will transfer data for (default 0s)
      --max-transfer SizeSuffix                     Maximum size of data to transfer (default off)
  -M, --metadata                                    If set, preserve metadata when copying objects
      --modify-window Duration                      Max time diff to be considered the same (default 1ns)
      --multi-thread-chunk-size SizeSuffix          Chunk size for multi-thread downloads / uploads, if not set by filesystem (default 64Mi)
      --multi-thread-cutoff SizeSuffix              Use multi-thread downloads for files above this size (default 256Mi)
      --multi-thread-streams int                    Number of streams to use for multi-thread downloads (default 4)
      --multi-thread-write-buffer-size SizeSuffix   In memory buffer size for writing when in multi-thread mode (default 128Ki)
      --name-transform stringArray                  Transform paths during the copy process
      --no-check-dest                               Don't check the destination, copy regardless
      --no-traverse                                 Don't traverse destination file system on copy
      --no-update-dir-modtime                       Don't update directory modification times
      --no-update-modtime                           Don't update destination modtime if files identical
      --order-by string                             Instructions on how to order the transfers, e.g. 'size,descending'
      --partial-suffix string                       Add partial-suffix to temporary file name when --inplace is not used (default ".partial")
      --refresh-times                               Refresh the modtime of remote files
      --server-side-across-configs                  Allow server-side operations (e.g. copy) to work across different configs
      --size-only                                   Skip based on size only, not modtime or checksum
      --streaming-upload-cutoff SizeSuffix          Cutoff for switching to chunked upload if file size is unknown, upload starts after reaching cutoff or when file ends (default 100Ki)
  -u, --update                                      Skip files that are newer on the destination

Sync Options

Flags used for sync commands

      --backup-dir string               Make backups into hierarchy based in DIR
      --delete-after                    When synchronizing, delete files on destination after transferring (default)
      --delete-before                   When synchronizing, delete files on destination before transferring
      --delete-during                   When synchronizing, delete files during transfer
      --fix-case                        Force rename of case insensitive dest to match source
      --ignore-errors                   Delete even if there are I/O errors
      --list-cutoff int                 To save memory, sort directory listings on disk above this threshold (default 1000000)
      --max-delete int                  When synchronizing, limit the number of deletes (default -1)
      --max-delete-size SizeSuffix      When synchronizing, limit the total size of deletes (default off)
      --suffix string                   Suffix to add to changed files
      --suffix-keep-extension           Preserve the extension when using --suffix
      --track-renames                   When synchronizing, track file renames and do a server-side move if possible
      --track-renames-strategy string   Strategies to use when synchronizing using track-renames hash|modtime|leaf (default "hash")

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone move

Move files from source to dest.

Synopsis

Moves the contents of the source directory to the destination directory. Rclone will error if the source and destination overlap and the remote does not support a server-side directory move operation.

To move single files, use the moveto command instead.

If no filters are in use and if possible this will server-side move source:path into dest:path. After this source:path will no longer exist.

Otherwise for each file in source:path selected by the filters (if any) this will move it into dest:path. If possible a server-side move will be used, otherwise it will copy it (server-side if possible) into dest:path then delete the original (if no errors on copy) in source:path.

If you want to delete empty source directories after move, use the --delete-empty-src-dirs flag.

See the --no-traverse option for controlling whether rclone lists the destination directory or not. Supplying this option when moving a small number of files into a large destination can speed transfers up greatly.

Rclone will sync the modification times of files and directories if the backend supports it. If metadata syncing is required then use the --metadata flag.

Note that the modification time and metadata for the root directory will not be synced. See https://github.com/rclone/rclone/issues/7652 for more info.

Important: Since this can cause data loss, test first with the --dry-run or the --interactive/-i flag.

Note: Use the -P/--progress flag to view real-time transfer statistics.

Logger Flags

The --differ, --missing-on-dst, --missing-on-src, --match and --error flags write paths, one per line, to the file name (or stdout if it is -) supplied. What they write is described in the help below. For example --differ will write all paths which are present on both the source and destination but different.

The --combined flag will write a file (or stdout) which contains all file paths with a symbol and then a space and then the path to tell you what happened to it. These are reminiscent of diff files.

  • = path means path was found in source and destination and was identical
  • - path means path was missing on the source, so only in the destination
  • + path means path was missing on the destination, so only in the source
  • * path means path was present in source and destination but different.
  • ! path means there was an error reading or hashing the source or dest.

The --dest-after flag writes a list file using the same format flags as lsf (including customizable options for hash, modtime, etc.) Conceptually it is similar to rsync's --itemize-changes, but not identical -- it should output an accurate list of what will be on the destination after the command is finished.

When the --no-traverse flag is set, all logs involving files that exist only on the destination will be incomplete or completely missing.

Note that these logger flags have a few limitations, and certain scenarios are not currently supported:

  • --max-duration / CutoffModeHard
  • --compare-dest / --copy-dest
  • server-side moves of an entire dir at once
  • High-level retries, because there would be duplicates (use --retries 1 to disable)
  • Possibly some unusual error scenarios

Note also that each file is logged during execution, as opposed to after, so it is most useful as a predictor of what SHOULD happen to each file (which may or may not match what actually DID).

rclone move source:path dest:path [flags]

Options

      --absolute                Put a leading / in front of path names
      --combined string         Make a combined report of changes to this file
      --create-empty-src-dirs   Create empty source dirs on destination after move
      --csv                     Output in CSV format
      --delete-empty-src-dirs   Delete empty source dirs after move
      --dest-after string       Report all files that exist on the dest post-sync
      --differ string           Report all non-matching files to this file
  -d, --dir-slash               Append a slash to directory names (default true)
      --dirs-only               Only list directories
      --error string            Report all files with errors (hashing or reading) to this file
      --files-only              Only list files (default true)
  -F, --format string           Output format - see lsf help for details (default "p")
      --hash h                  Use this hash when h is used in the format MD5|SHA-1|DropboxHash (default "md5")
  -h, --help                    help for move
      --match string            Report all matching files to this file
      --missing-on-dst string   Report all files missing from the destination to this file
      --missing-on-src string   Report all files missing from the source to this file
  -s, --separator string        Separator for the items in the format (default ";")
  -t, --timeformat string       Specify a custom time format - see docs for details (default: 2006-01-02 15:04:05)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Copy Options

Flags for anything which can copy a file

      --check-first                                 Do all the checks before starting transfers
  -c, --checksum                                    Check for changes with size & checksum (if available, or fallback to size only)
      --compare-dest stringArray                    Include additional server-side paths during comparison
      --copy-dest stringArray                       Implies --compare-dest but also copies files from paths into destination
      --cutoff-mode HARD|SOFT|CAUTIOUS              Mode to stop transfers when reaching the max transfer limit HARD|SOFT|CAUTIOUS (default HARD)
      --ignore-case-sync                            Ignore case when synchronizing
      --ignore-checksum                             Skip post copy check of checksums
      --ignore-existing                             Skip all files that exist on destination
      --ignore-size                                 Ignore size when skipping use modtime or checksum
  -I, --ignore-times                                Don't skip items that match size and time - transfer all unconditionally
      --immutable                                   Do not modify files, fail if existing files have been modified
      --inplace                                     Download directly to destination file instead of atomic download to temp/rename
  -l, --links                                       Translate symlinks to/from regular files with a '.rclonelink' extension
      --max-backlog int                             Maximum number of objects in sync or check backlog (default 10000)
      --max-duration Duration                       Maximum duration rclone will transfer data for (default 0s)
      --max-transfer SizeSuffix                     Maximum size of data to transfer (default off)
  -M, --metadata                                    If set, preserve metadata when copying objects
      --modify-window Duration                      Max time diff to be considered the same (default 1ns)
      --multi-thread-chunk-size SizeSuffix          Chunk size for multi-thread downloads / uploads, if not set by filesystem (default 64Mi)
      --multi-thread-cutoff SizeSuffix              Use multi-thread downloads for files above this size (default 256Mi)
      --multi-thread-streams int                    Number of streams to use for multi-thread downloads (default 4)
      --multi-thread-write-buffer-size SizeSuffix   In memory buffer size for writing when in multi-thread mode (default 128Ki)
      --name-transform stringArray                  Transform paths during the copy process
      --no-check-dest                               Don't check the destination, copy regardless
      --no-traverse                                 Don't traverse destination file system on copy
      --no-update-dir-modtime                       Don't update directory modification times
      --no-update-modtime                           Don't update destination modtime if files identical
      --order-by string                             Instructions on how to order the transfers, e.g. 'size,descending'
      --partial-suffix string                       Add partial-suffix to temporary file name when --inplace is not used (default ".partial")
      --refresh-times                               Refresh the modtime of remote files
      --server-side-across-configs                  Allow server-side operations (e.g. copy) to work across different configs
      --size-only                                   Skip based on size only, not modtime or checksum
      --streaming-upload-cutoff SizeSuffix          Cutoff for switching to chunked upload if file size is unknown, upload starts after reaching cutoff or when file ends (default 100Ki)
  -u, --update                                      Skip files that are newer on the destination

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone delete

Remove the files in path.

Synopsis

Remove the files in path. Unlike purge it obeys include/exclude filters so can be used to selectively delete files.

rclone delete only deletes files but leaves the directory structure alone. If you want to delete a directory and all of its contents use the purge command.

If you supply the --rmdirs flag, it will remove all empty directories along with it. You can also use the separate command rmdir or rmdirs to delete empty directories only.

For example, to delete all files bigger than 100 MiB, you may first want to check what would be deleted (use either):

rclone --min-size 100M lsl remote:path
rclone --dry-run --min-size 100M delete remote:path

Then proceed with the actual delete:

rclone --min-size 100M delete remote:path

That reads "delete everything with a minimum size of 100 MiB", hence delete all files bigger than 100 MiB.

Important: Since this can cause data loss, test first with the --dry-run or the --interactive/-i flag.

rclone delete remote:path [flags]

Options

  -h, --help     help for delete
      --rmdirs   rmdirs removes empty directories but leaves root intact

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone purge

Remove the path and all of its contents.

Synopsis

Remove the path and all of its contents. Note that this does not obey include/exclude filters - everything will be removed. Use the delete command if you want to selectively delete files. To delete empty directories only, use command rmdir or rmdirs.

The concurrency of this operation is controlled by the --checkers global flag. However, some backends will implement this command directly, in which case --checkers will be ignored.

Important: Since this can cause data loss, test first with the --dry-run or the --interactive/-i flag.

rclone purge remote:path [flags]

Options

  -h, --help   help for purge

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone mkdir

Make the path if it doesn't already exist.

rclone mkdir remote:path [flags]

Options

  -h, --help   help for mkdir

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone rmdir

Remove the empty directory at path.

Synopsis

This removes empty directory given by path. Will not remove the path if it has any objects in it, not even empty subdirectories. Use command rmdirs (or delete with option --rmdirs) to do that.

To delete a path and any objects in it, use purge command.

rclone rmdir remote:path [flags]

Options

  -h, --help   help for rmdir

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone check

Checks the files in the source and destination match.

Synopsis

Checks the files in the source and destination match. It compares sizes and hashes (MD5 or SHA1) and logs a report of files that don't match. It doesn't alter the source or destination.

For the crypt remote there is a dedicated command, cryptcheck, that are able to check the checksums of the encrypted files.

If you supply the --size-only flag, it will only compare the sizes not the hashes as well. Use this for a quick check.

If you supply the --download flag, it will download the data from both remotes and check them against each other on the fly. This can be useful for remotes that don't support hashes or if you really want to check all the data.

If you supply the --checkfile HASH flag with a valid hash name, the source:path must point to a text file in the SUM format.

If you supply the --one-way flag, it will only check that files in the source match the files in the destination, not the other way around. This means that extra files in the destination that are not in the source will not be detected.

The --differ, --missing-on-dst, --missing-on-src, --match and --error flags write paths, one per line, to the file name (or stdout if it is -) supplied. What they write is described in the help below. For example --differ will write all paths which are present on both the source and destination but different.

The --combined flag will write a file (or stdout) which contains all file paths with a symbol and then a space and then the path to tell you what happened to it. These are reminiscent of diff files.

  • = path means path was found in source and destination and was identical
  • - path means path was missing on the source, so only in the destination
  • + path means path was missing on the destination, so only in the source
  • * path means path was present in source and destination but different.
  • ! path means there was an error reading or hashing the source or dest.

The default number of parallel checks is 8. See the --checkers option for more information.

rclone check source:path dest:path [flags]

Options

  -C, --checkfile string        Treat source:path as a SUM file with hashes of given type
      --combined string         Make a combined report of changes to this file
      --differ string           Report all non-matching files to this file
      --download                Check by downloading rather than with hash
      --error string            Report all files with errors (hashing or reading) to this file
  -h, --help                    help for check
      --match string            Report all matching files to this file
      --missing-on-dst string   Report all files missing from the destination to this file
      --missing-on-src string   Report all files missing from the source to this file
      --one-way                 Check one way only, source files must exist on remote

Options shared with other commands are described next. See the global flags page for global options not listed here.

Check Options

Flags used for check commands

      --max-backlog int   Maximum number of objects in sync or check backlog (default 10000)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone ls

List the objects in the path with size and path.

Synopsis

Lists the objects in the source path to standard output in a human readable format with size and path. Recurses by default.

E.g.

$ rclone ls swift:bucket
    60295 bevajer5jef
    90613 canole
    94467 diwogej7
    37600 fubuwic

Any of the filtering options can be applied to this command.

There are several related list commands

  • ls to list size and path of objects only
  • lsl to list modification time, size and path of objects only
  • lsd to list directories only
  • lsf to list objects and directories in easy to parse format
  • lsjson to list objects and directories in JSON format

ls,lsl,lsd are designed to be human-readable. lsf is designed to be human and machine-readable. lsjson is designed to be machine-readable.

Note that ls and lsl recurse by default - use --max-depth 1 to stop the recursion.

The other list commands lsd,lsf,lsjson do not recurse by default - use -R to make them recurse.

On the bucket-based remotes (e.g. s3, gcs, b2) a directory is a key prefix rather than an object, so a listing without -R shows each prefix as one directory entry and not the objects stored under it.

List commands prefer a recursive method that uses more memory but fewer transactions by default. Use --disable ListR to suppress the behavior. See --fast-list for more details.

Listing a nonexistent directory will produce an error except for remotes which can't have empty directories (e.g. s3, swift, or gcs - the bucket-based remotes).

rclone ls remote:path [flags]

Options

  -h, --help   help for ls

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone lsd

List all directories/containers/buckets in the path.

Synopsis

Lists the directories in the source path to standard output. Does not recurse by default. Use the -R flag to recurse.

This command lists the total size of the directory (if known, -1 if not), the modification time (if known, the --default-time if not), the number of objects in the directory (if known, -1 if not) and the name of the directory, E.g.

$ rclone lsd swift:
      494000 2018-04-26 08:43:20     10000 10000files
          65 2018-04-26 08:43:20         1 1File

Or

$ rclone lsd drive:test
          -1 2016-10-17 17:41:53        -1 1000files
          -1 2017-01-03 14:40:54        -1 2500files
          -1 2017-07-08 14:39:28        -1 4000files

If you just want the directory names use rclone lsf --dirs-only.

Any of the filtering options can be applied to this command.

There are several related list commands

  • ls to list size and path of objects only
  • lsl to list modification time, size and path of objects only
  • lsd to list directories only
  • lsf to list objects and directories in easy to parse format
  • lsjson to list objects and directories in JSON format

ls,lsl,lsd are designed to be human-readable. lsf is designed to be human and machine-readable. lsjson is designed to be machine-readable.

Note that ls and lsl recurse by default - use --max-depth 1 to stop the recursion.

The other list commands lsd,lsf,lsjson do not recurse by default - use -R to make them recurse.

On the bucket-based remotes (e.g. s3, gcs, b2) a directory is a key prefix rather than an object, so a listing without -R shows each prefix as one directory entry and not the objects stored under it.

List commands prefer a recursive method that uses more memory but fewer transactions by default. Use --disable ListR to suppress the behavior. See --fast-list for more details.

Listing a nonexistent directory will produce an error except for remotes which can't have empty directories (e.g. s3, swift, or gcs - the bucket-based remotes).

rclone lsd remote:path [flags]

Options

  -h, --help        help for lsd
  -R, --recursive   Recurse into the listing

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone lsl

List the objects in path with modification time, size and path.

Synopsis

Lists the objects in the source path to standard output in a human readable format with modification time, size and path. Recurses by default.

E.g.

$ rclone lsl swift:bucket
    60295 2016-06-25 18:55:41.062626927 bevajer5jef
    90613 2016-06-25 18:55:43.302607074 canole
    94467 2016-06-25 18:55:43.046609333 diwogej7
    37600 2016-06-25 18:55:40.814629136 fubuwic

Any of the filtering options can be applied to this command.

There are several related list commands

  • ls to list size and path of objects only
  • lsl to list modification time, size and path of objects only
  • lsd to list directories only
  • lsf to list objects and directories in easy to parse format
  • lsjson to list objects and directories in JSON format

ls,lsl,lsd are designed to be human-readable. lsf is designed to be human and machine-readable. lsjson is designed to be machine-readable.

Note that ls and lsl recurse by default - use --max-depth 1 to stop the recursion.

The other list commands lsd,lsf,lsjson do not recurse by default - use -R to make them recurse.

On the bucket-based remotes (e.g. s3, gcs, b2) a directory is a key prefix rather than an object, so a listing without -R shows each prefix as one directory entry and not the objects stored under it.

List commands prefer a recursive method that uses more memory but fewer transactions by default. Use --disable ListR to suppress the behavior. See --fast-list for more details.

Listing a nonexistent directory will produce an error except for remotes which can't have empty directories (e.g. s3, swift, or gcs - the bucket-based remotes).

rclone lsl remote:path [flags]

Options

  -h, --help   help for lsl

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone md5sum

Produces an md5sum file for all the objects in the path.

Synopsis

Produces an md5sum file for all the objects in the path. This is in the same format as the standard md5sum tool produces.

By default, the hash is requested from the remote. If MD5 is not supported by the remote, no hash will be returned. With the download flag, the file will be downloaded from the remote and hashed locally enabling MD5 for any remote.

For other algorithms, see the hashsum command. Running rclone md5sum remote:path is equivalent to running rclone hashsum MD5 remote:path.

This command can also hash data received on standard input (stdin), by not passing a remote:path, or by passing a hyphen as remote:path when there is data to read (if not, the hyphen will be treated literally, as a relative path).

rclone md5sum remote:path [flags]

Options

      --base64               Output base64 encoded hashsum
  -C, --checkfile string     Validate hashes against a given SUM file instead of printing them
      --download             Download the file and hash it locally; if this flag is not specified, the hash is requested from the remote
  -h, --help                 help for md5sum
      --output-file string   Output hashsums to a file rather than the terminal

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone sha1sum

Produces an sha1sum file for all the objects in the path.

Synopsis

Produces an sha1sum file for all the objects in the path. This is in the same format as the standard sha1sum tool produces.

By default, the hash is requested from the remote. If SHA-1 is not supported by the remote, no hash will be returned. With the download flag, the file will be downloaded from the remote and hashed locally enabling SHA-1 for any remote.

For other algorithms, see the hashsum command. Running rclone sha1sum remote:path is equivalent to running rclone hashsum SHA1 remote:path.

This command can also hash data received on standard input (stdin), by not passing a remote:path, or by passing a hyphen as remote:path when there is data to read (if not, the hyphen will be treated literally, as a relative path).

This command can also hash data received on STDIN, if not passing a remote:path.

rclone sha1sum remote:path [flags]

Options

      --base64               Output base64 encoded hashsum
  -C, --checkfile string     Validate hashes against a given SUM file instead of printing them
      --download             Download the file and hash it locally; if this flag is not specified, the hash is requested from the remote
  -h, --help                 help for sha1sum
      --output-file string   Output hashsums to a file rather than the terminal

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone size

Prints the total size and number of objects in remote:path.

Synopsis

Counts objects in the path and calculates the total size. Prints the result to standard output.

By default the output is in human-readable format, but shows values in both human-readable format as well as the raw numbers (global option --human-readable is not considered). Use option --json to format output as JSON instead.

Recurses by default, use --max-depth 1 to stop the recursion.

Some backends do not always provide file sizes, see for example Google Photos and Google Docs. Rclone will then show a notice in the log indicating how many such files were encountered, and count them in as empty files in the output of the size command.

rclone size remote:path [flags]

Options

  -h, --help   help for size
      --json   Format output as JSON

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone version

Show the version number.

Synopsis

Show the rclone version number, the go version, the build target OS and architecture, the runtime OS and kernel version and bitness, build tags and the type of executable (static or dynamic).

For example:

$ rclone version
rclone v1.55.0
- os/version: ubuntu 18.04 (64 bit)
- os/kernel: 4.15.0-136-generic (x86_64)
- os/type: linux
- os/arch: amd64
- go/version: go1.16
- go/linking: static
- go/tags: none

Note: before rclone version 1.55 the os/type and os/arch lines were merged, and the "go/version" line was tagged as "go version".

If you supply the --check flag, then it will do an online check to compare your version with the latest release and the latest beta.

$ rclone version --check
yours:  1.42.0.6
latest: 1.42          (released 2018-06-16)
beta:   1.42.0.5      (released 2018-06-17)

Or

$ rclone version --check
yours:  1.41
latest: 1.42          (released 2018-06-16)
  upgrade: https://downloads.rclone.org/v1.42
beta:   1.42.0.5      (released 2018-06-17)
  upgrade: https://beta.rclone.org/v1.42-005-g56e1e820

If you supply the --deps flag then rclone will print a list of all the packages it depends on and their versions along with some other information about the build.

rclone version [flags]

Options

      --check   Check for new version
      --deps    Show the Go dependencies
  -h, --help    help for version

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone cleanup

Clean up the remote if possible.

Synopsis

Clean up the remote if possible. Empty the trash or delete old file versions. Not supported by all remotes.

rclone cleanup remote:path [flags]

Options

  -h, --help   help for cleanup

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone dedupe

Interactively find duplicate filenames and delete/rename them.

Synopsis

By default dedupe interactively finds files with duplicate names and offers to delete all but one or rename them to be different. This is known as deduping by name.

Deduping by name is only useful with a small group of backends (e.g. Google Drive, Opendrive) that can have duplicate file names. It can be run on wrapping backends (e.g. crypt) if they wrap a backend which supports duplicate file names.

However if --by-hash is passed in then dedupe will find files with duplicate hashes instead which will work on any backend which supports at least one hash. This can be used to find files with duplicate content. This is known as deduping by hash.

If deduping by name, first rclone will merge directories with the same name. It will do this iteratively until all the identically named directories have been merged.

Next, if deduping by name, for every group of duplicate file names / hashes, it will delete all but one identical file it finds without confirmation. This means that for most duplicated files the dedupe command will not be interactive.

dedupe considers files to be identical if they have the same file path and the same hash. If the backend does not support hashes (e.g. crypt wrapping Google Drive) then they will never be found to be identical. If you use the --size-only flag then files will be considered identical if they have the same size (any hash will be ignored). This can be useful on crypt backends which do not support hashes.

Next rclone will resolve the remaining duplicates. Exactly which action is taken depends on the dedupe mode. By default, rclone will interactively query the user for each one.

Important: Since this can cause data loss, test first with the --dry-run or the --interactive/-i flag.

Here is an example run.

Before - with duplicates

$ rclone lsl drive:dupes
  6048320 2016-03-05 16:23:16.798000000 one.txt
  6048320 2016-03-05 16:23:11.775000000 one.txt
   564374 2016-03-05 16:23:06.731000000 one.txt
  6048320 2016-03-05 16:18:26.092000000 one.txt
  6048320 2016-03-05 16:22:46.185000000 two.txt
  1744073 2016-03-05 16:22:38.104000000 two.txt
   564374 2016-03-05 16:22:52.118000000 two.txt

Now the dedupe session

$ rclone dedupe drive:dupes
2016/03/05 16:24:37 Google drive root 'dupes': Looking for duplicates using interactive mode.
one.txt: Found 4 files with duplicate names
one.txt: Deleting 2/3 identical duplicates (MD5 "1eedaa9fe86fd4b8632e2ac549403b36")
one.txt: 2 duplicates remain
  1:      6048320 bytes, 2016-03-05 16:23:16.798000000, MD5 1eedaa9fe86fd4b8632e2ac549403b36
  2:       564374 bytes, 2016-03-05 16:23:06.731000000, MD5 7594e7dc9fc28f727c42ee3e0749de81
s) Skip and do nothing
k) Keep just one (choose which in next step)
r) Rename all to be different (by changing file.jpg to file-1.jpg)
s/k/r> k
Enter the number of the file to keep> 1
one.txt: Deleted 1 extra copies
two.txt: Found 3 files with duplicate names
two.txt: 3 duplicates remain
  1:       564374 bytes, 2016-03-05 16:22:52.118000000, MD5 7594e7dc9fc28f727c42ee3e0749de81
  2:      6048320 bytes, 2016-03-05 16:22:46.185000000, MD5 1eedaa9fe86fd4b8632e2ac549403b36
  3:      1744073 bytes, 2016-03-05 16:22:38.104000000, MD5 851957f7fb6f0bc4ce76be966d336802
s) Skip and do nothing
k) Keep just one (choose which in next step)
r) Rename all to be different (by changing file.jpg to file-1.jpg)
s/k/r> r
two-1.txt: renamed from: two.txt
two-2.txt: renamed from: two.txt
two-3.txt: renamed from: two.txt

The result being

$ rclone lsl drive:dupes
  6048320 2016-03-05 16:23:16.798000000 one.txt
   564374 2016-03-05 16:22:52.118000000 two-1.txt
  6048320 2016-03-05 16:22:46.185000000 two-2.txt
  1744073 2016-03-05 16:22:38.104000000 two-3.txt

Dedupe can be run non interactively using the --dedupe-mode flag or by using an extra parameter with the same value

  • --dedupe-mode interactive - interactive as above.
  • --dedupe-mode skip - removes identical files then skips anything left.
  • --dedupe-mode first - removes identical files then keeps the first one.
  • --dedupe-mode newest - removes identical files then keeps the newest one.
  • --dedupe-mode oldest - removes identical files then keeps the oldest one.
  • --dedupe-mode largest - removes identical files then keeps the largest one.
  • --dedupe-mode smallest - removes identical files then keeps the smallest one.
  • --dedupe-mode rename - removes identical files then renames the rest to be different.
  • --dedupe-mode list - lists duplicate dirs and files only and changes nothing.

For example, to rename all the identically named photos in your Google Photos directory, do

rclone dedupe --dedupe-mode rename "drive:Google Photos"

Or

rclone dedupe rename "drive:Google Photos"
rclone dedupe [mode] remote:path [flags]

Options

      --by-hash              Find identical hashes rather than names
      --dedupe-mode string   Dedupe mode interactive|skip|first|newest|oldest|largest|smallest|rename (default "interactive")
  -h, --help                 help for dedupe

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone about

Get quota information from the remote.

Synopsis

Prints quota information about a remote to standard output. The output is typically used, free, quota and trash contents.

E.g. Typical output from rclone about remote: is:

Total:   17 GiB
Used:    7.444 GiB
Free:    1.315 GiB
Trashed: 100.000 MiB
Other:   8.241 GiB

Where the fields are:

  • Total: Total size available.
  • Used: Total size used.
  • Free: Total space available to this user.
  • Trashed: Total space used by trash.
  • Other: Total amount in other storage (e.g. Gmail, Google Photos).
  • Objects: Total number of objects in the storage.

All sizes are in number of bytes.

Applying a --full flag to the command prints the bytes in full, e.g.

Total:   18253611008
Used:    7993453766
Free:    1411001220
Trashed: 104857602
Other:   8849156022

A --json flag generates conveniently machine-readable output, e.g.

{
  "total": 18253611008,
  "used": 7993453766,
  "trashed": 104857602,
  "other": 8849156022,
  "free": 1411001220
}

Not all backends print all fields. Information is not included if it is not provided by a backend. Where the value is unlimited it is omitted.

Some backends does not support the rclone about command at all, see complete list in documentation.

rclone about remote: [flags]

Options

      --full   Full numbers instead of human-readable
  -h, --help   help for about
      --json   Format output as JSON

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone archive

Perform an action on an archive.

Synopsis

Perform an action on an archive. Requires the use of a subcommand to specify the protocol, e.g.

rclone archive list remote:file.zip

Each subcommand has its own options which you can see in their help.

See rclone archive create for the archive formats supported.

rclone archive <action> [opts] <source> [<destination>] [flags]

Options

  -h, --help   help for archive

See the global flags page for global options not listed here.

See Also

rclone archive create

Archive source file(s) to destination.

Synopsis

Creates an archive from the files in source:path and saves the archive to dest:path. If dest:path is missing, it will write to the console.

The valid formats for the --format flag are listed below. If --format is not set rclone will guess it from the extension of dest:path.

Format Extensions
zip .zip
tar .tar
tar.gz .tar.gz, .tgz, .taz
tar.bz2 .tar.bz2, .tb2, .tbz, .tbz2, .tz2
tar.lz .tar.lz
tar.lz4 .tar.lz4
tar.xz .tar.xz, .txz
tar.zst .tar.zst, .tzst
tar.br .tar.br
tar.sz .tar.sz
tar.mz .tar.mz

The --prefix and --full-path flags control the prefix for the files in the archive.

If the flag --full-path is set then the files will have the full source path as the prefix.

If the flag --prefix=<value> is set then the files will have <value> as prefix. It's possible to create invalid file names with --prefix=<value> so use with caution. Flag --prefix has priority over --full-path.

Given a directory /sourcedir with the following:

file1.txt
dir1/file2.txt

Running the command rclone archive create /sourcedir /dest.tar.gz will make an archive with the contents:

file1.txt
dir1/
dir1/file2.txt

Running the command rclone archive create --full-path /sourcedir /dest.tar.gz will make an archive with the contents:

sourcedir/file1.txt
sourcedir/dir1/
sourcedir/dir1/file2.txt

Running the command rclone archive create --prefix=my_new_path /sourcedir /dest.tar.gz will make an archive with the contents:

my_new_path/file1.txt
my_new_path/dir1/
my_new_path/dir1/file2.txt
rclone archive create [flags] <source> [<destination>]

Options

      --format string   Create the archive with format or guess from extension.
      --full-path       Set prefix for files in archive to source path
  -h, --help            help for create
      --prefix string   Set prefix for files in archive to entered value or source path

See the global flags page for global options not listed here.

See Also

rclone archive extract

Extract archives from source to destination.

Synopsis

Extract the archive contents to a destination directory auto detecting the format. See rclone archive create for the archive formats supported.

For example on this archive:

$ rclone archive list --long remote:archive.zip
        6 2025-10-30 09:46:23.000000000 file.txt
        0 2025-10-30 09:46:57.000000000 dir/
        4 2025-10-30 09:46:57.000000000 dir/bye.txt

You can run extract like this

$ rclone archive extract remote:archive.zip remote:extracted

Which gives this result

$ rclone tree remote:extracted
/
├── dir
│   └── bye.txt
└── file.txt

The source or destination or both can be local or remote.

Filters can be used to only extract certain files:

$ rclone archive extract archive.zip partial --include "bye.*"
$ rclone tree partial
/
└── dir
    └── bye.txt

The archive backend can also be used to extract files. It can be used to read only mount archives also but it supports a different set of archive formats to the archive commands.

rclone archive extract [flags] <source> <destination>

Options

  -h, --help   help for extract

See the global flags page for global options not listed here.

See Also

rclone archive list

List archive contents from source.

Synopsis

List the contents of an archive to the console, auto detecting the format. See rclone archive create for the archive formats supported.

For example:

$ rclone archive list remote:archive.zip
        6 file.txt
        0 dir/
        4 dir/bye.txt

Or with --long flag for more info:

$ rclone archive list --long remote:archive.zip
        6 2025-10-30 09:46:23.000000000 file.txt
        0 2025-10-30 09:46:57.000000000 dir/
        4 2025-10-30 09:46:57.000000000 dir/bye.txt

Or with --plain flag which is useful for scripting:

$ rclone archive list --plain /path/to/archive.zip
file.txt
dir/
dir/bye.txt

Or with --dirs-only:

$ rclone archive list --plain --dirs-only /path/to/archive.zip
dir/

Or with --files-only:

$ rclone archive list --plain --files-only /path/to/archive.zip
file.txt
dir/bye.txt

Filters may also be used:

$ rclone archive list --long archive.zip --include "bye.*"
        4 2025-10-30 09:46:57.000000000 dir/bye.txt

The archive backend can also be used to list files. It can be used to read only mount archives also but it supports a different set of archive formats to the archive commands.

rclone archive list [flags] <source>

Options

      --dirs-only    Only list directories
      --files-only   Only list files
  -h, --help         help for list
      --long         List extra attributtes
      --plain        Only list file names

See the global flags page for global options not listed here.

See Also

rclone authorize

Remote authorization.

Synopsis

Remote authorization. Used to authorize a remote or headless rclone from a machine with a browser. Use as instructed by rclone config. See also the remote setup documentation.

The command requires 1-3 arguments:

  • Name of a backend (e.g. "drive", "s3")
  • Either a base64 encoded JSON blob obtained from a previous rclone config session
  • Or a client_id and client_secret pair obtained from the remote service

Use --auth-no-open-browser to prevent rclone to open auth link in default browser automatically.

Use --template to generate HTML output via a custom Go template. If a blank string is provided as an argument to this flag, the default template is used.

rclone authorize <backendname> [base64_json_blob | client_id client_secret] [flags]

Options

      --auth-no-open-browser   Do not automatically open auth link in default browser
  -h, --help                   help for authorize
      --template string        The path to a custom Go template for generating HTML responses

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone backend

Run a backend-specific command.

Synopsis

This runs a backend-specific command. The commands themselves (except for "help" and "features") are defined by the backends and you should see the backend docs for definitions.

You can discover what commands a backend implements by using

rclone backend help remote:
rclone backend help <backendname>

You can also discover information about the backend using (see operations/fsinfo in the remote control docs for more info).

rclone backend features remote:

Pass options to the backend command with -o. This should be key=value or key, e.g.:

rclone backend stats remote:path stats -o format=json -o long

Pass arguments to the backend by placing them on the end of the line

rclone backend cleanup remote:path file1 file2 file3

Note to run these commands on a running backend then see backend/command in the rc docs.

rclone backend <command> remote:path [opts] <args> [flags]

Options

  -h, --help                 help for backend
      --json                 Always output in JSON format
  -o, --option stringArray   Option in the form name=value or name

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone bisync

Perform bidirectional synchronization between two paths.

Synopsis

Perform bidirectional synchronization between two paths.

Bisync provides a bidirectional cloud sync solution in rclone. It retains the Path1 and Path2 filesystem listings from the prior run. On each successive run it will:

  • list files on Path1 and Path2, and check for changes on each side. Changes include New, Newer, Older, and Deleted files.
  • Propagate changes on Path1 to Path2, and vice-versa.

Bisync is considered an advanced command, so use with care. Make sure you have read and understood the entire manual (especially the Limitations section) before using, or data loss can result. Questions can be asked in the Rclone Forum.

See full bisync description for details.

rclone bisync remote1:path1 remote2:path2 [flags]

Options

      --backup-dir1 string                   --backup-dir for Path1. Must be a non-overlapping path on the same remote.
      --backup-dir2 string                   --backup-dir for Path2. Must be a non-overlapping path on the same remote.
      --check-access                         Ensure expected RCLONE_TEST files are found on both Path1 and Path2 filesystems, else abort.
      --check-filename string                Filename for --check-access (default: RCLONE_TEST)
      --check-sync string                    Controls comparison of final listings: true|false|only (default: true) (default "true")
      --compare string                       Comma-separated list of bisync-specific compare options ex. 'size,modtime,checksum' (default: 'size,modtime')
      --conflict-loser ConflictLoserAction   Action to take on the loser of a sync conflict (when there is a winner) or on both files (when there is no winner): , num, pathname, delete (default: num)
      --conflict-resolve string              Automatically resolve conflicts by preferring the version that is: none, path1, path2, newer, older, larger, smaller (default: none) (default "none")
      --conflict-suffix string               Suffix to use when renaming a --conflict-loser. Can be either one string or two comma-separated strings to assign different suffixes to Path1/Path2. (default: 'conflict')
      --create-empty-src-dirs                Sync creation and deletion of empty directories. (Not compatible with --remove-empty-dirs)
      --download-hash                        Compute hash by downloading when otherwise unavailable. (warning: may be slow and use lots of data!)
      --filters-file string                  Read filtering patterns from a file
      --force                                Bypass --max-delete safety check and run the sync. Consider using with --verbose
  -h, --help                                 help for bisync
      --ignore-listing-checksum              Do not use checksums for listings (add --ignore-checksum to additionally skip post-copy checksum checks)
      --max-delete-renames-aware             Exclude tracked renames from the --max-delete safety check (requires --track-renames)
      --max-lock Duration                    Consider lock files older than this to be expired (default: 0 (never expire)) (minimum: 2m) (default 0s)
      --no-cleanup                           Retain working files (useful for troubleshooting and testing).
      --no-slow-hash                         Ignore listing checksums only on backends where they are slow
      --recover                              Automatically recover from interruptions without requiring --resync.
      --remove-empty-dirs                    Remove ALL empty directories at the final cleanup step.
      --resilient                            Allow future runs to retry after certain less-serious errors, instead of requiring --resync.
  -1, --resync                               Performs the resync run. Equivalent to --resync-mode path1. Consider using --verbose or --dry-run first.
      --resync-mode string                   During resync, prefer the version that is: path1, path2, newer, older, larger, smaller (default: path1 if --resync, otherwise none for no resync.) (default "none")
      --slow-hash-sync-only                  Ignore slow checksums for listings and deltas, but still consider them during sync calls.
      --workdir string                       Use custom working dir - useful for testing. (default: $HOME/.cache/rclone/bisync)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Copy Options

Flags for anything which can copy a file

      --check-first                                 Do all the checks before starting transfers
  -c, --checksum                                    Check for changes with size & checksum (if available, or fallback to size only)
      --compare-dest stringArray                    Include additional server-side paths during comparison
      --copy-dest stringArray                       Implies --compare-dest but also copies files from paths into destination
      --cutoff-mode HARD|SOFT|CAUTIOUS              Mode to stop transfers when reaching the max transfer limit HARD|SOFT|CAUTIOUS (default HARD)
      --ignore-case-sync                            Ignore case when synchronizing
      --ignore-checksum                             Skip post copy check of checksums
      --ignore-existing                             Skip all files that exist on destination
      --ignore-size                                 Ignore size when skipping use modtime or checksum
  -I, --ignore-times                                Don't skip items that match size and time - transfer all unconditionally
      --immutable                                   Do not modify files, fail if existing files have been modified
      --inplace                                     Download directly to destination file instead of atomic download to temp/rename
  -l, --links                                       Translate symlinks to/from regular files with a '.rclonelink' extension
      --max-backlog int                             Maximum number of objects in sync or check backlog (default 10000)
      --max-duration Duration                       Maximum duration rclone will transfer data for (default 0s)
      --max-transfer SizeSuffix                     Maximum size of data to transfer (default off)
  -M, --metadata                                    If set, preserve metadata when copying objects
      --modify-window Duration                      Max time diff to be considered the same (default 1ns)
      --multi-thread-chunk-size SizeSuffix          Chunk size for multi-thread downloads / uploads, if not set by filesystem (default 64Mi)
      --multi-thread-cutoff SizeSuffix              Use multi-thread downloads for files above this size (default 256Mi)
      --multi-thread-streams int                    Number of streams to use for multi-thread downloads (default 4)
      --multi-thread-write-buffer-size SizeSuffix   In memory buffer size for writing when in multi-thread mode (default 128Ki)
      --name-transform stringArray                  Transform paths during the copy process
      --no-check-dest                               Don't check the destination, copy regardless
      --no-traverse                                 Don't traverse destination file system on copy
      --no-update-dir-modtime                       Don't update directory modification times
      --no-update-modtime                           Don't update destination modtime if files identical
      --order-by string                             Instructions on how to order the transfers, e.g. 'size,descending'
      --partial-suffix string                       Add partial-suffix to temporary file name when --inplace is not used (default ".partial")
      --refresh-times                               Refresh the modtime of remote files
      --server-side-across-configs                  Allow server-side operations (e.g. copy) to work across different configs
      --size-only                                   Skip based on size only, not modtime or checksum
      --streaming-upload-cutoff SizeSuffix          Cutoff for switching to chunked upload if file size is unknown, upload starts after reaching cutoff or when file ends (default 100Ki)
  -u, --update                                      Skip files that are newer on the destination

Sync Options

Flags used for sync commands

      --backup-dir string               Make backups into hierarchy based in DIR
      --delete-after                    When synchronizing, delete files on destination after transferring (default)
      --delete-before                   When synchronizing, delete files on destination before transferring
      --delete-during                   When synchronizing, delete files during transfer
      --fix-case                        Force rename of case insensitive dest to match source
      --ignore-errors                   Delete even if there are I/O errors
      --list-cutoff int                 To save memory, sort directory listings on disk above this threshold (default 1000000)
      --max-delete int                  When synchronizing, limit the number of deletes (default -1)
      --max-delete-size SizeSuffix      When synchronizing, limit the total size of deletes (default off)
      --suffix string                   Suffix to add to changed files
      --suffix-keep-extension           Preserve the extension when using --suffix
      --track-renames                   When synchronizing, track file renames and do a server-side move if possible
      --track-renames-strategy string   Strategies to use when synchronizing using track-renames hash|modtime|leaf (default "hash")

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone cat

Concatenates any files and sends them to stdout.

Synopsis

Sends any files to standard output.

You can use it like this to output a single file

rclone cat remote:path/to/file

Or like this to output any file in dir or its subdirectories.

rclone cat remote:path/to/dir

Or like this to output any .txt files in dir or its subdirectories.

rclone --include "*.txt" cat remote:path/to/dir

Use the --head flag to print characters only at the start, --tail for the end and --offset and --count to print a section in the middle. Note that if offset is negative it will count from the end, so --offset -1 --count 1 is equivalent to --tail 1.

Use the --separator flag to print a separator value between files. Be sure to shell-escape special characters. For example, to print a newline between files, use:

  • bash:

    rclone --include "*.txt" --separator $'\n' cat remote:path/to/dir
    
  • powershell:

    rclone --include "*.txt" --separator "`n" cat remote:path/to/dir
    
rclone cat remote:path [flags]

Options

      --count int          Only print N characters (default -1)
      --discard            Discard the output instead of printing
      --head int           Only print the first N characters
  -h, --help               help for cat
      --offset int         Start printing at offset N (or from end if -ve)
      --separator string   Separator to use between objects when printing multiple files
      --tail int           Only print the last N characters

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone checksum

Checks the files in the destination against a SUM file.

Synopsis

Checks that hashsums of destination files match the SUM file. It compares hashes (MD5, SHA1, etc) and logs a report of files which don't match. It doesn't alter the file system.

The sumfile is treated as the source and the dst:path is treated as the destination for the purposes of the output.

If you supply the --download flag, it will download the data from the remote and calculate the content hash on the fly. This can be useful for remotes that don't support hashes or if you really want to check all the data.

Note that hash values in the SUM file are treated as case insensitive.

If you supply the --one-way flag, it will only check that files in the source match the files in the destination, not the other way around. This means that extra files in the destination that are not in the source will not be detected.

The --differ, --missing-on-dst, --missing-on-src, --match and --error flags write paths, one per line, to the file name (or stdout if it is -) supplied. What they write is described in the help below. For example --differ will write all paths which are present on both the source and destination but different.

The --combined flag will write a file (or stdout) which contains all file paths with a symbol and then a space and then the path to tell you what happened to it. These are reminiscent of diff files.

  • = path means path was found in source and destination and was identical
  • - path means path was missing on the source, so only in the destination
  • + path means path was missing on the destination, so only in the source
  • * path means path was present in source and destination but different.
  • ! path means there was an error reading or hashing the source or dest.

The default number of parallel checks is 8. See the --checkers option for more information.

rclone checksum <hash> sumfile dst:path [flags]

Options

      --combined string         Make a combined report of changes to this file
      --differ string           Report all non-matching files to this file
      --download                Check by hashing the contents
      --error string            Report all files with errors (hashing or reading) to this file
  -h, --help                    help for checksum
      --match string            Report all matching files to this file
      --missing-on-dst string   Report all files missing from the destination to this file
      --missing-on-src string   Report all files missing from the source to this file
      --one-way                 Check one way only, source files must exist on remote

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone completion

Output completion script for a given shell.

Synopsis

Generates a shell completion script for rclone. Run with --help to list the supported shells. Enum-valued flags such as --log-level and --vfs-cache-mode also complete their supported values.

Options

  -h, --help   help for completion

See the global flags page for global options not listed here.

See Also

rclone completion bash

Output bash completion script for rclone.

Synopsis

Generates a bash shell autocompletion script for rclone.

By default, when run without any arguments,

rclone completion bash

the generated script will be written to

/etc/bash_completion.d/rclone

and so rclone will probably need to be run as root, or with sudo.

If you supply a path to a file as the command line argument, then the generated script will be written to that file, in which case you should not need root privileges.

If output_file is "-", then the output will be written to stdout.

If you have installed the script into the default location, you can logout and login again to use the autocompletion script.

Alternatively, you can source the script directly

. /path/to/my_bash_completion_scripts/rclone

and the autocompletion functionality will be added to your current shell.

rclone completion bash [output_file] [flags]

Options

  -h, --help   help for bash

See the global flags page for global options not listed here.

See Also

rclone completion fish

Output fish completion script for rclone.

Synopsis

Generates a fish autocompletion script for rclone.

This writes to /etc/fish/completions/rclone.fish by default so will probably need to be run with sudo or as root, e.g.

sudo rclone completion fish

Logout and login again to use the autocompletion scripts, or source them directly

. /etc/fish/completions/rclone.fish

If you supply a command line argument the script will be written there.

If output_file is "-", then the output will be written to stdout.

rclone completion fish [output_file] [flags]

Options

  -h, --help   help for fish

See the global flags page for global options not listed here.

See Also

rclone completion powershell

Output powershell completion script for rclone.

Synopsis

Generate the autocompletion script for powershell.

To load completions in your current shell session:

rclone completion powershell | Out-String | Invoke-Expression

To load completions for every new session, add the output of the above command to your powershell profile.

If output_file is "-" or missing, then the output will be written to stdout.

rclone completion powershell [output_file] [flags]

Options

  -h, --help   help for powershell

See the global flags page for global options not listed here.

See Also

rclone completion zsh

Output zsh completion script for rclone.

Synopsis

Generates a zsh autocompletion script for rclone.

This writes to /usr/share/zsh/vendor-completions/_rclone by default so will probably need to be run with sudo or as root, e.g.

sudo rclone completion zsh

Logout and login again to use the autocompletion scripts, or source them directly

autoload -U compinit && compinit

If you supply a command line argument the script will be written there.

If output_file is "-", then the output will be written to stdout.

rclone completion zsh [output_file] [flags]

Options

  -h, --help   help for zsh

See the global flags page for global options not listed here.

See Also

rclone config create

Create a new remote with name, type and options.

Synopsis

Create a new remote of name with type and options. The options should be passed in pairs of key value or as key=value.

For example, to make a swift remote of name myremote using auto config you would do:

rclone config create myremote swift env_auth true
rclone config create myremote swift env_auth=true

So for example if you wanted to configure a Google Drive remote but using remote authorization you would do this:

rclone config create mydrive drive config_is_local=false

Note that if the config process would normally ask a question the default is taken (unless --non-interactive is used). Each time that happens rclone will print or DEBUG a message saying how to affect the value taken.

If any of the parameters passed is a password field, then rclone will automatically obscure them if they aren't already obscured before putting them in the config file.

NB If the password parameter is 22 characters or longer and consists only of base64 characters then rclone can get confused about whether the password is already obscured or not and put unobscured passwords into the config file. If you want to be 100% certain that the passwords get obscured then use the --obscure flag, or if you are 100% certain you are already passing obscured passwords then use --no-obscure. You can also set obscured passwords using the rclone config password command.

The flag --non-interactive is for use by applications that wish to configure rclone themselves, rather than using rclone's text based configuration questions. If this flag is set, and rclone needs to ask the user a question, a JSON blob will be returned with the question in it.

This will look something like (some irrelevant detail removed):

{
  "State": "*oauth-islocal,teamdrive,,",
  "Option": {
    "Name": "config_is_local",
    "Help": "Use web browser to automatically authenticate rclone with remote?\n * Say Y if the machine running rclone has a web browser you can use\n * Say N if running rclone on a (remote) machine without web browser access\nIf not sure try Y. If Y failed, try N.\n",
    "Default": true,
    "Examples": [
      {
        "Value": "true",
        "Help": "Yes"
      },
      {
        "Value": "false",
        "Help": "No"
      }
    ],
    "Required": false,
    "IsPassword": false,
    "Type": "bool",
    "Exclusive": true,
  },
  "Error": "",
}

The format of Option is the same as returned by rclone config providers. The question should be asked to the user and returned to rclone as the --result option along with the --state parameter.

The keys of Option are used as follows:

  • Name - name of variable - show to user
  • Help - help text. Hard wrapped at 80 chars. Any URLs should be clicky.
  • Default - default value - return this if the user just wants the default.
  • Examples - the user should be able to choose one of these
  • Required - the value should be non-empty
  • IsPassword - the value is a password and should be edited as such
  • Type - type of value, eg bool, string, int and others
  • Exclusive - if set no free-form entry allowed only the Examples
  • Irrelevant keys Provider, ShortOpt, Hide, NoPrefix, Advanced

If Error is set then it should be shown to the user at the same time as the question.

rclone config update name --continue --state "*oauth-islocal,teamdrive,," --result "true"

Note that when using --continue all passwords should be passed in the clear (not obscured). Any default config values should be passed in with each invocation of --continue.

At the end of the non interactive process, rclone will return a result with State as empty string.

If --all is passed then rclone will ask all the config questions, not just the post config questions. Any parameters are used as defaults for questions as usual.

Note that bin/config.py in the rclone source implements this protocol as a readable demonstration.

rclone config create name type [key value]* [flags]

Options

      --all               Ask the full set of config questions
      --continue          Continue the configuration process with an answer
  -h, --help              help for create
      --no-obscure        Force any passwords not to be obscured
      --no-output         Don't provide any output
      --non-interactive   Don't interact with user and return questions
      --obscure           Force any passwords to be obscured
      --result string     Result - use with --continue
      --state string      State - use with --continue

See the global flags page for global options not listed here.

See Also

rclone config delete

Delete an existing remote.

rclone config delete name [flags]

Options

  -h, --help   help for delete

See the global flags page for global options not listed here.

See Also

rclone config disconnect

Disconnects user from remote

Synopsis

This disconnects the remote: passed in to the cloud storage system.

This normally means revoking the oauth token.

To reconnect use "rclone config reconnect".

rclone config disconnect remote: [flags]

Options

  -h, --help   help for disconnect

See the global flags page for global options not listed here.

See Also

rclone config dump

Dump the config file as JSON.

rclone config dump [flags]

Options

  -h, --help   help for dump

See the global flags page for global options not listed here.

See Also

rclone config edit

Enter an interactive configuration session.

Synopsis

Enter an interactive configuration session where you can setup new remotes and manage existing ones. You may also set or remove a password to protect your configuration.

rclone config edit [flags]

Options

  -h, --help   help for edit

See the global flags page for global options not listed here.

See Also

rclone config encryption

set, remove and check the encryption for the config file

Synopsis

This command sets, clears and checks the encryption for the config file using the subcommands below.

Options

  -h, --help   help for encryption

See the global flags page for global options not listed here.

See Also

rclone config encryption check

Check that the config file is encrypted

Synopsis

This checks the config file is encrypted and that you can decrypt it.

It will attempt to decrypt the config using the password you supply.

If decryption fails it will return a non-zero exit code if using --password-command, otherwise it will prompt again for the password.

If the config file is not encrypted it will return a non zero exit code.

rclone config encryption check [flags]

Options

  -h, --help   help for check

See the global flags page for global options not listed here.

See Also

rclone config encryption remove

Remove the config file encryption password

Synopsis

Remove the config file encryption password

This removes the config file encryption, returning it to un-encrypted.

If --password-command is in use, this will be called to supply the old config password.

If the config was not encrypted then no error will be returned and this command will do nothing.

rclone config encryption remove [flags]

Options

  -h, --help   help for remove

See the global flags page for global options not listed here.

See Also

rclone config encryption set

Set or change the config file encryption password

Synopsis

This command sets or changes the config file encryption password.

If there was no config password set then it sets a new one, otherwise it changes the existing config password.

Note that if you are changing an encryption password using --password-command then this will be called once to decrypt the config using the old password and then again to read the new password to re-encrypt the config.

When --password-command is called to change the password then the environment variable RCLONE_PASSWORD_CHANGE=1 will be set. So if changing passwords programmatically you can use the environment variable to distinguish which password you must supply.

Alternatively you can remove the password first (with rclone config encryption remove), then set it again with this command which may be easier if you don't mind the unencrypted config file being on the disk briefly.

rclone config encryption set [flags]

Options

  -h, --help   help for set

See the global flags page for global options not listed here.

See Also

rclone config file

Show path of configuration file in use.

rclone config file [flags]

Options

  -h, --help   help for file

See the global flags page for global options not listed here.

See Also

rclone config password

Update password in an existing remote.

Synopsis

Update an existing remote's password. The password should be passed in pairs of key password or as key=password. The password should be passed in in clear (unobscured).

For example, to set password of a remote of name myremote you would do:

rclone config password myremote fieldname mypassword
rclone config password myremote fieldname=mypassword

This command is obsolete now that "config update" and "config create" both support obscuring passwords directly.

rclone config password name [key value]+ [flags]

Options

  -h, --help   help for password

See the global flags page for global options not listed here.

See Also

rclone config paths

Show paths used for configuration, cache, temp etc.

rclone config paths [flags]

Options

  -h, --help   help for paths

See the global flags page for global options not listed here.

See Also

rclone config providers

List in JSON format all the providers and options.

rclone config providers [flags]

Options

  -h, --help   help for providers

See the global flags page for global options not listed here.

See Also

rclone config reconnect

Re-authenticates user with remote.

Synopsis

This reconnects remote: passed in to the cloud storage system.

To disconnect the remote use "rclone config disconnect".

This normally means going through the interactive oauth flow again.

rclone config reconnect remote: [flags]

Options

  -h, --help   help for reconnect

See the global flags page for global options not listed here.

See Also

rclone config redacted

Print redacted (decrypted) config file, or the redacted config for a single remote.

Synopsis

This prints a redacted copy of the config file, either the whole config file or for a given remote.

The config file will be redacted by replacing all passwords and other sensitive info with XXX.

This makes the config file suitable for posting online for support.

It should be double checked before posting as the redaction may not be perfect.

rclone config redacted [<remote>] [flags]

Options

  -h, --help   help for redacted

See the global flags page for global options not listed here.

See Also

rclone config show

Print (decrypted) config file, or the config for a single remote.

rclone config show [<remote>] [flags]

Options

  -h, --help   help for show

See the global flags page for global options not listed here.

See Also

rclone config string

Print connection string for a single remote.

Synopsis

Print a connection string for a single remote.

The connection strings can be used wherever a remote is needed and can be more convenient than using the config file, especially if using the RC API.

Backend parameters may be provided to the command also.

Example:

$ rclone config string s3:rclone --s3-no-check-bucket
:s3,access_key_id=XXX,no_check_bucket,provider=AWS,region=eu-west-2,secret_access_key=YYY:rclone

NB the strings are not quoted for use in shells (eg bash, powershell, windows cmd). Most will work if enclosed in "double quotes", however connection strings that contain double quotes will require further quoting which is very shell dependent.

rclone config string <remote> [flags]

Options

  -h, --help   help for string

See the global flags page for global options not listed here.

See Also

rclone config touch

Ensure configuration file exists.

rclone config touch [flags]

Options

  -h, --help   help for touch

See the global flags page for global options not listed here.

See Also

rclone config unset

Unset options in an existing remote.

Synopsis

Remove one or more options from an existing remote. The options to remove should be passed in as a list of key names.

For example, to remove the client_id and client_secret options from a remote of name myremote you would do:

rclone config unset myremote client_id client_secret

This removes the keys from the config file entirely, which is different from setting them to an empty string with config update. Removing a key restores rclone's default behaviour for that option, whereas setting it to an empty string overrides the default with an empty value.

You can't unset the type of a remote - use config delete to remove the whole remote instead.

rclone config unset name [key]+ [flags]

Options

  -h, --help   help for unset

See the global flags page for global options not listed here.

See Also

rclone config update

Update options in an existing remote.

Synopsis

Update an existing remote's options. The options should be passed in pairs of key value or as key=value.

For example, to update the env_auth field of a remote of name myremote you would do:

rclone config update myremote env_auth true
rclone config update myremote env_auth=true

If the remote uses OAuth the token will be updated, if you don't require this add an extra parameter thus:

rclone config update myremote env_auth=true config_refresh_token=false

Note that if the config process would normally ask a question the default is taken (unless --non-interactive is used). Each time that happens rclone will print or DEBUG a message saying how to affect the value taken.

If any of the parameters passed is a password field, then rclone will automatically obscure them if they aren't already obscured before putting them in the config file.

NB If the password parameter is 22 characters or longer and consists only of base64 characters then rclone can get confused about whether the password is already obscured or not and put unobscured passwords into the config file. If you want to be 100% certain that the passwords get obscured then use the --obscure flag, or if you are 100% certain you are already passing obscured passwords then use --no-obscure. You can also set obscured passwords using the rclone config password command.

The flag --non-interactive is for use by applications that wish to configure rclone themselves, rather than using rclone's text based configuration questions. If this flag is set, and rclone needs to ask the user a question, a JSON blob will be returned with the question in it.

This will look something like (some irrelevant detail removed):

{
  "State": "*oauth-islocal,teamdrive,,",
  "Option": {
    "Name": "config_is_local",
    "Help": "Use web browser to automatically authenticate rclone with remote?\n * Say Y if the machine running rclone has a web browser you can use\n * Say N if running rclone on a (remote) machine without web browser access\nIf not sure try Y. If Y failed, try N.\n",
    "Default": true,
    "Examples": [
      {
        "Value": "true",
        "Help": "Yes"
      },
      {
        "Value": "false",
        "Help": "No"
      }
    ],
    "Required": false,
    "IsPassword": false,
    "Type": "bool",
    "Exclusive": true,
  },
  "Error": "",
}

The format of Option is the same as returned by rclone config providers. The question should be asked to the user and returned to rclone as the --result option along with the --state parameter.

The keys of Option are used as follows:

  • Name - name of variable - show to user
  • Help - help text. Hard wrapped at 80 chars. Any URLs should be clicky.
  • Default - default value - return this if the user just wants the default.
  • Examples - the user should be able to choose one of these
  • Required - the value should be non-empty
  • IsPassword - the value is a password and should be edited as such
  • Type - type of value, eg bool, string, int and others
  • Exclusive - if set no free-form entry allowed only the Examples
  • Irrelevant keys Provider, ShortOpt, Hide, NoPrefix, Advanced

If Error is set then it should be shown to the user at the same time as the question.

rclone config update name --continue --state "*oauth-islocal,teamdrive,," --result "true"

Note that when using --continue all passwords should be passed in the clear (not obscured). Any default config values should be passed in with each invocation of --continue.

At the end of the non interactive process, rclone will return a result with State as empty string.

If --all is passed then rclone will ask all the config questions, not just the post config questions. Any parameters are used as defaults for questions as usual.

Note that bin/config.py in the rclone source implements this protocol as a readable demonstration.

rclone config update name [key value]+ [flags]

Options

      --all               Ask the full set of config questions
      --continue          Continue the configuration process with an answer
  -h, --help              help for update
      --no-obscure        Force any passwords not to be obscured
      --no-output         Don't provide any output
      --non-interactive   Don't interact with user and return questions
      --obscure           Force any passwords to be obscured
      --result string     Result - use with --continue
      --state string      State - use with --continue

See the global flags page for global options not listed here.

See Also

rclone config userinfo

Prints info about logged in user of remote.

Synopsis

This prints the details of the person logged in to the cloud storage system.

rclone config userinfo remote: [flags]

Options

  -h, --help   help for userinfo
      --json   Format output as JSON

See the global flags page for global options not listed here.

See Also

rclone convmv

Convert file and directory names in place.

Synopsis

convmv supports advanced path name transformations for converting and renaming files and directories by applying prefixes, suffixes, and other alterations.

Command Description
--name-transform prefix=XXXX Prepends XXXX to the file name.
--name-transform suffix=XXXX Appends XXXX to the file name after the extension.
--name-transform suffix_keep_extension=XXXX Appends XXXX to the file name while preserving the original file extension.
--name-transform trimprefix=XXXX Removes XXXX if it appears at the start of the file name.
--name-transform trimsuffix=XXXX Removes XXXX if it appears at the end of the file name.
--name-transform regex=pattern/replacement Applies a regex-based transformation.
--name-transform replace=old:new Replaces occurrences of old with new in the file name.
--name-transform date={YYYYMMDD} Appends or prefixes the specified date format.
--name-transform truncate=N Truncates the file name to a maximum of N characters.
--name-transform truncate_keep_extension=N Truncates the file name to a maximum of N characters while preserving the original file extension.
--name-transform truncate_bytes=N Truncates the file name to a maximum of N bytes (not characters).
--name-transform truncate_bytes_keep_extension=N Truncates the file name to a maximum of N bytes (not characters) while preserving the original file extension.
--name-transform base64encode Encodes the file name in Base64.
--name-transform base64decode Decodes a Base64-encoded file name.
--name-transform encoder=ENCODING Converts the file name to the specified encoding (e.g., ISO-8859-1, Windows-1252, Macintosh).
--name-transform decoder=ENCODING Decodes the file name from the specified encoding.
--name-transform charmap=MAP Applies a character mapping transformation.
--name-transform lowercase Converts the file name to lowercase.
--name-transform uppercase Converts the file name to UPPERCASE.
--name-transform titlecase Converts the file name to Title Case.
--name-transform ascii Strips non-ASCII characters.
--name-transform url URL-encodes the file name.
--name-transform nfc Converts the file name to NFC Unicode normalization form.
--name-transform nfd Converts the file name to NFD Unicode normalization form.
--name-transform nfkc Converts the file name to NFKC Unicode normalization form.
--name-transform nfkd Converts the file name to NFKD Unicode normalization form.
--name-transform command=/path/to/my/programfile names. Executes an external program to transform.

Conversion modes:

none
nfc
nfd
nfkc
nfkd
replace
prefix
suffix
suffix_keep_extension
trimprefix
trimsuffix
index
date
truncate
truncate_keep_extension
truncate_bytes
truncate_bytes_keep_extension
base64encode
base64decode
encoder
decoder
ISO-8859-1
Windows-1252
Macintosh
charmap
lowercase
uppercase
titlecase
ascii
url
regex
command

Char maps:

IBM-Code-Page-037
IBM-Code-Page-437
IBM-Code-Page-850
IBM-Code-Page-852
IBM-Code-Page-855
Windows-Code-Page-858
IBM-Code-Page-860
IBM-Code-Page-862
IBM-Code-Page-863
IBM-Code-Page-865
IBM-Code-Page-866
IBM-Code-Page-1047
IBM-Code-Page-1140
ISO-8859-1
ISO-8859-2
ISO-8859-3
ISO-8859-4
ISO-8859-5
ISO-8859-6
ISO-8859-7
ISO-8859-8
ISO-8859-9
ISO-8859-10
ISO-8859-13
ISO-8859-14
ISO-8859-15
ISO-8859-16
KOI8-R
KOI8-U
Macintosh
Macintosh-Cyrillic
Windows-874
Windows-1250
Windows-1251
Windows-1252
Windows-1253
Windows-1254
Windows-1255
Windows-1256
Windows-1257
Windows-1258
X-User-Defined

Encoding masks:

Asterisk
BackQuote
BackSlash
Colon
CrLf
Ctl
Del
Dollar
Dot
DoubleQuote
Exclamation
Hash
InvalidUtf8
LeftCrLfHtVt
LeftPeriod
LeftSpace
LeftTilde
LtGt
None
Percent
Pipe
Question
Raw
RightCrLfHtVt
RightPeriod
RightSpace
Semicolon
SingleQuote
Slash
SquareBracket

Examples:

rclone convmv "stories/The Quick Brown Fox!.txt" --name-transform "all,uppercase"
// Output: STORIES/THE QUICK BROWN FOX!.TXT
rclone convmv "stories/The Quick Brown Fox!.txt" --name-transform "all,replace=Fox:Turtle" --name-transform "all,replace=Quick:Slow"
// Output: stories/The Slow Brown Turtle!.txt
rclone convmv "stories/The Quick Brown Fox!.txt" --name-transform "all,base64encode"
// Output: c3Rvcmllcw==/VGhlIFF1aWNrIEJyb3duIEZveCEudHh0
rclone convmv "c3Rvcmllcw==/VGhlIFF1aWNrIEJyb3duIEZveCEudHh0" --name-transform "all,base64decode"
// Output: stories/The Quick Brown Fox!.txt
rclone convmv "stories/The Quick Brown 🦊 Fox Went to the Café!.txt" --name-transform "all,nfc"
// Output: stories/The Quick Brown 🦊 Fox Went to the Café!.txt
rclone convmv "stories/The Quick Brown 🦊 Fox Went to the Café!.txt" --name-transform "all,nfd"
// Output: stories/The Quick Brown 🦊 Fox Went to the Café!.txt
rclone convmv "stories/The Quick Brown 🦊 Fox!.txt" --name-transform "all,ascii"
// Output: stories/The Quick Brown  Fox!.txt
rclone convmv "stories/The Quick Brown Fox!.txt" --name-transform "all,trimsuffix=.txt"
// Output: stories/The Quick Brown Fox!
rclone convmv "stories/The Quick Brown Fox!.txt" --name-transform "all,prefix=OLD_"
// Output: OLD_stories/OLD_The Quick Brown Fox!.txt
rclone convmv "stories/The Quick Brown 🦊 Fox Went to the Café!.txt" --name-transform "all,charmap=ISO-8859-7"
// Output: stories/The Quick Brown _ Fox Went to the Caf_!.txt
rclone convmv "stories/The Quick Brown Fox: A Memoir [draft].txt" --name-transform "all,encoder=Colon,SquareBracket"
// Output: stories/The Quick Brown Fox: A Memoir [draft].txt
rclone convmv "stories/The Quick Brown 🦊 Fox Went to the Café!.txt" --name-transform "all,truncate=21"
// Output: stories/The Quick Brown 🦊 Fox
rclone convmv "stories/The Quick Brown Fox!.txt" --name-transform "all,command=echo"
// Output: stories/The Quick Brown Fox!.txt
rclone convmv "stories/The Quick Brown Fox!" --name-transform "date=-{YYYYMMDD}"
// Output: stories/The Quick Brown Fox!-20261106
rclone convmv "stories/The Quick Brown Fox!" --name-transform "date=-{macfriendlytime}"
// Output: stories/The Quick Brown Fox!-2026-11-06 0220PM
rclone convmv "stories/The Quick Brown Fox!.txt" --name-transform "all,regex=[\\.\\w]/ab"
// Output: ababababababab/ababab ababababab ababababab ababab!abababab

The regex command generally accepts Perl-style regular expressions, the exact syntax is defined in the Go regular expression reference. The replacement string may contain capturing group variables, referencing capturing groups using the syntax $name or ${name}, where the name can refer to a named capturing group or it can simply be the index as a number. To insert a literal $, use $$.

Multiple transformations can be used in sequence, applied in the order they are specified on the command line.

The --name-transform flag is also available in sync, copy, and move.

Files vs Directories

By default --name-transform will only apply to file names. The means only the leaf file name will be transformed. However some of the transforms would be better applied to the whole path or just directories. To choose which which part of the file path is affected some tags can be added to the --name-transform.

Tag Effect
file Only transform the leaf name of files (DEFAULT)
dir Only transform name of directories - these may appear anywhere in the path
all Transform the entire path for files and directories

This is used by adding the tag into the transform name like this: --name-transform file,prefix=ABC or --name-transform dir,prefix=DEF.

For some conversions using all is more likely to be useful, for example --name-transform all,nfc.

Note that --name-transform may not add path separators / to the name. This will cause an error.

Ordering and Conflicts

  • Transformations will be applied in the order specified by the user.
    • If the file tag is in use (the default) then only the leaf name of files will be transformed.
    • If the dir tag is in use then directories anywhere in the path will be transformed
    • If the all tag is in use then directories and files anywhere in the path will be transformed
    • Each transformation will be run one path segment at a time.
    • If a transformation adds a / or ends up with an empty path segment then that will be an error.
  • It is up to the user to put the transformations in a sensible order.
    • Conflicting transformations, such as prefix followed by trimprefix or nfc followed by nfd, are possible.
    • Instead of enforcing mutual exclusivity, transformations are applied in sequence as specified by the user, allowing for intentional use cases (e.g., trimming one prefix before adding another).
    • Users should be aware that certain combinations may lead to unexpected results and should verify transformations using --dry-run before execution.

Race Conditions and Non-Deterministic Behavior

Some transformations, such as replace=old:new, may introduce conflicts where multiple source files map to the same destination name. This can lead to race conditions when performing concurrent transfers. It is up to the user to anticipate these.

  • If two files from the source are transformed into the same name at the destination, the final state may be non-deterministic.
  • Running rclone check after a sync using such transformations may erroneously report missing or differing files due to overwritten results.

To minimize risks, users should:

  • Carefully review transformations that may introduce conflicts.
  • Use --dry-run to inspect changes before executing a sync (but keep in mind that it won't show the effect of non-deterministic transformations).
  • Avoid transformations that cause multiple distinct source files to map to the same destination name.
  • Consider disabling concurrency with --transfers=1 if necessary.
  • Certain transformations (e.g. prefix) will have a multiplying effect every time they are used. Avoid these when using bisync.
rclone convmv dest:path --name-transform XXX [flags]

Options

      --create-empty-src-dirs   Create empty source dirs on destination after move
      --delete-empty-src-dirs   Delete empty source dirs after move
  -h, --help                    help for convmv

Options shared with other commands are described next. See the global flags page for global options not listed here.

Copy Options

Flags for anything which can copy a file

      --check-first                                 Do all the checks before starting transfers
  -c, --checksum                                    Check for changes with size & checksum (if available, or fallback to size only)
      --compare-dest stringArray                    Include additional server-side paths during comparison
      --copy-dest stringArray                       Implies --compare-dest but also copies files from paths into destination
      --cutoff-mode HARD|SOFT|CAUTIOUS              Mode to stop transfers when reaching the max transfer limit HARD|SOFT|CAUTIOUS (default HARD)
      --ignore-case-sync                            Ignore case when synchronizing
      --ignore-checksum                             Skip post copy check of checksums
      --ignore-existing                             Skip all files that exist on destination
      --ignore-size                                 Ignore size when skipping use modtime or checksum
  -I, --ignore-times                                Don't skip items that match size and time - transfer all unconditionally
      --immutable                                   Do not modify files, fail if existing files have been modified
      --inplace                                     Download directly to destination file instead of atomic download to temp/rename
  -l, --links                                       Translate symlinks to/from regular files with a '.rclonelink' extension
      --max-backlog int                             Maximum number of objects in sync or check backlog (default 10000)
      --max-duration Duration                       Maximum duration rclone will transfer data for (default 0s)
      --max-transfer SizeSuffix                     Maximum size of data to transfer (default off)
  -M, --metadata                                    If set, preserve metadata when copying objects
      --modify-window Duration                      Max time diff to be considered the same (default 1ns)
      --multi-thread-chunk-size SizeSuffix          Chunk size for multi-thread downloads / uploads, if not set by filesystem (default 64Mi)
      --multi-thread-cutoff SizeSuffix              Use multi-thread downloads for files above this size (default 256Mi)
      --multi-thread-streams int                    Number of streams to use for multi-thread downloads (default 4)
      --multi-thread-write-buffer-size SizeSuffix   In memory buffer size for writing when in multi-thread mode (default 128Ki)
      --name-transform stringArray                  Transform paths during the copy process
      --no-check-dest                               Don't check the destination, copy regardless
      --no-traverse                                 Don't traverse destination file system on copy
      --no-update-dir-modtime                       Don't update directory modification times
      --no-update-modtime                           Don't update destination modtime if files identical
      --order-by string                             Instructions on how to order the transfers, e.g. 'size,descending'
      --partial-suffix string                       Add partial-suffix to temporary file name when --inplace is not used (default ".partial")
      --refresh-times                               Refresh the modtime of remote files
      --server-side-across-configs                  Allow server-side operations (e.g. copy) to work across different configs
      --size-only                                   Skip based on size only, not modtime or checksum
      --streaming-upload-cutoff SizeSuffix          Cutoff for switching to chunked upload if file size is unknown, upload starts after reaching cutoff or when file ends (default 100Ki)
  -u, --update                                      Skip files that are newer on the destination

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone copyto

Copy files from source to dest, skipping identical files.

Synopsis

If source:path is a file or directory then it copies it to a file or directory named dest:path.

This can be used to copy a single file to a destination with a name different from its source - for example, uploading and renaming in one step. If the source is a directory then it acts exactly like the copy command.

So

rclone copyto src dst

where src and dst are rclone paths, either remote:path or /path/to/local or C:\windows\path\if\on\windows.

This will:

if src is file
    copy it to dst, overwriting an existing file if it exists
if src is directory
    copy it to dst, overwriting existing files if they exist
    see copy command for full details

This doesn't transfer files that are identical on src and dst, testing by size and modification time or MD5SUM. It doesn't delete files from the destination.

If you are looking to copy just a byte range of a file, please see rclone cat --offset X --count Y.

Note: Use the -P/--progress flag to view real-time transfer statistics.

Logger Flags

The --differ, --missing-on-dst, --missing-on-src, --match and --error flags write paths, one per line, to the file name (or stdout if it is -) supplied. What they write is described in the help below. For example --differ will write all paths which are present on both the source and destination but different.

The --combined flag will write a file (or stdout) which contains all file paths with a symbol and then a space and then the path to tell you what happened to it. These are reminiscent of diff files.

  • = path means path was found in source and destination and was identical
  • - path means path was missing on the source, so only in the destination
  • + path means path was missing on the destination, so only in the source
  • * path means path was present in source and destination but different.
  • ! path means there was an error reading or hashing the source or dest.

The --dest-after flag writes a list file using the same format flags as lsf (including customizable options for hash, modtime, etc.) Conceptually it is similar to rsync's --itemize-changes, but not identical -- it should output an accurate list of what will be on the destination after the command is finished.

When the --no-traverse flag is set, all logs involving files that exist only on the destination will be incomplete or completely missing.

Note that these logger flags have a few limitations, and certain scenarios are not currently supported:

  • --max-duration / CutoffModeHard
  • --compare-dest / --copy-dest
  • server-side moves of an entire dir at once
  • High-level retries, because there would be duplicates (use --retries 1 to disable)
  • Possibly some unusual error scenarios

Note also that each file is logged during execution, as opposed to after, so it is most useful as a predictor of what SHOULD happen to each file (which may or may not match what actually DID).

rclone copyto source:path dest:path [flags]

Options

      --absolute                Put a leading / in front of path names
      --combined string         Make a combined report of changes to this file
      --csv                     Output in CSV format
      --dest-after string       Report all files that exist on the dest post-sync
      --differ string           Report all non-matching files to this file
  -d, --dir-slash               Append a slash to directory names (default true)
      --dirs-only               Only list directories
      --error string            Report all files with errors (hashing or reading) to this file
      --files-only              Only list files (default true)
  -F, --format string           Output format - see lsf help for details (default "p")
      --hash h                  Use this hash when h is used in the format MD5|SHA-1|DropboxHash (default "md5")
  -h, --help                    help for copyto
      --match string            Report all matching files to this file
      --missing-on-dst string   Report all files missing from the destination to this file
      --missing-on-src string   Report all files missing from the source to this file
  -s, --separator string        Separator for the items in the format (default ";")
  -t, --timeformat string       Specify a custom time format - see docs for details (default: 2006-01-02 15:04:05)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Copy Options

Flags for anything which can copy a file

      --check-first                                 Do all the checks before starting transfers
  -c, --checksum                                    Check for changes with size & checksum (if available, or fallback to size only)
      --compare-dest stringArray                    Include additional server-side paths during comparison
      --copy-dest stringArray                       Implies --compare-dest but also copies files from paths into destination
      --cutoff-mode HARD|SOFT|CAUTIOUS              Mode to stop transfers when reaching the max transfer limit HARD|SOFT|CAUTIOUS (default HARD)
      --ignore-case-sync                            Ignore case when synchronizing
      --ignore-checksum                             Skip post copy check of checksums
      --ignore-existing                             Skip all files that exist on destination
      --ignore-size                                 Ignore size when skipping use modtime or checksum
  -I, --ignore-times                                Don't skip items that match size and time - transfer all unconditionally
      --immutable                                   Do not modify files, fail if existing files have been modified
      --inplace                                     Download directly to destination file instead of atomic download to temp/rename
  -l, --links                                       Translate symlinks to/from regular files with a '.rclonelink' extension
      --max-backlog int                             Maximum number of objects in sync or check backlog (default 10000)
      --max-duration Duration                       Maximum duration rclone will transfer data for (default 0s)
      --max-transfer SizeSuffix                     Maximum size of data to transfer (default off)
  -M, --metadata                                    If set, preserve metadata when copying objects
      --modify-window Duration                      Max time diff to be considered the same (default 1ns)
      --multi-thread-chunk-size SizeSuffix          Chunk size for multi-thread downloads / uploads, if not set by filesystem (default 64Mi)
      --multi-thread-cutoff SizeSuffix              Use multi-thread downloads for files above this size (default 256Mi)
      --multi-thread-streams int                    Number of streams to use for multi-thread downloads (default 4)
      --multi-thread-write-buffer-size SizeSuffix   In memory buffer size for writing when in multi-thread mode (default 128Ki)
      --name-transform stringArray                  Transform paths during the copy process
      --no-check-dest                               Don't check the destination, copy regardless
      --no-traverse                                 Don't traverse destination file system on copy
      --no-update-dir-modtime                       Don't update directory modification times
      --no-update-modtime                           Don't update destination modtime if files identical
      --order-by string                             Instructions on how to order the transfers, e.g. 'size,descending'
      --partial-suffix string                       Add partial-suffix to temporary file name when --inplace is not used (default ".partial")
      --refresh-times                               Refresh the modtime of remote files
      --server-side-across-configs                  Allow server-side operations (e.g. copy) to work across different configs
      --size-only                                   Skip based on size only, not modtime or checksum
      --streaming-upload-cutoff SizeSuffix          Cutoff for switching to chunked upload if file size is unknown, upload starts after reaching cutoff or when file ends (default 100Ki)
  -u, --update                                      Skip files that are newer on the destination

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone copyurl

Copy the contents of the URL supplied content to dest:path.

Synopsis

Download a URL's content and copy it to the destination without saving it in temporary storage.

Setting --auto-filename will attempt to automatically determine the filename from the URL (after any redirections) and used in the destination path.

With --header-filename in addition, if a specific filename is set in HTTP headers, it will be used instead of the name from the URL. With --print-filename in addition, the resulting file name will be printed.

Setting --no-clobber will prevent overwriting file on the destination if there is one with the same name.

Setting --stdout or making the output file name - will cause the output to be written to standard output.

Setting --urls allows you to input a CSV file of URLs in format: URL, FILENAME. If --urls is in use then replace the URL in the arguments with the file containing the URLs, e.g.:

rclone copyurl --urls myurls.csv remote:dir

Missing filenames will be autogenerated equivalent to using --auto-filename. Note that --stdout and --print-filename are incompatible with --urls. This will do --transfers copies in parallel. Note that if --auto-filename is desired for all URLs then a file with only URLs and no filename can be used.

Each FILENAME in the CSV file can start with a relative path which will be appended to the destination path provided at the command line. For example, running the command shown above with the following CSV file will write two files to the destination: remote:dir/local/path/bar.json and remote:dir/another/local/directory/qux.json

https://example.org/foo/bar.json,local/path/bar.json
https://example.org/qux/baz.json,another/local/directory/qux.json

Troubleshooting

If you can't get rclone copyurl to work then here are some things you can try:

  • --disable-http2 rclone will use HTTP2 if available - try disabling it
  • --bind 0.0.0.0 rclone will use IPv6 if available - try disabling it
  • --bind ::0 to disable IPv4
  • --user agent curl - some sites have whitelists for curl's user-agent - try that
  • Make sure the site works with curl directly
rclone copyurl https://example.com dest:path [flags]

Options

  -a, --auto-filename     Get the file name from the URL and use it for destination file path
      --header-filename   Get the file name from the Content-Disposition header
  -h, --help              help for copyurl
      --no-clobber        Prevent overwriting file with same name
  -p, --print-filename    Print the resulting name from --auto-filename
      --stdout            Write the output to stdout rather than a file
      --urls              Use a CSV file of links to process multiple URLs

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone cryptcheck

Cryptcheck checks the integrity of an encrypted remote.

Synopsis

Checks a remote against an encrypted remote. This is the equivalent of running rclone check, but able to check the checksums of the encrypted remote.

For it to work the underlying remote of the cryptedremote must support some kind of checksum.

It works by reading the nonce from each file on the cryptedremote: and using that to encrypt each file on the remote:. It then checks the checksum of the underlying file on the cryptedremote: against the checksum of the file it has just encrypted.

Use it like this

rclone cryptcheck /path/to/files encryptedremote:path

You can use it like this also, but that will involve downloading all the files in remote:path.

rclone cryptcheck remote:path encryptedremote:path

After it has run it will log the status of the encryptedremote:.

If you supply the --one-way flag, it will only check that files in the source match the files in the destination, not the other way around. This means that extra files in the destination that are not in the source will not be detected.

The --differ, --missing-on-dst, --missing-on-src, --match and --error flags write paths, one per line, to the file name (or stdout if it is -) supplied. What they write is described in the help below. For example --differ will write all paths which are present on both the source and destination but different.

The --combined flag will write a file (or stdout) which contains all file paths with a symbol and then a space and then the path to tell you what happened to it. These are reminiscent of diff files.

  • = path means path was found in source and destination and was identical
  • - path means path was missing on the source, so only in the destination
  • + path means path was missing on the destination, so only in the source
  • * path means path was present in source and destination but different.
  • ! path means there was an error reading or hashing the source or dest.

The default number of parallel checks is 8. See the --checkers option for more information.

rclone cryptcheck remote:path cryptedremote:path [flags]

Options

      --combined string         Make a combined report of changes to this file
      --differ string           Report all non-matching files to this file
      --error string            Report all files with errors (hashing or reading) to this file
  -h, --help                    help for cryptcheck
      --match string            Report all matching files to this file
      --missing-on-dst string   Report all files missing from the destination to this file
      --missing-on-src string   Report all files missing from the source to this file
      --one-way                 Check one way only, source files must exist on remote

Options shared with other commands are described next. See the global flags page for global options not listed here.

Check Options

Flags used for check commands

      --max-backlog int   Maximum number of objects in sync or check backlog (default 10000)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone cryptdecode

Cryptdecode returns unencrypted file names.

Synopsis

Returns unencrypted file names when provided with a list of encrypted file names. List limit is 10 items.

If you supply the --reverse flag, it will return encrypted file names.

use it like this

rclone cryptdecode encryptedremote: encryptedfilename1 encryptedfilename2
rclone cryptdecode --reverse encryptedremote: filename1 filename2

Another way to accomplish this is by using the rclone backend encode (or decode) command. See the documentation on the crypt overlay for more info.

rclone cryptdecode encryptedremote: encryptedfilename [flags]

Options

  -h, --help      help for cryptdecode
      --reverse   Reverse cryptdecode, encrypts filenames

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone deletefile

Remove a single file from remote.

Synopsis

Remove a single file from remote. Unlike delete it cannot be used to remove a directory and it doesn't obey include/exclude filters - if the specified file exists, it will always be removed from its original location.

The --backup-dir and --suffix flags are honored when set.

rclone deletefile remote:path [flags]

Options

  -h, --help   help for deletefile

Options shared with other commands are described next. See the global flags page for global options not listed here.

Sync Options

Flags used for sync commands

      --backup-dir string               Make backups into hierarchy based in DIR
      --delete-after                    When synchronizing, delete files on destination after transferring (default)
      --delete-before                   When synchronizing, delete files on destination before transferring
      --delete-during                   When synchronizing, delete files during transfer
      --fix-case                        Force rename of case insensitive dest to match source
      --ignore-errors                   Delete even if there are I/O errors
      --list-cutoff int                 To save memory, sort directory listings on disk above this threshold (default 1000000)
      --max-delete int                  When synchronizing, limit the number of deletes (default -1)
      --max-delete-size SizeSuffix      When synchronizing, limit the total size of deletes (default off)
      --suffix string                   Suffix to add to changed files
      --suffix-keep-extension           Preserve the extension when using --suffix
      --track-renames                   When synchronizing, track file renames and do a server-side move if possible
      --track-renames-strategy string   Strategies to use when synchronizing using track-renames hash|modtime|leaf (default "hash")

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone gendocs

Output markdown docs for rclone to the directory supplied.

Synopsis

This produces markdown docs for the rclone commands to the directory supplied. These are in a format suitable for hugo to render into the rclone.org website.

rclone gendocs output_directory [flags]

Options

  -h, --help   help for gendocs

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone gitannex

Speaks with git-annex over stdin/stdout.

Synopsis

Rclone's gitannex subcommand enables git-annex to store and retrieve content from an rclone remote. It is meant to be run by git-annex, not directly by users.

Installation on Linux

  1. Skip this step if your version of git-annex is 10.20240430 or newer. Otherwise, you must create a symlink somewhere on your PATH with a particular name. This symlink helps git-annex tell rclone it wants to run the "gitannex" subcommand.

    Create the helper symlink in "$HOME/bin":

    ln -s "$(realpath rclone)" "$HOME/bin/git-annex-remote-rclone-builtin"
    
    Verify the new symlink is on your PATH:
    
    ```console
    which git-annex-remote-rclone-builtin
    
  2. Add a new remote to your git-annex repo. This new remote will connect git-annex with the rclone gitannex subcommand.

    Start by asking git-annex to describe the remote's available configuration parameters.

    If you skipped step 1:

    git annex initremote MyRemote type=rclone --whatelse
    

    If you created a symlink in step 1:

    git annex initremote MyRemote type=external externaltype=rclone-builtin --whatelse
    

    NOTE: If you're porting an existing git-annex-remote-rclone remote to use rclone gitannex, you can probably reuse the configuration parameters verbatim without renaming them. Check parameter synonyms with --whatelse as shown above.

    The following example creates a new git-annex remote named "MyRemote" that will use the rclone remote named "SomeRcloneRemote". That rclone remote must be one configured in your rclone.conf file, which can be located with rclone config file.

    git annex initremote MyRemote         \
        type=external                     \
        externaltype=rclone-builtin       \
        encryption=none                   \
        rcloneremotename=SomeRcloneRemote \
        rcloneprefix=git-annex-content    \
        rclonelayout=nodir
    
  3. Before you trust this command with your precious data, be sure to test the remote. This command is very new and has not been tested on many rclone backends. Caveat emptor!

    git annex testremote MyRemote
    

Happy annexing!

rclone gitannex [flags]

Options

  -h, --help   help for gitannex

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone gui

Open the web based GUI.

Synopsis

This command starts an embedded web GUI for rclone and opens it in your default browser.

This starts an RC API server and a GUI server on separate localhost ports, generates login credentials automatically unless --no-auth is specified, and opens the browser already authenticated.

rclone gui

By default rclone gui serves the web GUI that was embedded into the rclone binary at build time from https://github.com/rclone/rclone-web/ You can override this by passing a path to either an unpacked GUI directory or a dist.zip archive (e.g. one downloaded from the rclone-web releases page):

rclone gui ./my-dist/
rclone gui ./dist.zip

This is useful for iterating on the GUI locally without rebuilding rclone, or for serving a different GUI release than the one embedded.

Use --no-open-browser to skip opening the browser automatically:

rclone gui --no-open-browser

Use --addr to bind the GUI to a specific address:

rclone gui --addr localhost:5580

Use --user and --pass to set specific credentials:

rclone gui --user admin --pass secret

Use --no-auth to disable authentication entirely:

rclone gui --no-auth

For more help see the GUI docs.

rclone gui [path] [flags]

Options

      --addr stringArray       IPaddress:Port for the GUI server (default auto-chosen localhost port)
      --api-addr stringArray   IPaddress:Port for the RC API server (default auto-chosen localhost port)
      --enable-metrics         Enable OpenMetrics/Prometheus compatible endpoint at /metrics
  -h, --help                   help for gui
      --no-auth                Don't require auth for the RC API
      --no-open-browser        Skip opening the browser automatically
      --pass string            Password for RC authentication
      --user string            User name for RC authentication

Options shared with other commands are described next. See the global flags page for global options not listed here.

RC Options

Flags to control the Remote Control API

      --rc                                 Enable the remote control server
      --rc-addr stringArray                IPaddress:Port or :Port to bind server to (default localhost:5572)
      --rc-allow-origin string             Origin which cross-domain request (CORS) can be executed from
      --rc-baseurl string                  Prefix for URLs - leave blank for root
      --rc-cert string                     TLS PEM key (concatenation of certificate and CA certificate)
      --rc-client-ca string                Client certificate authority to verify clients with
      --rc-enable-metrics                  Enable the Prometheus metrics path at the remote control server
      --rc-files string                    Path to local files to serve on the HTTP server
      --rc-htpasswd string                 A htpasswd file - if not provided no authentication is done
      --rc-job-expire-duration Duration    Expire finished async jobs older than this value (default 1m0s)
      --rc-job-expire-interval Duration    Interval to check for expired async jobs (default 10s)
      --rc-key string                      TLS PEM Private key
      --rc-max-header-bytes int            Maximum size of request header (default 4096)
      --rc-min-tls-version string          Minimum TLS version that is acceptable (default "tls1.0")
      --rc-no-auth                         Don't require auth for certain methods
      --rc-pass string                     Password for authentication
      --rc-realm string                    Realm for authentication
      --rc-response-header stringArray     Set HTTP header for all responses, overriding existing values
      --rc-salt string                     Password hashing salt (default "dlPL2MqE")
      --rc-serve                           Enable the serving of remote objects
      --rc-serve-no-modtime                Don't read the modification time (can speed things up)
      --rc-server-read-timeout Duration    Timeout for server reading data (default 1h0m0s)
      --rc-server-write-timeout Duration   Timeout for server writing data (default 1h0m0s)
      --rc-template string                 User-specified template
      --rc-user string                     User name for authentication
      --rc-user-from-header string         User name from a defined HTTP header

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone hashsum

Produces a hashsum file for all the objects in the path.

Synopsis

Produces a hash file for all the objects in the path using the hash named. The output is in the same format as the standard md5sum/sha1sum tool.

By default, the hash is requested from the remote. If the hash is not supported by the remote, no hash will be returned. With the download flag, the file will be downloaded from the remote and hashed locally enabling any hash for any remote.

For the MD5 and SHA1 algorithms there are also dedicated commands, md5sum and sha1sum.

This command can also hash data received on standard input (stdin), by not passing a remote:path, or by passing a hyphen as remote:path when there is data to read (if not, the hyphen will be treated literally, as a relative path).

Run without a hash to see the list of all supported hashes, e.g.

$ rclone hashsum
Supported hashes are:
- md5
- sha1
- whirlpool
- crc32
- sha256
- sha512
- blake3
- xxh3
- xxh128
- crc32c

Then

rclone hashsum MD5 remote:path

Note that hash names are case insensitive and values are output in lower case.

rclone hashsum [<hash> remote:path] [flags]

Options

      --base64               Output base64 encoded hashsum
  -C, --checkfile string     Validate hashes against a given SUM file instead of printing them
      --download             Download the file and hash it locally; if this flag is not specified, the hash is requested from the remote
  -h, --help                 help for hashsum
      --output-file string   Output hashsums to a file rather than the terminal

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone index

Write static directory listings into a remote.

Synopsis

Write a directory listing, index.html by default, into every directory of remote:path so that it can be browsed when it is served as a static website, for example from an S3 website endpoint, a bucket behind a CDN, or a plain web server.

This works like rclone sync: each run makes the minimum changes needed to bring the listings into line with the files. Listings which haven't changed aren't rewritten, so a run with no changes makes no uploads. The listings never show the listing files themselves.

Note that rclone index treats every file with an output name (index.html by default) in the directories it indexes as its own, and will overwrite it or delete it without checking where it came from. To protect a hand-written index, exclude its directory with --index-exclude.

Outputs

--output NAME=FORMAT says what to write in each directory and can be repeated. NAME is the file name, and FORMAT is one of the built-in formats below or the path of a Go template file.

Format MIME type Contents
html text/html; charset=utf-8 The HTML listing from the rclone serve http template
json application/json JSON in the format of rclone lsjson for the directory
caddy application/json JSON in the format of Caddy's file_server browse

The default is --output index.html=html. The built-in formats are templates, so they can be printed with --print-template FORMAT, copied and changed. Outputs whose name ends in .html or .htm are rendered with Go's html/template and everything else with text/template, which has a json function for safe encoding. The template data is that of rclone serve http with .Static set (see rclone serve http). Outputs are uploaded with the MIME type of the built-in format, or that of the output name's extension for template outputs, since static hosts serve the stored type.

Filters

The normal filters (--include, --exclude, --filter-from, --max-age, ...) control what appears in the listings. An excluded file isn't listed, and an excluded directory isn't listed or indexed.

The --index-include, --index-exclude, --index-filter flags and their -from variants control which directories get listings. They take the same patterns as the normal filters, applied to directory paths, and --index-include implies excluding all other directories. A directory excluded here still appears in its parent's listing, but nothing is written or deleted inside it, so it can have its own hand-made index. --index-max-depth limits how deep listings are written.

Modification times

Listings show the modification time of each file. When using the s3, oracleobjectstorage and swift backends rclone stores the modification time as metadata which isn't returned by the bucket listing, so every object needs a HEAD request, which can mean thousands of transactions per run. Avoid this with one of:

  • --use-server-modtime uses the time the object was uploaded instead, which is what the web server sends as Last-Modified. Recommended for s3 unless the original modification times matter.
  • --no-modtime shows no times at all, which is the cheapest option.

--dir-time sets the time shown for directories:

  • newest (default): the newest modification time of anything below the directory. Note that this means one new file changes the listing of every directory above it.
  • dir: the directory's own modification time. On most file systems writing a listing into a directory changes its time, so this can take more than one run to settle.
  • none: no time.

The listing files are given the newest modification time below their directory, so the Last-Modified header of a listing is meaningful.

Links

Directory links are dir/ by default, which works on hosts that serve index documents: S3 website endpoints, GCS and Azure static websites, nginx and Apache, and R2 custom domains with a URL rewrite rule. For hosts that don't (plain S3 REST URLs, B2 friendly URLs) use --link-index to make directory links dir/index.html instead.

Partial runs

After a sync which changed a known set of files there is no need to walk the whole remote. Tell rclone index what changed and it re-indexes only the directories that could be affected:

rclone index r2:bucket --changed v1.76.0/ --changed version.txt
rclone index r2:bucket --changed-from changes.txt
rclone sync ./public r2:bucket --exclude index.html --combined - | rclone index r2:bucket --changed-combined -
  • --changed PATH names a changed file or directory relative to remote:path and can be repeated. A directory, given with a trailing slash or found to be one, is re-indexed completely.
  • --changed-from FILE reads one path per line exactly as written.
  • --changed-combined FILE reads the --combined report of rclone sync: lines starting with +, -, * and ! are changes and = lines are ignored.
  • Both files accept - for standard input and all three flags can be combined.

A changed file re-indexes its directory and every directory above it, since their times and their listings can change too, and on bucket based storage a directory can cease to exist. Each of those is one directory listing. With --dir-time newest the directories above also need the times of their unchanged subdirectories, which are read from the modification times of those subdirectories' listings at the cost of one small request each. With --use-server-modtime that is the time the listing was written rather than the time of the newest file, so partial runs can drift by the indexing delay until the next full run. --dir-time dir and none need no such lookups.

Directories which are neither named nor above a named path are left alone, as are directories deeper than --max-depth. An empty change list does nothing, and naming the root does a full run. There is no automatic cutover to a full run since that depends on the size of the whole remote: a partial run costs one listing per affected directory, a full run one request per thousand objects on S3-like storage or one per directory elsewhere. The number of directories a partial run will list is logged with -v, and --changed-max-dirs makes it fall back to a full run above that number. A good pattern is a partial run after each upload and a scheduled full run to catch anything missed.

Site icon

The listings carry no icon or branding of their own. To give a site an icon, upload a favicon.ico to the root of the remote, which browsers fetch from there for every page. Exclude it from the listings with --exclude /favicon.ico so it stays on the site but isn't shown.

Deleting listings

On backends which can't have empty directories (S3, B2 and other bucket-based storage) a directory only exists while it has files in it. Once the last file in a directory has been removed its listing is deleted, since the directory only existed because of it. Backends which can have empty directories keep a listing for each empty directory.

Using with sync

A sync from a local build deletes files on the destination which aren't in the source, which includes the listings. Either index the local build before syncing:

rclone index ./public
rclone sync ./public r2:bucket

or exclude the listings from the sync so it neither uploads nor deletes them, then index the remote:

rclone sync ./public r2:bucket --exclude index.html
rclone index r2:bucket

Examples

An S3 website bucket, using the upload times so no object needs a HEAD request:

rclone index s3:my-site --use-server-modtime

HTML and rclone JSON listings, setting Cache-Control on them:

rclone index s3:bucket --output index.html=html --output index.json=json --header-upload "Cache-Control: max-age=300"

Only list release files, and don't write listings under /private:

rclone index r2:bucket --include "*.{zip,deb,rpm}" --index-exclude "/private/**"

A B2 bucket served from its friendly URLs:

rclone index b2:bucket --link-index

See what would change:

rclone index r2:bucket --dry-run -v

Listings which haven't changed aren't normally rewritten, so changing the headers or metadata set on them needs --index-rewrite, which writes every listing a run would render whether it changed or not:

rclone index r2:bucket --index-rewrite --header-upload "Cache-Control: max-age=3600"
rclone index remote:path [flags]

Options

      --changed stringArray              Only re-index directories affected by this changed path (directories end in /)
      --changed-combined stringArray     Read changed paths from a sync --combined report (use - to read from stdin)
      --changed-from stringArray         Read changed paths from file, one per line (use - to read from stdin)
      --changed-max-dirs int             Do a full index if a partial one would list more than this many directories (0 for no limit)
      --dir-time newest|dir|none         How to work out the time shown for a directory (default newest)
  -h, --help                             help for index
      --index-exclude stringArray        Don't write listings in directories matching pattern
      --index-exclude-from stringArray   Read directory exclude patterns from file (use - to read from stdin)
      --index-filter stringArray         Add a rule for which directories get listings
      --index-filter-from stringArray    Read directory rules from a file (use - to read from stdin)
      --index-include stringArray        Only write listings in directories matching pattern
      --index-include-from stringArray   Read directory include patterns from file (use - to read from stdin)
      --index-max-depth int              Only write listings this many directories deep (-1 for no limit) (default -1)
      --index-rewrite                    Write every listing even if it is unchanged
      --link-index                       Link to dir/NAME rather than dir/ for hosts without index documents
      --no-modtime                       Don't show modification times in listings
      --output stringArray               Listing to write in each directory as NAME=FORMAT (default [index.html=html])
      --print-template string            Print the built-in template for FORMAT and exit

Options shared with other commands are described next. See the global flags page for global options not listed here.

Copy Options

Flags for anything which can copy a file

      --check-first                                 Do all the checks before starting transfers
  -c, --checksum                                    Check for changes with size & checksum (if available, or fallback to size only)
      --compare-dest stringArray                    Include additional server-side paths during comparison
      --copy-dest stringArray                       Implies --compare-dest but also copies files from paths into destination
      --cutoff-mode HARD|SOFT|CAUTIOUS              Mode to stop transfers when reaching the max transfer limit HARD|SOFT|CAUTIOUS (default HARD)
      --ignore-case-sync                            Ignore case when synchronizing
      --ignore-checksum                             Skip post copy check of checksums
      --ignore-existing                             Skip all files that exist on destination
      --ignore-size                                 Ignore size when skipping use modtime or checksum
  -I, --ignore-times                                Don't skip items that match size and time - transfer all unconditionally
      --immutable                                   Do not modify files, fail if existing files have been modified
      --inplace                                     Download directly to destination file instead of atomic download to temp/rename
  -l, --links                                       Translate symlinks to/from regular files with a '.rclonelink' extension
      --max-backlog int                             Maximum number of objects in sync or check backlog (default 10000)
      --max-duration Duration                       Maximum duration rclone will transfer data for (default 0s)
      --max-transfer SizeSuffix                     Maximum size of data to transfer (default off)
  -M, --metadata                                    If set, preserve metadata when copying objects
      --modify-window Duration                      Max time diff to be considered the same (default 1ns)
      --multi-thread-chunk-size SizeSuffix          Chunk size for multi-thread downloads / uploads, if not set by filesystem (default 64Mi)
      --multi-thread-cutoff SizeSuffix              Use multi-thread downloads for files above this size (default 256Mi)
      --multi-thread-streams int                    Number of streams to use for multi-thread downloads (default 4)
      --multi-thread-write-buffer-size SizeSuffix   In memory buffer size for writing when in multi-thread mode (default 128Ki)
      --name-transform stringArray                  Transform paths during the copy process
      --no-check-dest                               Don't check the destination, copy regardless
      --no-traverse                                 Don't traverse destination file system on copy
      --no-update-dir-modtime                       Don't update directory modification times
      --no-update-modtime                           Don't update destination modtime if files identical
      --order-by string                             Instructions on how to order the transfers, e.g. 'size,descending'
      --partial-suffix string                       Add partial-suffix to temporary file name when --inplace is not used (default ".partial")
      --refresh-times                               Refresh the modtime of remote files
      --server-side-across-configs                  Allow server-side operations (e.g. copy) to work across different configs
      --size-only                                   Skip based on size only, not modtime or checksum
      --streaming-upload-cutoff SizeSuffix          Cutoff for switching to chunked upload if file size is unknown, upload starts after reaching cutoff or when file ends (default 100Ki)
  -u, --update                                      Skip files that are newer on the destination

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone link

Generate public link to file/folder.

Synopsis

Create, retrieve or remove a public link to the given file or folder.

rclone link remote:path/to/file
rclone link remote:path/to/folder/
rclone link --unlink remote:path/to/folder/
rclone link --expire 1d remote:path/to/file

If you supply the --expire flag, it will set the expiration time otherwise it will use the default (100 years). Note not all backends support the --expire flag - if the backend doesn't support it then the link returned won't expire.

Use the --unlink flag to remove existing public links to the file or folder. Note not all backends support "--unlink" flag - those that don't will just ignore it.

If successful, the last line of the output will contain the link. Exact capabilities depend on the remote, but the link will always by default be created with the least constraints - e.g. no expiry, no password protection, accessible without account.

rclone link remote:path [flags]

Options

      --expire Duration   The amount of time that the link will be valid (default off)
  -h, --help              help for link
      --unlink            Remove existing public link to file/folder

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone listremotes

List all the remotes in the config file and defined in environment variables.

Synopsis

Lists all the available remotes from the config file, or the remotes matching an optional filter.

Prints the result in human-readable format by default, and as a simple list of remote names, or if used with flag --long a tabular format including the remote names, types and descriptions. Using flag --json produces machine-readable output instead, which always includes all attributes - including the source (file or environment).

Result can be filtered by a filter argument which applies to all attributes, and/or filter flags specific for each attribute. The values must be specified according to regular rclone filtering pattern syntax.

By default filtering uses non-anchored matching, so --type box also matches dropbox. Use --exact to match complete values only.

rclone listremotes [<filter>] [flags]

Options

      --description string   Filter remotes by description
      --exact                Match filter strings exactly instead of using non-anchored glob matching
  -h, --help                 help for listremotes
      --json                 Format output as JSON
      --long                 Show type and description in addition to name
      --name string          Filter remotes by name
      --order-by string      Instructions on how to order the result, e.g. 'type,name=descending'
      --source string        Filter remotes by source, e.g. 'file' or 'environment'
      --type string          Filter remotes by type

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone lsf

List directories and objects in remote:path formatted for parsing.

Synopsis

List the contents of the source path (directories and objects) to standard output in a form which is easy to parse by scripts. By default this will just be the names of the objects and directories, one per line. The directories will have a / suffix.

E.g.

$ rclone lsf swift:bucket
bevajer5jef
canole
diwogej7
ferejej3gux/
fubuwic

Use the --format option to control what gets listed. By default this is just the path, but you can use these parameters to control the output:

p - path
s - size
t - modification time
h - hash
i - ID of object
o - Original ID of underlying object
m - MimeType of object if known
e - encrypted name
T - tier of storage if known, e.g. "Hot" or "Cool"
M - Metadata of object in JSON blob format, eg {"key":"value"}

Some backends use composite IDs. In particular, Google Drive shortcuts are normally dereferenced, and their "i" field contains the target object's ID followed by the shortcut's ID, separated by a tab. This tab is part of the ID field and is not changed by --separator. Use --drive-skip-shortcuts to omit shortcut entries.

So if you wanted the path, size and modification time, you would use --format "pst", or maybe --format "tsp" to put the path last.

E.g.

$ rclone lsf  --format "tsp" swift:bucket
2016-06-25 18:55:41;60295;bevajer5jef
2016-06-25 18:55:43;90613;canole
2016-06-25 18:55:43;94467;diwogej7
2018-04-26 08:50:45;0;ferejej3gux/
2016-06-25 18:55:40;37600;fubuwic

If you specify "h" in the format you will get the MD5 hash by default, use the --hash flag to change which hash you want. Note that this can be returned as an empty string if it isn't available on the object (and for directories), "ERROR" if there was an error reading it from the object and "UNSUPPORTED" if that object does not support that hash type.

For example, to emulate the md5sum command you can use

rclone lsf -R --hash MD5 --format hp --separator "  " --files-only .

E.g.

$ rclone lsf -R --hash MD5 --format hp --separator "  " --files-only swift:bucket
7908e352297f0f530b84a756f188baa3  bevajer5jef
cd65ac234e6fea5925974a51cdd865cc  canole
03b5341b4f234b9d984d03ad076bae91  diwogej7
8fd37c3810dd660778137ac3a66cc06d  fubuwic
99713e14a4c4ff553acaf1930fad985b  gixacuh7ku

(Though "rclone md5sum ." is an easier way of typing this.)

By default the separator is ";" this can be changed with the --separator flag. Note that separators aren't escaped in the path so putting it last is a good strategy.

E.g.

$ rclone lsf  --separator "," --format "tshp" swift:bucket
2016-06-25 18:55:41,60295,7908e352297f0f530b84a756f188baa3,bevajer5jef
2016-06-25 18:55:43,90613,cd65ac234e6fea5925974a51cdd865cc,canole
2016-06-25 18:55:43,94467,03b5341b4f234b9d984d03ad076bae91,diwogej7
2018-04-26 08:52:53,0,,ferejej3gux/
2016-06-25 18:55:40,37600,8fd37c3810dd660778137ac3a66cc06d,fubuwic

You can output in CSV standard format. This will escape things in " if they contain,

E.g.

$ rclone lsf --csv --files-only --format ps remote:path
test.log,22355
test.sh,449
"this file contains a comma, in the file name.txt",6

Note that the --absolute parameter is useful for making lists of files to pass to an rclone copy with the --files-from-raw flag.

For example, to find all the files modified within one day and copy those only (without traversing the whole directory structure):

rclone lsf --absolute --files-only --max-age 1d /path/to/local > new_files
rclone copy --files-from-raw new_files /path/to/local remote:path

The default time format is '2006-01-02 15:04:05'. Other formats can be specified with the --time-format flag. Examples:

rclone lsf remote:path --format pt --time-format 'Jan 2, 2006 at 3:04pm (MST)'
rclone lsf remote:path --format pt --time-format '2006-01-02 15:04:05.000000000'
rclone lsf remote:path --format pt --time-format '2006-01-02T15:04:05.999999999Z07:00'
rclone lsf remote:path --format pt --time-format RFC3339
rclone lsf remote:path --format pt --time-format DateOnly
rclone lsf remote:path --format pt --time-format max
rclone lsf remote:path --format pt --time-format unix
rclone lsf remote:path --format pt --time-format unixnano

--time-format max will automatically truncate 2006-01-02 15:04:05.000000000 to the maximum precision supported by the remote.

Any of the filtering options can be applied to this command.

There are several related list commands

  • ls to list size and path of objects only
  • lsl to list modification time, size and path of objects only
  • lsd to list directories only
  • lsf to list objects and directories in easy to parse format
  • lsjson to list objects and directories in JSON format

ls,lsl,lsd are designed to be human-readable. lsf is designed to be human and machine-readable. lsjson is designed to be machine-readable.

Note that ls and lsl recurse by default - use --max-depth 1 to stop the recursion.

The other list commands lsd,lsf,lsjson do not recurse by default - use -R to make them recurse.

On the bucket-based remotes (e.g. s3, gcs, b2) a directory is a key prefix rather than an object, so a listing without -R shows each prefix as one directory entry and not the objects stored under it.

List commands prefer a recursive method that uses more memory but fewer transactions by default. Use --disable ListR to suppress the behavior. See --fast-list for more details.

Listing a nonexistent directory will produce an error except for remotes which can't have empty directories (e.g. s3, swift, or gcs - the bucket-based remotes).

rclone lsf remote:path [flags]

Options

      --absolute             Put a leading / in front of path names
      --csv                  Output in CSV format
  -d, --dir-slash            Append a slash to directory names (default true)
      --dirs-only            Only list directories
      --files-only           Only list files
  -F, --format string        Output format - see  help for details (default "p")
      --hash h               Use this hash when h is used in the format MD5|SHA-1|DropboxHash (default "md5")
  -h, --help                 help for lsf
  -R, --recursive            Recurse into the listing
  -s, --separator string     Separator for the items in the format (default ";")
  -t, --time-format string   Specify a custom time format - see docs for details (default: 2006-01-02 15:04:05)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone lsjson

List directories and objects in the path in JSON format.

Synopsis

List directories and objects in the path in JSON format.

The output is an array of Items, where each Item looks like this:

{
  "Hashes" : {
    "SHA-1" : "f572d396fae9206628714fb2ce00f72e94f2258f",
    "MD5" : "b1946ac92492d2347c6235b4d2611184",
    "DropboxHash" : "ecb65bb98f9d905b70458986c39fcbad7715e5f2fcc3b1f07767d7c83e2438cc"
  },
  "ID": "y2djkhiujf83u33",
  "OrigID": "UYOJVTUW00Q1RzTDA",
  "IsBucket" : false,
  "IsDir" : false,
  "MimeType" : "application/octet-stream",
  "ModTime" : "2017-05-31T16:15:57.034468261+01:00",
  "Name" : "file.txt",
  "Encrypted" : "v0qpsdq8anpci8n929v3uu9338",
  "EncryptedPath" : "kja9098349023498/v0qpsdq8anpci8n929v3uu9338",
  "Path" : "full/path/goes/here/file.txt",
  "Size" : 6,
  "Tier" : "hot",
}

The exact set of properties included depends on the backend:

  • The property IsBucket will only be included for bucket-based remotes, and only for directories that are buckets. It will always be omitted when value is not true.
  • Properties Encrypted and EncryptedPath will only be included for encrypted remotes, and (as mentioned below) only if the --encrypted option is set.

Different options may also affect which properties are included:

  • If --hash is not specified, the Hashes property will be omitted. The types of hash can be specified with the --hash-type parameter (which may be repeated). If --hash-type is set then it implies --hash.
  • If --no-modtime is specified then ModTime will be blank. This can speed things up on remotes where reading the ModTime takes an extra request (e.g. s3, swift).
  • If --no-mimetype is specified then MimeType will be blank. This can speed things up on remotes where reading the MimeType takes an extra request (e.g. s3, swift).
  • If --encrypted is not specified the Encrypted and EncryptedPath properties will be omitted - even for encrypted remotes.
  • If --metadata is set then an additional Metadata property will be returned. This will have metadata in rclone standard format as a JSON object.

The default is to list directories and files/objects, but this can be changed with the following options:

  • If --dirs-only is specified then directories will be returned only, no files/objects.
  • If --files-only is specified then files will be returned only, no directories.

If --stat is set then the output is not an array of items, but instead a single JSON blob will be returned about the item pointed to. This will return an error if the item isn't found, however on bucket based backends (like s3, gcs, b2, azureblob etc) if the item isn't found it will return an empty directory, as it isn't possible to tell empty directories from missing directories there.

The Path field will only show folders below the remote path being listed. If "remote:path" contains the file "subfolder/file.txt", the Path for "file.txt" will be "subfolder/file.txt", not "remote:path/subfolder/file.txt". When used without --recursive the Path will always be the same as Name.

The time is in RFC3339 format with up to nanosecond precision. The number of decimal digits in the seconds will depend on the precision that the remote can hold the times, so if times are accurate to the nearest millisecond (e.g. Google Drive) then 3 digits will always be shown ("2017-05-31T16:15:57.034+01:00") whereas if the times are accurate to the nearest second (Dropbox, Box, WebDav, etc.) no digits will be shown ("2017-05-31T16:15:57+01:00").

The whole output can be processed as a JSON blob, or alternatively it can be processed line by line as each item is written on individual lines (except with --stat).

Any of the filtering options can be applied to this command.

There are several related list commands

  • ls to list size and path of objects only
  • lsl to list modification time, size and path of objects only
  • lsd to list directories only
  • lsf to list objects and directories in easy to parse format
  • lsjson to list objects and directories in JSON format

ls,lsl,lsd are designed to be human-readable. lsf is designed to be human and machine-readable. lsjson is designed to be machine-readable.

Note that ls and lsl recurse by default - use --max-depth 1 to stop the recursion.

The other list commands lsd,lsf,lsjson do not recurse by default - use -R to make them recurse.

On the bucket-based remotes (e.g. s3, gcs, b2) a directory is a key prefix rather than an object, so a listing without -R shows each prefix as one directory entry and not the objects stored under it.

List commands prefer a recursive method that uses more memory but fewer transactions by default. Use --disable ListR to suppress the behavior. See --fast-list for more details.

Listing a nonexistent directory will produce an error except for remotes which can't have empty directories (e.g. s3, swift, or gcs - the bucket-based remotes).

rclone lsjson remote:path [flags]

Options

      --dirs-only               Show only directories in the listing
      --encrypted               Show the encrypted names
      --files-only              Show only files in the listing
      --hash                    Include hashes in the output (may take longer)
      --hash-type stringArray   Show only this hash type (may be repeated)
  -h, --help                    help for lsjson
  -M, --metadata                Add metadata to the listing
      --no-mimetype             Don't read the mime type (can speed things up)
      --no-modtime              Don't read the modification time (can speed things up)
      --original                Show the ID of the underlying Object
  -R, --recursive               Recurse into the listing
      --stat                    Just return the info for the pointed to file

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone mount

Mount the remote as file system on a mountpoint.

Synopsis

Rclone mount allows Linux, FreeBSD, macOS and Windows to mount any of Rclone's cloud storage systems as a file system with FUSE.

First set up your remote using rclone config. Check it works with rclone ls etc.

On Linux and macOS, you can run mount in either foreground or background (aka daemon) mode. Mount runs in foreground mode by default. Use the --daemon flag to force background mode. On Windows you can run mount in foreground only, the flag is ignored.

In background mode rclone acts as a generic Unix mount program: the main program starts, spawns background rclone process to setup and maintain the mount, waits until success or timeout and exits with appropriate code (killing the child process if it fails).

On Linux/macOS/FreeBSD start the mount like this, where /path/to/local/mount is an empty existing directory:

rclone mount remote:path/to/files /path/to/local/mount

On Windows you can start a mount in different ways. See below for details. If foreground mount is used interactively from a console window, rclone will serve the mount and occupy the console so another window should be used to work with the mount until rclone is interrupted e.g. by pressing Ctrl-C.

The following examples will mount to an automatically assigned drive, to specific drive letter X:, to path C:\path\parent\mount (where parent directory or drive must exist, and mount must not exist, and is not supported when mounting as a network drive), and the last example will mount as network share \\cloud\remote and map it to an automatically assigned drive:

rclone mount remote:path/to/files *
rclone mount remote:path/to/files X:
rclone mount remote:path/to/files C:\path\parent\mount
rclone mount remote:path/to/files \\cloud\remote

When the program ends while in foreground mode, either via Ctrl+C or receiving a SIGINT or SIGTERM signal, the mount should be automatically stopped.

When running in background mode the user will have to stop the mount manually:

# Linux
fusermount -u /path/to/local/mount
#... or on some systems
fusermount3 -u /path/to/local/mount
# OS X or Linux when using nfsmount
umount /path/to/local/mount

The umount operation can fail, for example when the mountpoint is busy. When that happens, it is the user's responsibility to stop the mount manually.

The size of the mounted file system will be set according to information retrieved from the remote, the same as returned by the rclone about command. Remotes with unlimited storage may report the used size only, then an additional 1 PiB of free space is assumed. If the remote does not support the about feature at all, then 1 PiB is set as both the total and the free size.

Installing on Windows

To run rclone mount on Windows, you will need to download and install WinFsp.

WinFsp is an open-source Windows File System Proxy which makes it easy to write user space file systems for Windows. It provides a FUSE emulation layer which rclone uses combination with cgofuse. Both of these packages are by Bill Zissimopoulos who was very helpful during the implementation of rclone mount for Windows.

Mounting modes on windows

Unlike other operating systems, Microsoft Windows provides a different filesystem type for network and fixed drives. It optimises access on the assumption fixed disk drives are fast and reliable, while network drives have relatively high latency and less reliability. Some settings can also be differentiated between the two types, for example that Windows Explorer should just display icons and not create preview thumbnails for image and video files on network drives.

In most cases, rclone will mount the remote as a normal, fixed disk drive by default. However, you can also choose to mount it as a remote network drive, often described as a network share. If you mount an rclone remote using the default, fixed drive mode and experience unexpected program errors, freezes or other issues, consider mounting as a network drive instead.

When mounting as a fixed disk drive you can either mount to an unused drive letter, or to a path representing a nonexistent subdirectory of an existing parent directory or drive. Using the special value * will tell rclone to automatically assign the next available drive letter, starting with Z: and moving backward. Examples:

rclone mount remote:path/to/files *
rclone mount remote:path/to/files X:
rclone mount remote:path/to/files C:\path\parent\mount
rclone mount remote:path/to/files X:

Option --volname can be used to set a custom volume name for the mounted file system. The default is to use the remote name and path.

To mount as network drive, you can add option --network-mode to your mount command. Mounting to a directory path is not supported in this mode, it is a limitation Windows imposes on junctions, so the remote must always be mounted to a drive letter.

rclone mount remote:path/to/files X: --network-mode

A volume name specified with --volname will be used to create the network share path. A complete UNC path, such as \\cloud\remote, optionally with path \\cloud\remote\madeup\path, will be used as is. Any other string will be used as the share part, after a default prefix \\server\. If no volume name is specified then \\server\share will be used. You must make sure the volume name is unique when you are mounting more than one drive, or else the mount command will fail. The share name will treated as the volume label for the mapped drive, shown in Windows Explorer etc, while the complete \\server\share will be reported as the remote UNC path by net use etc, just like a normal network drive mapping.

If you specify a full network share UNC path with --volname, this will implicitly set the --network-mode option, so the following two examples have same result:

rclone mount remote:path/to/files X: --network-mode
rclone mount remote:path/to/files X: --volname \\server\share

You may also specify the network share UNC path as the mountpoint itself. Then rclone will automatically assign a drive letter, same as with * and use that as mountpoint, and instead use the UNC path specified as the volume name, as if it were specified with the --volname option. This will also implicitly set the --network-mode option. This means the following two examples have same result:

rclone mount remote:path/to/files \\cloud\remote
rclone mount remote:path/to/files * --volname \\cloud\remote

There is yet another way to enable network mode, and to set the share path, and that is to pass the "native" libfuse/WinFsp option directly: --fuse-flag --VolumePrefix=\server\share. Note that the path must be with just a single backslash prefix in this case.

Note: In previous versions of rclone this was the only supported method.

Read more about drive mapping

See also Limitations section below.

Windows filesystem permissions

The FUSE emulation layer on Windows must convert between the POSIX-based permission model used in FUSE, and the permission model used in Windows, based on access-control lists (ACL).

The mounted filesystem will normally get three entries in its access-control list (ACL), representing permissions for the POSIX permission scopes: Owner, group and others. By default, the owner and group will be taken from the current user, and the built-in group "Everyone" will be used to represent others. The user/group can be customized with FUSE options "UserName" and "GroupName", e.g. -o UserName=user123 -o GroupName="Authenticated Users". The permissions on each entry will be set according to options --dir-perms and --file-perms, which takes a value in traditional Unix numeric notation.

The default permissions corresponds to --file-perms 0666 --dir-perms 0777, i.e. read and write permissions to everyone. This means you will not be able to start any programs from the mount. To be able to do that you must add execute permissions, e.g. --file-perms 0777 --dir-perms 0777 to add it to everyone. If the program needs to write files, chances are you will have to enable VFS File Caching as well (see also limitations). Note that the default write permission have some restrictions for accounts other than the owner, specifically it lacks the "write extended attributes", as explained next.

The mapping of permissions is not always trivial, and the result you see in Windows Explorer may not be exactly like you expected. For example, when setting a value that includes write access for the group or others scope, this will be mapped to individual permissions "write attributes", "write data" and "append data", but not "write extended attributes". Windows will then show this as basic permission "Special" instead of "Write", because "Write" also covers the "write extended attributes" permission. When setting digit 0 for group or others, to indicate no permissions, they will still get individual permissions "read attributes", "read extended attributes" and "read permissions". This is done for compatibility reasons, e.g. to allow users without additional permissions to be able to read basic metadata about files like in Unix.

WinFsp 2021 (version 1.9) introduced a new FUSE option "FileSecurity", that allows the complete specification of file security descriptors using SDDL. With this you get detailed control of the resulting permissions, compared to use of the POSIX permissions described above, and no additional permissions will be added automatically for compatibility with Unix. Some example use cases will following.

If you set POSIX permissions for only allowing access to the owner, using --file-perms 0600 --dir-perms 0700, the user group and the built-in "Everyone" group will still be given some special permissions, as described above. Some programs may then (incorrectly) interpret this as the file being accessible by everyone, for example an SSH client may warn about "unprotected private key file". You can work around this by specifying -o FileSecurity="D:P(A;;FA;;;OW)", which sets file all access (FA) to the owner (OW), and nothing else.

When setting write permissions then, except for the owner, this does not include the "write extended attributes" permission, as mentioned above. This may prevent applications from writing to files, giving permission denied error instead. To set working write permissions for the built-in "Everyone" group, similar to what it gets by default but with the addition of the "write extended attributes", you can specify -o FileSecurity="D:P(A;;FRFW;;;WD)", which sets file read (FR) and file write (FW) to everyone (WD). If file execute (FX) is also needed, then change to -o FileSecurity="D:P(A;;FRFWFX;;;WD)", or set file all access (FA) to get full access permissions, including delete, with -o FileSecurity="D:P(A;;FA;;;WD)".

Windows caveats

Drives created as Administrator are not visible to other accounts, not even an account that was elevated to Administrator with the User Account Control (UAC) feature. A result of this is that if you mount to a drive letter from a Command Prompt run as Administrator, and then try to access the same drive from Windows Explorer (which does not run as Administrator), you will not be able to see the mounted drive.

If you don't need to access the drive from applications running with administrative privileges, the easiest way around this is to always create the mount from a non-elevated command prompt.

To make mapped drives available to the user account that created them regardless if elevated or not, there is a special Windows setting called linked connections that can be enabled.

It is also possible to make a drive mount available to everyone on the system, by running the process creating it as the built-in SYSTEM account. There are several ways to do this: One is to use the command-line utility PsExec, from Microsoft's Sysinternals suite, which has option -s to start processes as the SYSTEM account. Another alternative is to run the mount command from a Windows Scheduled Task, or a Windows Service, configured to run as the SYSTEM account. A third alternative is to use the WinFsp.Launcher infrastructure). Read more in the install documentation. Note that when running rclone as another user, it will not use the configuration file from your profile unless you tell it to with the --config option. Note also that it is now the SYSTEM account that will have the owner permissions, and other accounts will have permissions according to the group or others scopes. As mentioned above, these will then not get the "write extended attributes" permission, and this may prevent writing to files. You can work around this with the FileSecurity option, see example above.

Note that mapping to a directory path, instead of a drive letter, does not suffer from the same limitations.

Mounting on macOS

Mounting on macOS can be done either via built-in NFS server, macFUSE (also known as osxfuse) or FUSE-T. macFUSE is a traditional FUSE driver utilizing a macOS kernel extension (kext). FUSE-T is an alternative FUSE system which "mounts" via an NFSv4 local server.

Unicode Normalization

It is highly recommended to keep the default of --no-unicode-normalization=false for all mount and serve commands on macOS. For details, see vfs-case-sensitivity.

NFS mount

For macOS (and other platforms where this path is supported), prefer the dedicated rclone nfsmount command. It starts the NFS server and performs the mount for you. rclone mount itself still uses FUSE (macFUSE/FUSE-T) and does not switch to NFS via a flag.

This method spins up an NFS server using serve nfs command and mounts it to the specified mountpoint. If you run this in background mode using |--daemon|, you will need to send SIGTERM signal to the rclone process using |kill| command to stop the mount.

Note that --nfs-cache-handle-limit controls the maximum number of cached file handles stored by the nfsmount caching handler. This should not be set too low or you may experience errors when trying to access files. The default is 1000000, but consider lowering this limit if the server's system resource usage causes problems.

macFUSE Notes

If installing macFUSE using dmg packages from the website, rclone will locate the macFUSE libraries without any further intervention. If however, macFUSE is installed using the macports package manager, the following addition steps are required.

sudo mkdir /usr/local/lib
cd /usr/local/lib
sudo ln -s /opt/local/lib/libfuse.2.dylib

FUSE-T Limitations, Caveats, and Notes

There are some limitations, caveats, and notes about how it works. These are current as of FUSE-T version 1.0.14.

ModTime update on read

As per the FUSE-T wiki:

File access and modification times cannot be set separately as it seems to be an issue with the NFS client which always modifies both. Can be reproduced with touch -m and touch -a commands

This means that viewing files with various tools, notably macOS Finder, will cause rlcone to update the modification time of the file. This may make rclone upload a full new copy of the file.

Read Only mounts

When mounting with --read-only, attempts to write to files will fail silently as opposed to with a clear warning as in macFUSE.

Mounting on Linux

On newer versions of Ubuntu, you may encounter the following error when running rclone mount:

NOTICE: mount helper error: fusermount3: mount failed: Permission denied CRITICAL: Fatal error: failed to mount FUSE fs: fusermount: exit status 1

This may be due to newer Apparmor restrictions, which can be disabled with sudo aa-disable /usr/bin/fusermount3 (you may need to sudo apt install apparmor-utils beforehand).

Mounting on OpenBSD

rclone nfsmount works on OpenBSD by spinning up the serve nfs server and mounting it with the system mount_nfs(8), the same approach used for the NFS mount method on macOS. There is no FUSE-based mount option on OpenBSD, so nfsmount is the only way to mount rclone remotes as a local file system there.

portmap registration

OpenBSD's kernel NFS client locates mountd through the system portmapper rather than connecting to a fixed port, so rclone's in-process NFS server needs to be registered with portmap before it can be mounted. Start the serve nfs server on a fixed port, stop the system NFS services so they don't conflict with it, then register that port with portmap for both the nfs and mountd RPC programs (this is the recipe from hjicks in #8578):

rclone serve nfs --addr localhost:<PORT> remote:path

doas rcctl stop nfsd portmap
doas pkill -9 mountd   # mountd can't be stopped via rcctl(8)
doas rcctl -f start portmap
doas rpcinfo -s nfs 3 <PORT>
doas rpcinfo -s mountd 3 <PORT>

Replace <PORT> with the port passed to --addr. Port 2049 works too, but needs either running rclone as root or adjusting sysctl(8) at securelevel 0, so a non-privileged port is simpler for testing.

Use mount_nfs -T, not mount

Mount the exported NFS share directly with mount_nfs(8) using the -T flag to force TCP, rather than the generic mount(8) wrapper - OpenBSD's mount(8) does not understand -T and rejects it with unknown option -- T:

mount_nfs -T localhost:/ /path/to/local/mount

rclone nfsmount does this for you automatically on OpenBSD (it calls mount_nfs directly with -T instead of going through mount), so this is only needed if you are mounting the serve nfs export by hand.

AUTH_UNIX and existing limitations

OpenBSD's (and the other BSDs') kernel NFS client requires the server to offer the AUTH_UNIX authentication flavor during the MOUNT RPC handshake; a server that only offers AUTH_NULL is rejected with mount_nfs: can't access /: Authentication error. rclone's NFS server advertises AUTH_UNIX for this reason, though it does not implement any per-user access control on top of it - see issue #8578 for the background on this and for the current state of OpenBSD support.

Limitations

Without the use of --vfs-cache-mode this can only write files sequentially, it can only seek when reading. This means that many applications won't work with their files on an rclone mount without --vfs-cache-mode writes or --vfs-cache-mode full. See the VFS File Caching section for more info. When using NFS mount on macOS, if you don't specify |--vfs-cache-mode| the mount point will be read-only.

Bucket-based remotes - Azure Blob, Swift, S3, Google Cloud Storage and B2 - can't store empty directories. Of these, only Azure Blob, Google Cloud Storage and S3 can preserve them when you add --xxx-directory_markers; otherwise, empty directories will vanish once they drop out of the directory cache.

When rclone mount is invoked on Unix with --daemon flag, the main rclone program will wait for the background mount to become ready or until the timeout specified by the --daemon-wait flag. On Linux it can check mount status using ProcFS so the flag in fact sets maximum time to wait, while the real wait can be less. On macOS / BSD the time to wait is constant and the check is performed only at the end. We advise you to set wait time on macOS reasonably.

Only supported on Linux, FreeBSD, OS X and Windows at the moment.

rclone mount vs rclone sync/copy

File systems expect things to be 100% reliable, whereas cloud storage systems are a long way from 100% reliable. The rclone sync/copy commands cope with this with lots of retries. However rclone mount can't use retries in the same way without making local copies of the uploads. Look at the VFS File Caching for solutions to make mount more reliable.

Attribute caching

You can use the flag --attr-timeout to set the time the kernel caches the attributes (size, modification time, etc.) for directory entries.

The default is 1s which caches files just long enough to avoid too many callbacks to rclone from the kernel.

In theory 0s should be the correct value for filesystems which can change outside the control of the kernel. However this causes quite a few problems such as rclone using too much memory, rclone not serving files to samba and excessive time listing directories.

The kernel can cache the info about a file for the time given by --attr-timeout. You may see corruption if the remote file changes length during this window. It will show up as either a truncated file or a file with garbage on the end. With --attr-timeout 1s this is very unlikely but not impossible. The higher you set --attr-timeout the more likely it is. The default setting of "1s" is the lowest setting which mitigates the problems above.

If you set it higher (10s or 1m say) then the kernel will call back to rclone less often making it more efficient, however there is more chance of the corruption issue above.

If files don't change on the remote outside of the control of rclone then there is no chance of corruption.

This is the same as setting the attr_timeout option in mount.fuse.

Filters

Note that all the rclone filters can be used to select a subset of the files to be visible in the mount.

systemd

When running rclone mount as a systemd service, it is possible to use Type=notify. In this case the service will enter the started state after the mountpoint has been successfully set up. Units having the rclone mount service specified as a requirement will see all files and folders immediately in this mode.

Note that systemd runs mount units without any environment variables including PATH or HOME. This means that tilde (~) expansion will not work and you should provide --config and --cache-dir explicitly as absolute paths via rclone arguments. Since mounting requires the fusermount or fusermount3 program, rclone will use the fallback PATH of /bin:/usr/bin in this scenario. Please ensure that fusermount/fusermount3 is present on this PATH.

Rclone as Unix mount helper

The core Unix program /bin/mount normally takes the -t FSTYPE argument then runs the /sbin/mount.FSTYPE helper program passing it mount options as -o key=val,... or --opt=.... Automount (classic or systemd) behaves in a similar way.

rclone by default expects GNU-style flags --key val. To run it as a mount helper you should symlink rclone binary to /sbin/mount.rclone and optionally /usr/bin/rclonefs, e.g. ln -s /usr/bin/rclone /sbin/mount.rclone. rclone will detect it and translate command-line arguments appropriately.

Now you can run classic mounts like this:

mount sftp1:subdir /mnt/data -t rclone -o vfs_cache_mode=writes,sftp_key_file=/path/to/pem

or create systemd mount units:

# /etc/systemd/system/mnt-data.mount
[Unit]
Description=Mount for /mnt/data
[Mount]
Type=rclone
What=sftp1:subdir
Where=/mnt/data
Options=rw,_netdev,allow_other,args2env,vfs-cache-mode=writes,config=/etc/rclone.conf,cache-dir=/var/rclone

optionally accompanied by systemd automount unit

# /etc/systemd/system/mnt-data.automount
[Unit]
Description=AutoMount for /mnt/data
[Automount]
Where=/mnt/data
TimeoutIdleSec=600
[Install]
WantedBy=multi-user.target

or add in /etc/fstab a line like

sftp1:subdir /mnt/data rclone rw,noauto,nofail,_netdev,x-systemd.automount,args2env,vfs_cache_mode=writes,config=/etc/rclone.conf,cache_dir=/var/cache/rclone 0 0

or use classic Automountd. Remember to provide explicit config=...,cache-dir=... as a workaround for mount units being run without HOME.

Rclone in the mount helper mode will split -o argument(s) by comma, replace _ by - and prepend -- to get the command-line flags. Options containing commas or spaces can be wrapped in single or double quotes. Any inner quotes inside outer quotes of the same type should be doubled.

Mount option syntax includes a few extra options treated specially:

  • env.NAME=VALUE will set an environment variable for the mount process. This helps with Automountd and Systemd.mount which don't allow setting custom environment for mount helpers. Typically you will use env.HTTPS_PROXY=proxy.host:3128 or env.HOME=/root
  • command=cmount can be used to run cmount or any other rclone command rather than the default mount.
  • args2env will pass mount options to the mount helper running in background via environment variables instead of command line arguments. This allows to hide secrets from such commands as ps or pgrep.
  • vv... will be transformed into appropriate --verbose=N
  • standard mount options like x-systemd.automount, _netdev, nosuid and alike are intended only for Automountd and ignored by rclone.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

rclone mount remote:path /path/to/mountpoint [flags]

Options

      --allow-idmap                            Allow id-mapped mounts (Linux 6.12+, mount2 only)
      --allow-non-empty                        Allow mounting over a non-empty directory (not supported on Windows)
      --allow-other                            Allow access to other users (not supported on Windows)
      --allow-root                             Allow access to root user (not supported on Windows)
      --async-read                             Use asynchronous reads (not supported on Windows) (default true)
      --attr-timeout Duration                  Time for which file/directory attributes are cached (default 1s)
      --daemon                                 Run mount in background and exit parent process (as background output is suppressed, use --log-file with --log-format=pid,... to monitor) (not supported on Windows)
      --daemon-timeout Duration                Time limit for rclone to respond to kernel (not supported on Windows) (default 0s)
      --daemon-wait Duration                   Time to wait for ready mount from daemon (maximum time on Linux, constant sleep time on OSX/BSD) (not supported on Windows) (default 1m0s)
      --debug-fuse                             Debug the FUSE internals - needs -v
      --default-permissions                    Makes kernel enforce access control based on the file mode (not supported on Windows)
      --devname string                         Set the device name - default is remote:path
      --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                     Directory permissions (default 777)
      --direct-io                              Use Direct IO, disables caching of data
      --file-perms FileMode                    File permissions (default 666)
      --fuse-flag stringArray                  Flags or arguments to be passed direct to libfuse/WinFsp (repeat if required)
      --gid uint32                             Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                   help for mount
      --link-perms FileMode                    Link permissions (default 666)
      --max-read-ahead SizeSuffix              The number of bytes that can be prefetched for sequential reads (not supported on Windows) (default 128Ki)
      --mount-case-insensitive Tristate        Tell the OS the mount is case insensitive (true) or sensitive (false) regardless of the backend (auto) (default unset)
      --network-mode                           Mount as remote network drive, instead of fixed disk drive (supported on Windows only)
      --no-checksum                            Don't compare checksums on up/download
      --no-modtime                             Don't read/write the modification time (can speed things up)
      --no-seek                                Don't allow seeking in files
      --noappledouble                          Ignore Apple Double (._) and .DS_Store files (supported on OSX only) (default true)
      --noapplexattr                           Ignore all "com.apple.*" extended attributes (supported on OSX only)
  -o, --option stringArray                     Option for libfuse/WinFsp (repeat if required)
      --poll-interval Duration                 Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --read-only                              Only allow read-only access
      --uid uint32                             Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                         Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --vfs-block-norm-dupes                   If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration             Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix          Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix    Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode               Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration       Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                   If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix   Specify the total space of disk (default off)
      --vfs-fast-fingerprint                   Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration            Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                              Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string          Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix              Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix         Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix   If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int             The number of parallel streams to read at once
      --vfs-read-wait Duration                 Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                            Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size           Use the rclone size algorithm for Used size
      --vfs-write-back Duration                Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                Time to wait for in-sequence write before giving error (default 1s)
      --volname string                         Set the volume name (supported on Windows and OSX only)
      --write-back-cache                       Makes kernel buffer writes before sending them to rclone (without this, writethrough caching is used) (not supported on Windows)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone moveto

Move file or directory from source to dest.

Synopsis

If source:path is a file or directory then it moves it to a file or directory named dest:path.

This can be used to rename files or upload single files to other than their existing name. If the source is a directory then it acts exactly like the move command.

So

rclone moveto src dst

where src and dst are rclone paths, either remote:path or /path/to/local or C:\windows\path\if\on\windows.

This will:

if src is file
    move it to dst, overwriting an existing file if it exists
if src is directory
    move it to dst, overwriting existing files if they exist
    see move command for full details

This doesn't transfer files that are identical on src and dst, testing by size and modification time or MD5SUM. src will be deleted on successful transfer.

Important: Since this can cause data loss, test first with the --dry-run or the --interactive/-i flag.

Note: Use the -P/--progress flag to view real-time transfer statistics.

Logger Flags

The --differ, --missing-on-dst, --missing-on-src, --match and --error flags write paths, one per line, to the file name (or stdout if it is -) supplied. What they write is described in the help below. For example --differ will write all paths which are present on both the source and destination but different.

The --combined flag will write a file (or stdout) which contains all file paths with a symbol and then a space and then the path to tell you what happened to it. These are reminiscent of diff files.

  • = path means path was found in source and destination and was identical
  • - path means path was missing on the source, so only in the destination
  • + path means path was missing on the destination, so only in the source
  • * path means path was present in source and destination but different.
  • ! path means there was an error reading or hashing the source or dest.

The --dest-after flag writes a list file using the same format flags as lsf (including customizable options for hash, modtime, etc.) Conceptually it is similar to rsync's --itemize-changes, but not identical -- it should output an accurate list of what will be on the destination after the command is finished.

When the --no-traverse flag is set, all logs involving files that exist only on the destination will be incomplete or completely missing.

Note that these logger flags have a few limitations, and certain scenarios are not currently supported:

  • --max-duration / CutoffModeHard
  • --compare-dest / --copy-dest
  • server-side moves of an entire dir at once
  • High-level retries, because there would be duplicates (use --retries 1 to disable)
  • Possibly some unusual error scenarios

Note also that each file is logged during execution, as opposed to after, so it is most useful as a predictor of what SHOULD happen to each file (which may or may not match what actually DID).

rclone moveto source:path dest:path [flags]

Options

      --absolute                Put a leading / in front of path names
      --combined string         Make a combined report of changes to this file
      --csv                     Output in CSV format
      --dest-after string       Report all files that exist on the dest post-sync
      --differ string           Report all non-matching files to this file
  -d, --dir-slash               Append a slash to directory names (default true)
      --dirs-only               Only list directories
      --error string            Report all files with errors (hashing or reading) to this file
      --files-only              Only list files (default true)
  -F, --format string           Output format - see lsf help for details (default "p")
      --hash h                  Use this hash when h is used in the format MD5|SHA-1|DropboxHash (default "md5")
  -h, --help                    help for moveto
      --match string            Report all matching files to this file
      --missing-on-dst string   Report all files missing from the destination to this file
      --missing-on-src string   Report all files missing from the source to this file
  -s, --separator string        Separator for the items in the format (default ";")
  -t, --timeformat string       Specify a custom time format - see docs for details (default: 2006-01-02 15:04:05)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Copy Options

Flags for anything which can copy a file

      --check-first                                 Do all the checks before starting transfers
  -c, --checksum                                    Check for changes with size & checksum (if available, or fallback to size only)
      --compare-dest stringArray                    Include additional server-side paths during comparison
      --copy-dest stringArray                       Implies --compare-dest but also copies files from paths into destination
      --cutoff-mode HARD|SOFT|CAUTIOUS              Mode to stop transfers when reaching the max transfer limit HARD|SOFT|CAUTIOUS (default HARD)
      --ignore-case-sync                            Ignore case when synchronizing
      --ignore-checksum                             Skip post copy check of checksums
      --ignore-existing                             Skip all files that exist on destination
      --ignore-size                                 Ignore size when skipping use modtime or checksum
  -I, --ignore-times                                Don't skip items that match size and time - transfer all unconditionally
      --immutable                                   Do not modify files, fail if existing files have been modified
      --inplace                                     Download directly to destination file instead of atomic download to temp/rename
  -l, --links                                       Translate symlinks to/from regular files with a '.rclonelink' extension
      --max-backlog int                             Maximum number of objects in sync or check backlog (default 10000)
      --max-duration Duration                       Maximum duration rclone will transfer data for (default 0s)
      --max-transfer SizeSuffix                     Maximum size of data to transfer (default off)
  -M, --metadata                                    If set, preserve metadata when copying objects
      --modify-window Duration                      Max time diff to be considered the same (default 1ns)
      --multi-thread-chunk-size SizeSuffix          Chunk size for multi-thread downloads / uploads, if not set by filesystem (default 64Mi)
      --multi-thread-cutoff SizeSuffix              Use multi-thread downloads for files above this size (default 256Mi)
      --multi-thread-streams int                    Number of streams to use for multi-thread downloads (default 4)
      --multi-thread-write-buffer-size SizeSuffix   In memory buffer size for writing when in multi-thread mode (default 128Ki)
      --name-transform stringArray                  Transform paths during the copy process
      --no-check-dest                               Don't check the destination, copy regardless
      --no-traverse                                 Don't traverse destination file system on copy
      --no-update-dir-modtime                       Don't update directory modification times
      --no-update-modtime                           Don't update destination modtime if files identical
      --order-by string                             Instructions on how to order the transfers, e.g. 'size,descending'
      --partial-suffix string                       Add partial-suffix to temporary file name when --inplace is not used (default ".partial")
      --refresh-times                               Refresh the modtime of remote files
      --server-side-across-configs                  Allow server-side operations (e.g. copy) to work across different configs
      --size-only                                   Skip based on size only, not modtime or checksum
      --streaming-upload-cutoff SizeSuffix          Cutoff for switching to chunked upload if file size is unknown, upload starts after reaching cutoff or when file ends (default 100Ki)
  -u, --update                                      Skip files that are newer on the destination

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone ncdu

Explore a remote with a text based user interface.

Synopsis

This displays a text based user interface allowing the navigation of a remote. It is most useful for answering the question - "What is using all my disk space?".

To make the user interface it first scans the entire remote given and builds an in memory representation. rclone ncdu can be used during this scanning phase and you will see it building up the directory structure as it goes along.

You can interact with the user interface using key presses, press '?' to toggle the help on and off. The supported keys are:

 ↑,↓ or k,j to Move
 →,l to enter
 ←,h to return
 g toggle graph
 c toggle counts
 a toggle average size in directory
 m toggle modified time
 u toggle human-readable format
 n,s,C,A,M sort by name,size,count,asize,mtime
 d delete file/directory
 v select file/directory
 V enter visual select mode
 D delete selected files/directories
 y copy current path to clipboard
 Y display current path
 ^L refresh screen (fix screen corruption)
 r recalculate file sizes
 ? to toggle help on and off
 ESC to close the menu box
 q/^c to quit

Listed files/directories may be prefixed by a one-character flag, some of them combined with a description in brackets at end of line. These flags have the following meaning:

e means this is an empty directory, i.e. contains no files (but
  may contain empty subdirectories)
~ means this is a directory where some of the files (possibly in
  subdirectories) have unknown size, and therefore the directory
  size may be underestimated (and average size inaccurate, as it
  is average of the files with known sizes).
. means an error occurred while reading a subdirectory, and
  therefore the directory size may be underestimated (and average
  size inaccurate)
! means an error occurred while reading this directory

This an homage to the ncdu tool but for rclone remotes. It is missing lots of features at the moment but is useful as it stands. Unlike ncdu it does not show excluded files.

Note that it might take some time to delete big files/directories. The UI won't respond in the meantime since the deletion is done synchronously.

For a non-interactive listing of the remote, see the tree command. To just get the total size of the remote you can also use the size command.

rclone ncdu remote:path [flags]

Options

  -h, --help   help for ncdu

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone nfsmount

Mount the remote as file system on a mountpoint.

Synopsis

Rclone nfsmount allows Linux, FreeBSD, macOS and Windows to mount any of Rclone's cloud storage systems as a file system with FUSE.

First set up your remote using rclone config. Check it works with rclone ls etc.

On Linux and macOS, you can run mount in either foreground or background (aka daemon) mode. Mount runs in foreground mode by default. Use the --daemon flag to force background mode. On Windows you can run mount in foreground only, the flag is ignored.

In background mode rclone acts as a generic Unix mount program: the main program starts, spawns background rclone process to setup and maintain the mount, waits until success or timeout and exits with appropriate code (killing the child process if it fails).

On Linux/macOS/FreeBSD start the mount like this, where /path/to/local/mount is an empty existing directory:

rclone nfsmount remote:path/to/files /path/to/local/mount

On Windows you can start a mount in different ways. See below for details. If foreground mount is used interactively from a console window, rclone will serve the mount and occupy the console so another window should be used to work with the mount until rclone is interrupted e.g. by pressing Ctrl-C.

The following examples will mount to an automatically assigned drive, to specific drive letter X:, to path C:\path\parent\mount (where parent directory or drive must exist, and mount must not exist, and is not supported when mounting as a network drive), and the last example will mount as network share \\cloud\remote and map it to an automatically assigned drive:

rclone nfsmount remote:path/to/files *
rclone nfsmount remote:path/to/files X:
rclone nfsmount remote:path/to/files C:\path\parent\mount
rclone nfsmount remote:path/to/files \\cloud\remote

When the program ends while in foreground mode, either via Ctrl+C or receiving a SIGINT or SIGTERM signal, the mount should be automatically stopped.

When running in background mode the user will have to stop the mount manually:

# Linux
fusermount -u /path/to/local/mount
#... or on some systems
fusermount3 -u /path/to/local/mount
# OS X or Linux when using nfsmount
umount /path/to/local/mount

The umount operation can fail, for example when the mountpoint is busy. When that happens, it is the user's responsibility to stop the mount manually.

The size of the mounted file system will be set according to information retrieved from the remote, the same as returned by the rclone about command. Remotes with unlimited storage may report the used size only, then an additional 1 PiB of free space is assumed. If the remote does not support the about feature at all, then 1 PiB is set as both the total and the free size.

Installing on Windows

To run rclone nfsmount on Windows, you will need to download and install WinFsp.

WinFsp is an open-source Windows File System Proxy which makes it easy to write user space file systems for Windows. It provides a FUSE emulation layer which rclone uses combination with cgofuse. Both of these packages are by Bill Zissimopoulos who was very helpful during the implementation of rclone nfsmount for Windows.

Mounting modes on windows

Unlike other operating systems, Microsoft Windows provides a different filesystem type for network and fixed drives. It optimises access on the assumption fixed disk drives are fast and reliable, while network drives have relatively high latency and less reliability. Some settings can also be differentiated between the two types, for example that Windows Explorer should just display icons and not create preview thumbnails for image and video files on network drives.

In most cases, rclone will mount the remote as a normal, fixed disk drive by default. However, you can also choose to mount it as a remote network drive, often described as a network share. If you mount an rclone remote using the default, fixed drive mode and experience unexpected program errors, freezes or other issues, consider mounting as a network drive instead.

When mounting as a fixed disk drive you can either mount to an unused drive letter, or to a path representing a nonexistent subdirectory of an existing parent directory or drive. Using the special value * will tell rclone to automatically assign the next available drive letter, starting with Z: and moving backward. Examples:

rclone nfsmount remote:path/to/files *
rclone nfsmount remote:path/to/files X:
rclone nfsmount remote:path/to/files C:\path\parent\mount
rclone nfsmount remote:path/to/files X:

Option --volname can be used to set a custom volume name for the mounted file system. The default is to use the remote name and path.

To mount as network drive, you can add option --network-mode to your nfsmount command. Mounting to a directory path is not supported in this mode, it is a limitation Windows imposes on junctions, so the remote must always be mounted to a drive letter.

rclone nfsmount remote:path/to/files X: --network-mode

A volume name specified with --volname will be used to create the network share path. A complete UNC path, such as \\cloud\remote, optionally with path \\cloud\remote\madeup\path, will be used as is. Any other string will be used as the share part, after a default prefix \\server\. If no volume name is specified then \\server\share will be used. You must make sure the volume name is unique when you are mounting more than one drive, or else the mount command will fail. The share name will treated as the volume label for the mapped drive, shown in Windows Explorer etc, while the complete \\server\share will be reported as the remote UNC path by net use etc, just like a normal network drive mapping.

If you specify a full network share UNC path with --volname, this will implicitly set the --network-mode option, so the following two examples have same result:

rclone nfsmount remote:path/to/files X: --network-mode
rclone nfsmount remote:path/to/files X: --volname \\server\share

You may also specify the network share UNC path as the mountpoint itself. Then rclone will automatically assign a drive letter, same as with * and use that as mountpoint, and instead use the UNC path specified as the volume name, as if it were specified with the --volname option. This will also implicitly set the --network-mode option. This means the following two examples have same result:

rclone nfsmount remote:path/to/files \\cloud\remote
rclone nfsmount remote:path/to/files * --volname \\cloud\remote

There is yet another way to enable network mode, and to set the share path, and that is to pass the "native" libfuse/WinFsp option directly: --fuse-flag --VolumePrefix=\server\share. Note that the path must be with just a single backslash prefix in this case.

Note: In previous versions of rclone this was the only supported method.

Read more about drive mapping

See also Limitations section below.

Windows filesystem permissions

The FUSE emulation layer on Windows must convert between the POSIX-based permission model used in FUSE, and the permission model used in Windows, based on access-control lists (ACL).

The mounted filesystem will normally get three entries in its access-control list (ACL), representing permissions for the POSIX permission scopes: Owner, group and others. By default, the owner and group will be taken from the current user, and the built-in group "Everyone" will be used to represent others. The user/group can be customized with FUSE options "UserName" and "GroupName", e.g. -o UserName=user123 -o GroupName="Authenticated Users". The permissions on each entry will be set according to options --dir-perms and --file-perms, which takes a value in traditional Unix numeric notation.

The default permissions corresponds to --file-perms 0666 --dir-perms 0777, i.e. read and write permissions to everyone. This means you will not be able to start any programs from the mount. To be able to do that you must add execute permissions, e.g. --file-perms 0777 --dir-perms 0777 to add it to everyone. If the program needs to write files, chances are you will have to enable VFS File Caching as well (see also limitations). Note that the default write permission have some restrictions for accounts other than the owner, specifically it lacks the "write extended attributes", as explained next.

The mapping of permissions is not always trivial, and the result you see in Windows Explorer may not be exactly like you expected. For example, when setting a value that includes write access for the group or others scope, this will be mapped to individual permissions "write attributes", "write data" and "append data", but not "write extended attributes". Windows will then show this as basic permission "Special" instead of "Write", because "Write" also covers the "write extended attributes" permission. When setting digit 0 for group or others, to indicate no permissions, they will still get individual permissions "read attributes", "read extended attributes" and "read permissions". This is done for compatibility reasons, e.g. to allow users without additional permissions to be able to read basic metadata about files like in Unix.

WinFsp 2021 (version 1.9) introduced a new FUSE option "FileSecurity", that allows the complete specification of file security descriptors using SDDL. With this you get detailed control of the resulting permissions, compared to use of the POSIX permissions described above, and no additional permissions will be added automatically for compatibility with Unix. Some example use cases will following.

If you set POSIX permissions for only allowing access to the owner, using --file-perms 0600 --dir-perms 0700, the user group and the built-in "Everyone" group will still be given some special permissions, as described above. Some programs may then (incorrectly) interpret this as the file being accessible by everyone, for example an SSH client may warn about "unprotected private key file". You can work around this by specifying -o FileSecurity="D:P(A;;FA;;;OW)", which sets file all access (FA) to the owner (OW), and nothing else.

When setting write permissions then, except for the owner, this does not include the "write extended attributes" permission, as mentioned above. This may prevent applications from writing to files, giving permission denied error instead. To set working write permissions for the built-in "Everyone" group, similar to what it gets by default but with the addition of the "write extended attributes", you can specify -o FileSecurity="D:P(A;;FRFW;;;WD)", which sets file read (FR) and file write (FW) to everyone (WD). If file execute (FX) is also needed, then change to -o FileSecurity="D:P(A;;FRFWFX;;;WD)", or set file all access (FA) to get full access permissions, including delete, with -o FileSecurity="D:P(A;;FA;;;WD)".

Windows caveats

Drives created as Administrator are not visible to other accounts, not even an account that was elevated to Administrator with the User Account Control (UAC) feature. A result of this is that if you mount to a drive letter from a Command Prompt run as Administrator, and then try to access the same drive from Windows Explorer (which does not run as Administrator), you will not be able to see the mounted drive.

If you don't need to access the drive from applications running with administrative privileges, the easiest way around this is to always create the mount from a non-elevated command prompt.

To make mapped drives available to the user account that created them regardless if elevated or not, there is a special Windows setting called linked connections that can be enabled.

It is also possible to make a drive mount available to everyone on the system, by running the process creating it as the built-in SYSTEM account. There are several ways to do this: One is to use the command-line utility PsExec, from Microsoft's Sysinternals suite, which has option -s to start processes as the SYSTEM account. Another alternative is to run the mount command from a Windows Scheduled Task, or a Windows Service, configured to run as the SYSTEM account. A third alternative is to use the WinFsp.Launcher infrastructure). Read more in the install documentation. Note that when running rclone as another user, it will not use the configuration file from your profile unless you tell it to with the --config option. Note also that it is now the SYSTEM account that will have the owner permissions, and other accounts will have permissions according to the group or others scopes. As mentioned above, these will then not get the "write extended attributes" permission, and this may prevent writing to files. You can work around this with the FileSecurity option, see example above.

Note that mapping to a directory path, instead of a drive letter, does not suffer from the same limitations.

Mounting on macOS

Mounting on macOS can be done either via built-in NFS server, macFUSE (also known as osxfuse) or FUSE-T. macFUSE is a traditional FUSE driver utilizing a macOS kernel extension (kext). FUSE-T is an alternative FUSE system which "mounts" via an NFSv4 local server.

Unicode Normalization

It is highly recommended to keep the default of --no-unicode-normalization=false for all mount and serve commands on macOS. For details, see vfs-case-sensitivity.

NFS mount

For macOS (and other platforms where this path is supported), prefer the dedicated rclone nfsmount command. It starts the NFS server and performs the mount for you. rclone mount itself still uses FUSE (macFUSE/FUSE-T) and does not switch to NFS via a flag.

This method spins up an NFS server using serve nfs command and mounts it to the specified mountpoint. If you run this in background mode using |--daemon|, you will need to send SIGTERM signal to the rclone process using |kill| command to stop the mount.

Note that --nfs-cache-handle-limit controls the maximum number of cached file handles stored by the nfsmount caching handler. This should not be set too low or you may experience errors when trying to access files. The default is 1000000, but consider lowering this limit if the server's system resource usage causes problems.

macFUSE Notes

If installing macFUSE using dmg packages from the website, rclone will locate the macFUSE libraries without any further intervention. If however, macFUSE is installed using the macports package manager, the following addition steps are required.

sudo mkdir /usr/local/lib
cd /usr/local/lib
sudo ln -s /opt/local/lib/libfuse.2.dylib

FUSE-T Limitations, Caveats, and Notes

There are some limitations, caveats, and notes about how it works. These are current as of FUSE-T version 1.0.14.

ModTime update on read

As per the FUSE-T wiki:

File access and modification times cannot be set separately as it seems to be an issue with the NFS client which always modifies both. Can be reproduced with touch -m and touch -a commands

This means that viewing files with various tools, notably macOS Finder, will cause rlcone to update the modification time of the file. This may make rclone upload a full new copy of the file.

Read Only mounts

When mounting with --read-only, attempts to write to files will fail silently as opposed to with a clear warning as in macFUSE.

Mounting on Linux

On newer versions of Ubuntu, you may encounter the following error when running rclone mount:

NOTICE: mount helper error: fusermount3: mount failed: Permission denied CRITICAL: Fatal error: failed to mount FUSE fs: fusermount: exit status 1

This may be due to newer Apparmor restrictions, which can be disabled with sudo aa-disable /usr/bin/fusermount3 (you may need to sudo apt install apparmor-utils beforehand).

Mounting on OpenBSD

rclone nfsmount works on OpenBSD by spinning up the serve nfs server and mounting it with the system mount_nfs(8), the same approach used for the NFS mount method on macOS. There is no FUSE-based mount option on OpenBSD, so nfsmount is the only way to mount rclone remotes as a local file system there.

portmap registration

OpenBSD's kernel NFS client locates mountd through the system portmapper rather than connecting to a fixed port, so rclone's in-process NFS server needs to be registered with portmap before it can be mounted. Start the serve nfs server on a fixed port, stop the system NFS services so they don't conflict with it, then register that port with portmap for both the nfs and mountd RPC programs (this is the recipe from hjicks in #8578):

rclone serve nfs --addr localhost:<PORT> remote:path

doas rcctl stop nfsd portmap
doas pkill -9 mountd   # mountd can't be stopped via rcctl(8)
doas rcctl -f start portmap
doas rpcinfo -s nfs 3 <PORT>
doas rpcinfo -s mountd 3 <PORT>

Replace <PORT> with the port passed to --addr. Port 2049 works too, but needs either running rclone as root or adjusting sysctl(8) at securelevel 0, so a non-privileged port is simpler for testing.

Use mount_nfs -T, not mount

Mount the exported NFS share directly with mount_nfs(8) using the -T flag to force TCP, rather than the generic mount(8) wrapper - OpenBSD's mount(8) does not understand -T and rejects it with unknown option -- T:

mount_nfs -T localhost:/ /path/to/local/mount

rclone nfsmount does this for you automatically on OpenBSD (it calls mount_nfs directly with -T instead of going through mount), so this is only needed if you are mounting the serve nfs export by hand.

AUTH_UNIX and existing limitations

OpenBSD's (and the other BSDs') kernel NFS client requires the server to offer the AUTH_UNIX authentication flavor during the MOUNT RPC handshake; a server that only offers AUTH_NULL is rejected with mount_nfs: can't access /: Authentication error. rclone's NFS server advertises AUTH_UNIX for this reason, though it does not implement any per-user access control on top of it - see issue #8578 for the background on this and for the current state of OpenBSD support.

Limitations

Without the use of --vfs-cache-mode this can only write files sequentially, it can only seek when reading. This means that many applications won't work with their files on an rclone mount without --vfs-cache-mode writes or --vfs-cache-mode full. See the VFS File Caching section for more info. When using NFS mount on macOS, if you don't specify |--vfs-cache-mode| the mount point will be read-only.

Bucket-based remotes - Azure Blob, Swift, S3, Google Cloud Storage and B2 - can't store empty directories. Of these, only Azure Blob, Google Cloud Storage and S3 can preserve them when you add --xxx-directory_markers; otherwise, empty directories will vanish once they drop out of the directory cache.

When rclone mount is invoked on Unix with --daemon flag, the main rclone program will wait for the background mount to become ready or until the timeout specified by the --daemon-wait flag. On Linux it can check mount status using ProcFS so the flag in fact sets maximum time to wait, while the real wait can be less. On macOS / BSD the time to wait is constant and the check is performed only at the end. We advise you to set wait time on macOS reasonably.

Only supported on Linux, FreeBSD, OS X and Windows at the moment.

rclone nfsmount vs rclone sync/copy

File systems expect things to be 100% reliable, whereas cloud storage systems are a long way from 100% reliable. The rclone sync/copy commands cope with this with lots of retries. However rclone nfsmount can't use retries in the same way without making local copies of the uploads. Look at the VFS File Caching for solutions to make nfsmount more reliable.

Attribute caching

You can use the flag --attr-timeout to set the time the kernel caches the attributes (size, modification time, etc.) for directory entries.

The default is 1s which caches files just long enough to avoid too many callbacks to rclone from the kernel.

In theory 0s should be the correct value for filesystems which can change outside the control of the kernel. However this causes quite a few problems such as rclone using too much memory, rclone not serving files to samba and excessive time listing directories.

The kernel can cache the info about a file for the time given by --attr-timeout. You may see corruption if the remote file changes length during this window. It will show up as either a truncated file or a file with garbage on the end. With --attr-timeout 1s this is very unlikely but not impossible. The higher you set --attr-timeout the more likely it is. The default setting of "1s" is the lowest setting which mitigates the problems above.

If you set it higher (10s or 1m say) then the kernel will call back to rclone less often making it more efficient, however there is more chance of the corruption issue above.

If files don't change on the remote outside of the control of rclone then there is no chance of corruption.

This is the same as setting the attr_timeout option in mount.fuse.

Filters

Note that all the rclone filters can be used to select a subset of the files to be visible in the mount.

systemd

When running rclone nfsmount as a systemd service, it is possible to use Type=notify. In this case the service will enter the started state after the mountpoint has been successfully set up. Units having the rclone nfsmount service specified as a requirement will see all files and folders immediately in this mode.

Note that systemd runs mount units without any environment variables including PATH or HOME. This means that tilde (~) expansion will not work and you should provide --config and --cache-dir explicitly as absolute paths via rclone arguments. Since mounting requires the fusermount or fusermount3 program, rclone will use the fallback PATH of /bin:/usr/bin in this scenario. Please ensure that fusermount/fusermount3 is present on this PATH.

Rclone as Unix mount helper

The core Unix program /bin/mount normally takes the -t FSTYPE argument then runs the /sbin/mount.FSTYPE helper program passing it mount options as -o key=val,... or --opt=.... Automount (classic or systemd) behaves in a similar way.

rclone by default expects GNU-style flags --key val. To run it as a mount helper you should symlink rclone binary to /sbin/mount.rclone and optionally /usr/bin/rclonefs, e.g. ln -s /usr/bin/rclone /sbin/mount.rclone. rclone will detect it and translate command-line arguments appropriately.

Now you can run classic mounts like this:

mount sftp1:subdir /mnt/data -t rclone -o vfs_cache_mode=writes,sftp_key_file=/path/to/pem

or create systemd mount units:

# /etc/systemd/system/mnt-data.mount
[Unit]
Description=Mount for /mnt/data
[Mount]
Type=rclone
What=sftp1:subdir
Where=/mnt/data
Options=rw,_netdev,allow_other,args2env,vfs-cache-mode=writes,config=/etc/rclone.conf,cache-dir=/var/rclone

optionally accompanied by systemd automount unit

# /etc/systemd/system/mnt-data.automount
[Unit]
Description=AutoMount for /mnt/data
[Automount]
Where=/mnt/data
TimeoutIdleSec=600
[Install]
WantedBy=multi-user.target

or add in /etc/fstab a line like

sftp1:subdir /mnt/data rclone rw,noauto,nofail,_netdev,x-systemd.automount,args2env,vfs_cache_mode=writes,config=/etc/rclone.conf,cache_dir=/var/cache/rclone 0 0

or use classic Automountd. Remember to provide explicit config=...,cache-dir=... as a workaround for mount units being run without HOME.

Rclone in the mount helper mode will split -o argument(s) by comma, replace _ by - and prepend -- to get the command-line flags. Options containing commas or spaces can be wrapped in single or double quotes. Any inner quotes inside outer quotes of the same type should be doubled.

Mount option syntax includes a few extra options treated specially:

  • env.NAME=VALUE will set an environment variable for the mount process. This helps with Automountd and Systemd.mount which don't allow setting custom environment for mount helpers. Typically you will use env.HTTPS_PROXY=proxy.host:3128 or env.HOME=/root
  • command=cmount can be used to run cmount or any other rclone command rather than the default mount.
  • args2env will pass mount options to the mount helper running in background via environment variables instead of command line arguments. This allows to hide secrets from such commands as ps or pgrep.
  • vv... will be transformed into appropriate --verbose=N
  • standard mount options like x-systemd.automount, _netdev, nosuid and alike are intended only for Automountd and ignored by rclone.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

rclone nfsmount remote:path /path/to/mountpoint [flags]

Options

      --addr string                            IPaddress:Port or :Port to bind server to
      --allow-idmap                            Allow id-mapped mounts (Linux 6.12+, mount2 only)
      --allow-non-empty                        Allow mounting over a non-empty directory (not supported on Windows)
      --allow-other                            Allow access to other users (not supported on Windows)
      --allow-root                             Allow access to root user (not supported on Windows)
      --async-read                             Use asynchronous reads (not supported on Windows) (default true)
      --attr-timeout Duration                  Time for which file/directory attributes are cached (default 1s)
      --daemon                                 Run mount in background and exit parent process (as background output is suppressed, use --log-file with --log-format=pid,... to monitor) (not supported on Windows)
      --daemon-timeout Duration                Time limit for rclone to respond to kernel (not supported on Windows) (default 0s)
      --daemon-wait Duration                   Time to wait for ready mount from daemon (maximum time on Linux, constant sleep time on OSX/BSD) (not supported on Windows) (default 1m0s)
      --debug-fuse                             Debug the FUSE internals - needs -v
      --default-permissions                    Makes kernel enforce access control based on the file mode (not supported on Windows)
      --devname string                         Set the device name - default is remote:path
      --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                     Directory permissions (default 777)
      --direct-io                              Use Direct IO, disables caching of data
      --file-perms FileMode                    File permissions (default 666)
      --fuse-flag stringArray                  Flags or arguments to be passed direct to libfuse/WinFsp (repeat if required)
      --gid uint32                             Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                   help for nfsmount
      --link-perms FileMode                    Link permissions (default 666)
      --max-read-ahead SizeSuffix              The number of bytes that can be prefetched for sequential reads (not supported on Windows) (default 128Ki)
      --mount-case-insensitive Tristate        Tell the OS the mount is case insensitive (true) or sensitive (false) regardless of the backend (auto) (default unset)
      --network-mode                           Mount as remote network drive, instead of fixed disk drive (supported on Windows only)
      --nfs-cache-dir string                   The directory the NFS handle cache will use if set
      --nfs-cache-handle-limit int             max file handles cached simultaneously (min 5) (default 1000000)
      --nfs-cache-type memory|disk|symlink     Type of NFS handle cache to use (default memory)
      --nfs-mount-path string                  Subpath of the remote to mount via NFS (must be an existing directory). (default "/")
      --no-checksum                            Don't compare checksums on up/download
      --no-modtime                             Don't read/write the modification time (can speed things up)
      --no-seek                                Don't allow seeking in files
      --noappledouble                          Ignore Apple Double (._) and .DS_Store files (supported on OSX only) (default true)
      --noapplexattr                           Ignore all "com.apple.*" extended attributes (supported on OSX only)
  -o, --option stringArray                     Option for libfuse/WinFsp (repeat if required)
      --poll-interval Duration                 Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --read-only                              Only allow read-only access
      --sudo                                   Use sudo to run the mount/umount commands as root.
      --uid uint32                             Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                         Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --vfs-block-norm-dupes                   If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration             Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix          Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix    Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode               Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration       Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                   If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix   Specify the total space of disk (default off)
      --vfs-fast-fingerprint                   Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration            Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                              Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string          Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix              Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix         Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix   If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int             The number of parallel streams to read at once
      --vfs-read-wait Duration                 Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                            Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size           Use the rclone size algorithm for Used size
      --vfs-write-back Duration                Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                Time to wait for in-sequence write before giving error (default 1s)
      --volname string                         Set the volume name (supported on Windows and OSX only)
      --write-back-cache                       Makes kernel buffer writes before sending them to rclone (without this, writethrough caching is used) (not supported on Windows)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone obscure

Obscure password for use in the rclone config file.

Synopsis

In the rclone config file, human-readable passwords are obscured. Obscuring them is done by encrypting them and writing them out in base64. This is not a secure way of encrypting these passwords as rclone can decrypt them - it is to prevent "eyedropping" - namely someone seeing a password in the rclone config file by accident.

Many equally important things (like access tokens) are not obscured in the config file. However it is very hard to shoulder surf a 64 character hex token.

This command can also accept a password through STDIN instead of an argument by passing a hyphen as an argument. This will use the first line of STDIN as the password not including the trailing newline.

echo 'secretpassword' | rclone obscure -

If there is no data on STDIN to read, rclone obscure will default to obfuscating the hyphen itself.

If you want to encrypt the config file then please use config file encryption - see rclone config for more info.

rclone obscure password [flags]

Options

  -h, --help   help for obscure

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone rc

Run a command against a running rclone.

Synopsis

This runs a command against a running rclone. Use the --url flag to specify an non default URL to connect on. This can be either a ":port" which is taken to mean http://localhost:port or a "host:port" which is taken to mean http://host:port.

A username and password can be passed in with --user and --pass.

Note that --rc-addr, --rc-user, --rc-pass will be read also for --url, --user, --pass.

The --unix-socket flag can be used to connect over a unix socket like this

# start server on /tmp/my.socket
rclone rcd --rc-addr unix:///tmp/my.socket
# Connect to it
rclone rc --unix-socket /tmp/my.socket core/stats

Arguments should be passed in as parameter=value.

The result will be returned as a JSON object by default.

The --json parameter can be used to pass in a JSON blob as an input instead of key=value arguments. This is the only way of passing in more complicated values.

The -o/--opt option can be used to set a key "opt" with key, value options in the form -o key=value or -o key. It can be repeated as many times as required. This is useful for rc commands which take the "opt" parameter which by convention is a dictionary of strings.

-o key=value -o key2

Will place this in the "opt" value

{"key":"value", "key2","")

The -a/--arg option can be used to set strings in the "arg" value. It can be repeated as many times as required. This is useful for rc commands which take the "arg" parameter which by convention is a list of strings.

-a value -a value2

Will place this in the "arg" value

["value", "value2"]

Use --loopback to connect to the rclone instance running rclone rc. This is very useful for testing commands without having to run an rclone rc server, e.g.:

rclone rc --loopback operations/about fs=/

Use rclone rc to see a list of all possible commands.

rclone rc commands parameter [flags]

Options

  -a, --arg stringArray      Argument placed in the "arg" array
  -h, --help                 help for rc
      --json string          Input JSON - use instead of key=value args
      --loopback             If set connect to this rclone instance not via HTTP
      --no-output            If set, don't output the JSON result
  -o, --opt stringArray      Option in the form name=value or name placed in the "opt" array
      --pass string          Password to use to connect to rclone remote control
      --unix-socket string   Path to a unix domain socket to dial to, instead of opening a TCP connection directly
      --url string           URL to connect to rclone remote control (default "http://localhost:5572/")
      --user string          Username to use to rclone remote control

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone rcat

Copies standard input to file on remote.

Synopsis

Reads from standard input (stdin) and copies it to a single remote file.

echo "hello world" | rclone rcat remote:path/to/file
ffmpeg - | rclone rcat remote:path/to/file

If the remote file already exists, it will be overwritten.

rcat will try to upload small files in a single request, which is usually more efficient than the streaming/chunked upload endpoints, which use multiple requests. Exact behaviour depends on the remote. What is considered a small file may be set through --streaming-upload-cutoff. Uploading only starts after the cutoff is reached or if the file ends before that. The data must fit into RAM. The cutoff needs to be small enough to adhere the limits of your remote, please see there. Generally speaking, setting this cutoff too high will decrease your performance.

Use the --size flag to preallocate the file in advance at the remote end and actually stream it, even if remote backend doesn't support streaming.

--size should be the exact size of the input stream in bytes. If the size of the stream is different in length to the --size passed in then the transfer will likely fail.

Note that the upload cannot be retried because the data is not stored. If the backend supports multipart uploading then individual chunks can be retried. If you need to transfer a lot of data, you may be better off caching it locally and then rclone move it to the destination which can use retries.

rclone rcat remote:path [flags]

Options

  -h, --help       help for rcat
      --size int   File size hint to preallocate (default -1)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone rcd

Run rclone listening to remote control commands only.

Synopsis

This runs rclone so that it only listens to remote control commands.

This is useful if you are controlling rclone via the rc API.

If you pass in a path to a directory, rclone will serve that directory for GET requests on the URL passed in. It will also open the URL in the browser when rclone is run.

See the rc documentation for more info on the rc flags.

Server options

Use --rc-addr to specify which IP address and port the server should listen on, eg --rc-addr 1.2.3.4:8000 or --rc-addr :8080 to listen to all IPs. By default it only listens on localhost. You can use port :0 to let the OS choose an available port.

If you set --rc-addr to listen on a public or LAN accessible IP address then using Authentication is advised - see the next section for info.

You can use a unix socket by setting the url to unix:///path/to/socket or just by using an absolute path name.

--rc-addr may be repeated to listen on multiple IPs/ports/sockets. Socket activation, described further below, can also be used to accomplish the same.

--rc-server-read-timeout and --rc-server-write-timeout can be used to control the timeouts on the server. Note that this is the total time for a transfer.

--rc-max-header-bytes controls the maximum number of bytes the server will accept in the HTTP header.

--rc-response-header can be used to set an HTTP header for all responses, will overriding existing values. The flag may be repeated to add multiple headers. Use the format Header-Name: value.

--rc-allow-origin enables cross-origin (CORS) requests from browsers by setting the Access-Control-Allow-Origin header on every response to the origin given, e.g. --rc-allow-origin https://app.example.com. Browser CORS preflight OPTIONS requests are answered by rclone without needing authentication as browsers never send credentials with them. Setting this to * allows any origin which is a security risk on an authenticated server.

--rc-baseurl controls the URL prefix that rclone serves from. By default rclone will serve from the root. If you used --rc-baseurl "/rclone" then rclone would serve from a URL starting with "/rclone/". This is useful if you wish to proxy rclone serve. Rclone automatically inserts leading and trailing "/" on --rc-baseurl, so --rc-baseurl "rclone", --rc-baseurl "/rclone" and --rc-baseurl "/rclone/" are all treated identically.

--rc-disable-zip may be set to disable the zipping download option.

Protocol

The server supports HTTP/1.1 and HTTP/2. HTTP/2 is used automatically for TLS connections. For non-TLS connections, HTTP/2 cleartext (h2c) is supported, allowing HTTP/2 without encryption.

TLS (SSL)

By default this will serve over http. If you want you can serve over https. You will need to supply the --rc-cert and --rc-key flags. If you wish to do client side certificate validation then you will need to supply --rc-client-ca also.

When TLS is configured every listener given with --rc-addr serves TLS. An individual listener can be prefixed with http:// to serve unencrypted HTTP on that address, or with tls:// to state explicitly that it must serve TLS. Using a tls:// address without --rc-cert and --rc-key is an error.

--rc-cert must be set to the path of a file containing either a PEM encoded certificate, or a concatenation of that with the CA certificate. --rc-key must be set to the path of a file with the PEM encoded private key. If setting --rc-client-ca, it should be set to the path of a file with PEM encoded client certificate authority certificates.

--rc-min-tls-version is minimum TLS version that is acceptable. Valid values are "tls1.0", "tls1.1", "tls1.2" and "tls1.3" (default "tls1.0").

Socket activation

Instead of the listening addresses specified above, rclone will listen to all FDs passed by the service manager, if any (and ignore any arguments passed by --rc-addr).

This allows rclone to be a socket-activated service. It can be configured with .socket and .service unit files as described in https://www.freedesktop.org/software/systemd/man/latest/systemd.socket.html.

Socket activation can be tested ad-hoc with the systemd-socket-activatecommand

systemd-socket-activate -l 8000 -- rclone serve

This will socket-activate rclone on the first connection to port 8000 over TCP.

Template

--rc-template allows a user to specify a custom markup template for HTTP and WebDAV serve functions. The server exports the following markup to be used within the template to server pages:

Parameter Subparameter Description
.Name The full path of a file/directory.
.Title Directory listing of '.Name'.
.Sort The current sort used. This is changeable via '?sort=' parameter. Possible values: namedirfirst, name, size, time (default namedirfirst).
.Order The current ordering used. This is changeable via '?order=' parameter. Possible values: asc, desc (default asc).
.Query Currently unused.
.Path The path of the directory from the root with leading and trailing slashes, e.g. '/dir/'.
.IsRoot Boolean for if this is the root directory.
.UpLink The link to the parent directory.
.Static Boolean for if the listing is being written as a static page by rclone index.
.NumDirs The number of directories in the listing.
.NumFiles The number of files in the listing.
.TotalSize The total size in bytes of the files in the listing.
.Breadcrumb Allows for creating a relative navigation.
.Link The link of the Text relative to the root.
.Text The Name of the directory.
.Entries Information about a specific file/directory.
.URL The url of an entry.
.Leaf Currently same as '.URL' but intended to be just the name.
.IsDir Boolean for if an entry is a directory or not.
.Size Size in bytes of the entry.
.ModTime The UTC timestamp of an entry.
.MimeType The MIME type of an entry guessed from its name, or 'inode/directory'.

The server also makes the following functions available so that they can be used within the template. These functions help extend the options for dynamic rendering of HTML. They can be used to render HTML based on specific conditions.

Function Description
afterEpoch Returns the time since the epoch for the given time.
contains Checks whether a given substring is present or not in a given string.
hasPrefix Checks whether the given string begins with the specified prefix.
hasSuffix Checks whether the given string end with the specified suffix.

Authentication

By default this will serve files without needing a login.

You can either use an htpasswd file which can take lots of users, or set a single username and password with the --rc-user and --rc-pass flags.

Alternatively, you can have the reverse proxy manage authentication and use the username provided in the configured header with --user-from-header (e.g., --rc-user-from-header=x-remote-user). Ensure the proxy is trusted and headers cannot be spoofed, as misconfiguration may lead to unauthorized access.

If either of the above authentication methods is not configured and client certificates are required by the --client-ca flag passed to the server, the client certificate common name will be considered as the username.

Use --rc-htpasswd /path/to/htpasswd to provide an htpasswd file. This is in standard apache format and supports MD5, SHA1 and BCrypt for basic authentication. Bcrypt is recommended.

To create an htpasswd file:

touch htpasswd
htpasswd -B htpasswd user
htpasswd -B htpasswd anotherUser

Use the same hash format and cost for every entry in the file. If the entries differ, the time taken to reject a wrong password can reveal which usernames exist.

The password file can be updated while rclone is running.

When authentication is enabled every request must be authenticated, including HTTP OPTIONS requests, as the answer to an OPTIONS request can reveal whether a path exists. Health checks or monitoring which use unauthenticated OPTIONS requests will need to send credentials. The exception is browser CORS preflight requests when --rc-allow-origin is set, which are answered without authentication as browsers never send credentials with them.

Use --rc-realm to set the authentication realm.

Use --rc-salt to change the password hashing salt from the default.

rclone rcd <path to files to serve>* [flags]

Options

  -h, --help   help for rcd

Options shared with other commands are described next. See the global flags page for global options not listed here.

RC Options

Flags to control the Remote Control API

      --rc                                 Enable the remote control server
      --rc-addr stringArray                IPaddress:Port or :Port to bind server to (default localhost:5572)
      --rc-allow-origin string             Origin which cross-domain request (CORS) can be executed from
      --rc-baseurl string                  Prefix for URLs - leave blank for root
      --rc-cert string                     TLS PEM key (concatenation of certificate and CA certificate)
      --rc-client-ca string                Client certificate authority to verify clients with
      --rc-enable-metrics                  Enable the Prometheus metrics path at the remote control server
      --rc-files string                    Path to local files to serve on the HTTP server
      --rc-htpasswd string                 A htpasswd file - if not provided no authentication is done
      --rc-job-expire-duration Duration    Expire finished async jobs older than this value (default 1m0s)
      --rc-job-expire-interval Duration    Interval to check for expired async jobs (default 10s)
      --rc-key string                      TLS PEM Private key
      --rc-max-header-bytes int            Maximum size of request header (default 4096)
      --rc-min-tls-version string          Minimum TLS version that is acceptable (default "tls1.0")
      --rc-no-auth                         Don't require auth for certain methods
      --rc-pass string                     Password for authentication
      --rc-realm string                    Realm for authentication
      --rc-response-header stringArray     Set HTTP header for all responses, overriding existing values
      --rc-salt string                     Password hashing salt (default "dlPL2MqE")
      --rc-serve                           Enable the serving of remote objects
      --rc-serve-no-modtime                Don't read the modification time (can speed things up)
      --rc-server-read-timeout Duration    Timeout for server reading data (default 1h0m0s)
      --rc-server-write-timeout Duration   Timeout for server writing data (default 1h0m0s)
      --rc-template string                 User-specified template
      --rc-user string                     User name for authentication
      --rc-user-from-header string         User name from a defined HTTP header

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone rmdirs

Remove empty directories under the path.

Synopsis

This recursively removes any empty directories (including directories that only contain empty directories), that it finds under the path. The root path itself will also be removed if it is empty, unless you supply the --leave-root flag.

Use command rmdir to delete just the empty directory given by path, not recurse.

This is useful for tidying up remotes that rclone has left a lot of empty directories in. For example the delete command will delete files but leave the directory structure (unless used with option --rmdirs).

This will delete --checkers directories concurrently so if you have thousands of empty directories consider increasing this number.

To delete a path and any objects in it, use the purge command.

rclone rmdirs remote:path [flags]

Options

  -h, --help         help for rmdirs
      --leave-root   Do not remove root directory if empty

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone selfupdate

Update the rclone binary.

Synopsis

This command downloads the latest release of rclone and replaces the currently running binary. The download is verified with a hashsum and cryptographically signed signature; see the release signing docs for details.

If used without flags (or with implied --stable flag), this command will install the latest stable release. However, some issues may be fixed (or features added) only in the latest beta release. In such cases you should run the command with the --beta flag, i.e. rclone selfupdate --beta. You can check in advance what version would be installed by adding the --check flag, then repeat the command without it when you are satisfied.

Sometimes the rclone team may recommend you a concrete beta or stable rclone release to troubleshoot your issue or add a bleeding edge feature. The --version VER flag, if given, will update to the concrete version instead of the latest one. If you omit micro version from VER (for example 1.53), the latest matching micro version will be used.

Upon successful update rclone will print a message that contains a previous version number. You will need it if you later decide to revert your update for some reason. Then you'll have to note the previous version and run the following command: rclone selfupdate [--beta] OLDVER. If the old version contains only dots and digits (for example v1.54.0) then it's a stable release so you won't need the --beta flag. Beta releases have an additional information similar to v1.54.0-beta.5111.06f1c0c61. (if you are a developer and use a locally built rclone, the version number will end with -DEV, you will have to rebuild it as it obviously can't be distributed).

If you previously installed rclone via a package manager, the package may include local documentation or configure services. You may wish to update with the flag --package deb or --package rpm (whichever is correct for your OS) to update these too. This command with the default --package zip will update only the rclone executable so the local manual may become inaccurate after it.

The rclone mount command may or may not support extended FUSE options depending on the build and OS. selfupdate will refuse to update if the capability would be discarded.

Note: Windows forbids deletion of a currently running executable so this command will rename the old executable to 'rclone.old.exe' upon success.

Please note that this command was not available before rclone version 1.55. If it fails for you with the message unknown command "selfupdate" then you will need to update manually following the install documentation.

rclone selfupdate [flags]

Options

      --beta             Install beta release
      --check            Check for latest release, do not download
  -h, --help             help for selfupdate
      --output string    Save the downloaded binary at a given path (default: replace running binary)
      --package string   Package format: zip|deb|rpm (default: zip)
      --stable           Install stable release (this is the default)
      --version string   Install the given rclone version (default: latest)

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone serve

Serve a remote over a protocol.

Synopsis

Serve a remote over a given protocol. Requires the use of a subcommand to specify the protocol, e.g.

rclone serve http remote:

When the "--metadata" flag is enabled, the following metadata fields will be provided as headers:

  • "content-disposition"
  • "cache-control"
  • "content-language"
  • "content-encoding" Note: The availability of these fields depends on whether the remote supports metadata.

Each subcommand has its own options which you can see in their help.

rclone serve <protocol> [opts] <remote> [flags]

Options

  -h, --help   help for serve

See the global flags page for global options not listed here.

See Also

rclone serve dlna

Serve remote:path over DLNA

Synopsis

Run a DLNA media server for media stored in an rclone remote. Many devices, such as the Xbox and PlayStation, can automatically discover this server in the LAN and play audio/video from it. VLC is also supported. Service discovery uses UDP multicast packets (SSDP) and will thus only work on LANs.

Rclone will list all files present in the remote, without filtering based on media formats or file extensions. Additionally, there is no media transcoding support. This means that some players might show files that they are not able to play back correctly.

Rclone will add external subtitle files (.srt) to videos if they have the same filename as the video file itself (except the extension), either in the same directory as the video, or in a "Subs" subdirectory.

Server options

Use --addr to specify which IP address and port the server should listen on, e.g. --addr 1.2.3.4:8000 or --addr :8080 to listen to all IPs.

Use --name to choose the friendly server name, which is by default "rclone (hostname)".

Use --log-trace in conjunction with -vv to enable additional debug logging of all UPNP traffic.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

rclone serve dlna remote:path [flags]

Options

      --addr string                            The ip:port or :port to bind the DLNA http server to (default ":7879")
      --announce-interval Duration             The interval between SSDP announcements (default 12m0s)
      --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                     Directory permissions (default 777)
      --file-perms FileMode                    File permissions (default 666)
      --gid uint32                             Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                   help for dlna
      --interface stringArray                  The interface to use for SSDP (repeat as necessary)
      --link-perms FileMode                    Link permissions (default 666)
      --log-trace                              Enable trace logging of SOAP traffic
      --name string                            Name of DLNA server
      --no-checksum                            Don't compare checksums on up/download
      --no-modtime                             Don't read/write the modification time (can speed things up)
      --no-seek                                Don't allow seeking in files
      --poll-interval Duration                 Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --read-only                              Only allow read-only access
      --uid uint32                             Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                         Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --vfs-block-norm-dupes                   If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration             Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix          Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix    Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode               Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration       Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                   If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix   Specify the total space of disk (default off)
      --vfs-fast-fingerprint                   Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration            Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                              Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string          Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix              Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix         Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix   If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int             The number of parallel streams to read at once
      --vfs-read-wait Duration                 Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                            Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size           Use the rclone size algorithm for Used size
      --vfs-write-back Duration                Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                Time to wait for in-sequence write before giving error (default 1s)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

rclone serve docker

Serve any remote on docker's volume plugin API.

Synopsis

This command implements the Docker volume plugin API allowing docker to use rclone as a data storage mechanism for various cloud providers. rclone provides docker volume plugin based on it.

To create a docker plugin, one must create a Unix or TCP socket that Docker will look for when you use the plugin and then it listens for commands from docker daemon and runs the corresponding code when necessary. Docker plugins can run as a managed plugin under control of the docker daemon or as an independent native service. For testing, you can just run it directly from the command line, for example:

sudo rclone serve docker --base-dir /tmp/rclone-volumes --socket-addr localhost:8787 -vv

Running rclone serve docker will create the said socket, listening for commands from Docker to create the necessary Volumes. Normally you need not give the --socket-addr flag. The API will listen on the unix domain socket at /run/docker/plugins/rclone.sock. In the example above rclone will create a TCP socket and a small file /etc/docker/plugins/rclone.spec containing the socket address. We use sudo because both paths are writeable only by the root user.

If you later decide to change listening socket, the docker daemon must be restarted to reconnect to /run/docker/plugins/rclone.sock or parse new /etc/docker/plugins/rclone.spec. Until you restart, any volume related docker commands will timeout trying to access the old socket. Running directly is supported on Linux only, not on Windows or MacOS. This is not a problem with managed plugin mode described in details in the full documentation.

The command will create volume mounts under the path given by --base-dir (by default /var/lib/docker-volumes/rclone available only to root) and maintain the JSON formatted file docker-plugin.state in the rclone cache directory with book-keeping records of created and mounted volumes.

All mount and VFS options are submitted by the docker daemon via API, but you can also provide defaults on the command line as well as set path to the config file and cache directory or adjust logging verbosity.

Restarting or upgrading the plugin

When the plugin is restarted (for example with docker plugin disable rclone && docker plugin enable rclone, when upgrading the plugin, or when the host reboots) rclone reads its docker-plugin.state file and restores the volumes and mounts that were active before. The plugin starts serving its socket straight away and re-establishes the mounts in the background, so a slow or unreachable remote no longer prevents the plugin from coming back up.

However, restarting the plugin necessarily stops and restarts the process that serves the FUSE mounts. Any container that is already running and holding files open on an rclone volume keeps a handle to the old, now dead mount, so those handles start returning transport endpoint is not connected until the container is restarted. This is a limitation of replacing the process behind a live FUSE mount and cannot be avoided by rclone - the plugin itself recovers and newly started containers work normally, but:

  • Restart any containers that were using rclone volumes after you restart or upgrade the plugin (e.g. docker restart <container>), or stop them before and start them after.
  • Databases and other applications that keep files open continuously (Grafana, Prometheus, SQLite-backed apps, etc.) are the most affected and should always be restarted.

Security

The plugin API accepts a remote (aka fs) option on volume creation, and this is parsed exactly like an rclone connection string. Connection strings are trusted configuration: they may carry inline backend options, and some backends use those options to run local commands (for example the sftp backend's ssh option spawns an external binary). Anyone who can send requests to the plugin socket can therefore make rclone run arbitrary commands as the user running rclone serve docker (typically root). Treat access to the socket as equivalent to that level of access and only expose it to trusted callers.

When listening on the default unix socket at /run/docker/plugins/rclone.sock rclone creates it with mode 0660 owned by root and the group given by --socket-gid (the process GID by default), so only root and members of that group - normally just the docker daemon - can reach it. Do not loosen these permissions or hand the group to untrusted users.

When using --socket-addr to listen on a TCP socket there is no authentication and the API is reachable by anyone who can open the port, so bind it to a loopback or otherwise trusted address and protect it with a firewall. Note that holding Docker daemon access is already equivalent to root on the host, so a caller able to issue docker volume create does not gain anything new from this; the concern is exposing the socket more widely than the daemon itself.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

rclone serve docker [flags]

Options

      --allow-idmap                            Allow id-mapped mounts (Linux 6.12+, mount2 only)
      --allow-non-empty                        Allow mounting over a non-empty directory (not supported on Windows)
      --allow-other                            Allow access to other users (not supported on Windows)
      --allow-root                             Allow access to root user (not supported on Windows)
      --async-read                             Use asynchronous reads (not supported on Windows) (default true)
      --attr-timeout Duration                  Time for which file/directory attributes are cached (default 1s)
      --base-dir string                        Base directory for volumes (default "/var/lib/docker-volumes/rclone")
      --daemon                                 Run mount in background and exit parent process (as background output is suppressed, use --log-file with --log-format=pid,... to monitor) (not supported on Windows)
      --daemon-timeout Duration                Time limit for rclone to respond to kernel (not supported on Windows) (default 0s)
      --daemon-wait Duration                   Time to wait for ready mount from daemon (maximum time on Linux, constant sleep time on OSX/BSD) (not supported on Windows) (default 1m0s)
      --debug-fuse                             Debug the FUSE internals - needs -v
      --default-permissions                    Makes kernel enforce access control based on the file mode (not supported on Windows)
      --devname string                         Set the device name - default is remote:path
      --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                     Directory permissions (default 777)
      --direct-io                              Use Direct IO, disables caching of data
      --file-perms FileMode                    File permissions (default 666)
      --forget-state                           Skip restoring previous state
      --fuse-flag stringArray                  Flags or arguments to be passed direct to libfuse/WinFsp (repeat if required)
      --gid uint32                             Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                   help for docker
      --link-perms FileMode                    Link permissions (default 666)
      --max-read-ahead SizeSuffix              The number of bytes that can be prefetched for sequential reads (not supported on Windows) (default 128Ki)
      --mount-case-insensitive Tristate        Tell the OS the mount is case insensitive (true) or sensitive (false) regardless of the backend (auto) (default unset)
      --network-mode                           Mount as remote network drive, instead of fixed disk drive (supported on Windows only)
      --no-checksum                            Don't compare checksums on up/download
      --no-modtime                             Don't read/write the modification time (can speed things up)
      --no-seek                                Don't allow seeking in files
      --no-spec                                Do not write spec file
      --noappledouble                          Ignore Apple Double (._) and .DS_Store files (supported on OSX only) (default true)
      --noapplexattr                           Ignore all "com.apple.*" extended attributes (supported on OSX only)
  -o, --option stringArray                     Option for libfuse/WinFsp (repeat if required)
      --poll-interval Duration                 Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --read-only                              Only allow read-only access
      --socket-addr string                     Address <host:port> or absolute path (default: /run/docker/plugins/rclone.sock)
      --socket-gid int                         GID for unix socket (default: current process GID) (default 1000)
      --uid uint32                             Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                         Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --vfs-block-norm-dupes                   If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration             Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix          Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix    Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode               Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration       Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                   If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix   Specify the total space of disk (default off)
      --vfs-fast-fingerprint                   Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration            Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                              Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string          Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix              Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix         Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix   If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int             The number of parallel streams to read at once
      --vfs-read-wait Duration                 Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                            Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size           Use the rclone size algorithm for Used size
      --vfs-write-back Duration                Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                Time to wait for in-sequence write before giving error (default 1s)
      --volname string                         Set the volume name (supported on Windows and OSX only)
      --write-back-cache                       Makes kernel buffer writes before sending them to rclone (without this, writethrough caching is used) (not supported on Windows)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

rclone serve ftp

Serve remote:path over FTP.

Synopsis

Run a basic FTP server to serve a remote over FTP protocol. This can be viewed with a FTP client or you can make a remote of type FTP to read and write it.

Server options

Use --addr to specify which IP address and port the server should listen on, e.g. --addr 1.2.3.4:8000 or --addr :8080 to listen to all IPs. By default it only listens on localhost. You can use port :0 to let the OS choose an available port.

If you set --addr to listen on a public or LAN accessible IP address then using Authentication is advised - see the next section for info.

Authentication

By default this will serve files without needing a login.

You can set a single username and password with the --user and --pass flags.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

Auth Proxy

If you supply the parameter --auth-proxy /path/to/program then rclone will use that program to generate backends on the fly which then are used to authenticate incoming requests. This uses a simple JSON based protocol with input on STDIN and output on STDOUT.

PLEASE NOTE: --auth-proxy and --authorized-keys cannot be used together, if --auth-proxy is set the authorized keys option will be ignored.

There is an example program bin/test_proxy.py in the rclone source code.

The program's job is to take a user and pass on the input and turn those into the config for a backend on STDOUT in JSON format. This config will have any default parameters for the backend added, but it won't use configuration from environment variables or command line options - it is the job of the proxy program to make a complete config.

This config generated must have this extra parameter

  • _root - root to use for the backend

And it may have these parameters

  • _obscure - comma separated strings for parameters to obscure
  • _secret_access_key - the secret for S3 access key auth (see below)

If password authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "pass": "mypassword",
  "client_ip": "192.168.1.1"
}

If public-key authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "public_key": "AAAAB3NzaC1yc2EAAAADAQABAAABAQDuwESFdAe14hVS6omeyX7edc...JQdf",
  "client_ip": "192.168.1.1"
}

If the client authenticated with an S3 access key (rclone serve s3), the client never sends its secret, only a signature made with it, so the input contains just the access key ID as the user with no pass or public_key:

{
  "user": "AKIAIOSFODNN7EXAMPLE",
  "client_ip": "192.168.1.1"
}

In this case the program must look up the secret access key for that access key ID and return it in the _secret_access_key field of the output. Rclone then uses that secret to verify the signature on the request, refusing the request if it does not match. This means the proxy program is the source of truth for both the credentials and the backend they map to. If the program does not return _secret_access_key or returns it empty the request is refused.

The program's answer for an access key ID is cached (see below) but is checked with the program again after 5 minutes even if the access key ID is in constant use, so revoking an access key ID in the program takes effect within 5 minutes. A rotated secret takes effect on the first request signed with it.

The client_ip key holds the IP address the client connected from, without a port number. It can be used to restrict logins to certain networks, or to log authentication attempts centrally. It is omitted if the client has no IP address, for example when connecting over a unix socket. Note that if rclone is behind a reverse proxy this will be the address of the reverse proxy and not the original client.

And as an example return this on STDOUT

{
  "type": "sftp",
  "_root": "",
  "_obscure": "pass",
  "user": "me",
  "pass": "mypassword",
  "host": "sftp.example.com"
}

This would mean that an SFTP backend would be created on the fly for the user and pass/public_key returned in the output to the host given. Note that since _obscure is set to pass, rclone will obscure the pass parameter before creating the backend (which is required for sftp backends).

The program can manipulate the supplied user in any way, for example to make proxy to many different sftp backends, you could make the user be [email protected] and then set the host to example.com in the output and the user to user. For security you'd probably want to restrict the host to a limited list.

An internal cache of backends is keyed on the user, the kind of credential presented, a hash of the pass or public_key, and the client_ip. This means that if a user's password or public-key changes, the client connects from a new IP address, or the proxy returns different config parameters (eg a rotated api_key), a fresh backend will be created on the next request rather than the cached one being reused. A backend created for a public key login is never reused for a password login, so a public key cannot be sent as a password to skip the program.

A cached answer is checked with the program again after 5 minutes even if the user is in constant use, so revoking a user in the program takes effect within 5 minutes. For serve sftp this applies to new logins - an existing connection is not closed.

This can be used to build general purpose proxies to any kind of backend that rclone supports.

rclone serve ftp remote:path [flags]

Options

      --addr string                            IPaddress:Port or :Port to bind server to (default "localhost:2121")
      --auth-proxy string                      A program to use to create the backend from the auth
      --cert string                            TLS PEM key (concatenation of certificate and CA certificate)
      --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                     Directory permissions (default 777)
      --file-perms FileMode                    File permissions (default 666)
      --gid uint32                             Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                   help for ftp
      --key string                             TLS PEM Private key
      --link-perms FileMode                    Link permissions (default 666)
      --no-checksum                            Don't compare checksums on up/download
      --no-modtime                             Don't read/write the modification time (can speed things up)
      --no-seek                                Don't allow seeking in files
      --pass string                            Password for authentication (empty value allow every password)
      --passive-port string                    Passive port range to use (default "30000-32000")
      --poll-interval Duration                 Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --public-ip string                       Public IP address to advertise for passive connections
      --read-only                              Only allow read-only access
      --uid uint32                             Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                         Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --user string                            User name for authentication (default "anonymous")
      --vfs-block-norm-dupes                   If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration             Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix          Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix    Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode               Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration       Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                   If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix   Specify the total space of disk (default off)
      --vfs-fast-fingerprint                   Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration            Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                              Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string          Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix              Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix         Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix   If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int             The number of parallel streams to read at once
      --vfs-read-wait Duration                 Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                            Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size           Use the rclone size algorithm for Used size
      --vfs-write-back Duration                Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                Time to wait for in-sequence write before giving error (default 1s)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

rclone serve http

Serve the remote over HTTP.

Synopsis

Run a basic web server to serve a remote over HTTP. This can be viewed in a web browser or you can make a remote of type http read from it.

You can use the filter flags (e.g. --include, --exclude) to control what is served.

The server will log errors. Use -v to see access logs.

--bwlimit will be respected for file transfers. Use --stats to control the stats printing.

Server options

Use --addr to specify which IP address and port the server should listen on, eg --addr 1.2.3.4:8000 or --addr :8080 to listen to all IPs. By default it only listens on localhost. You can use port :0 to let the OS choose an available port.

If you set --addr to listen on a public or LAN accessible IP address then using Authentication is advised - see the next section for info.

You can use a unix socket by setting the url to unix:///path/to/socket or just by using an absolute path name.

--addr may be repeated to listen on multiple IPs/ports/sockets. Socket activation, described further below, can also be used to accomplish the same.

--server-read-timeout and --server-write-timeout can be used to control the timeouts on the server. Note that this is the total time for a transfer.

--max-header-bytes controls the maximum number of bytes the server will accept in the HTTP header.

--response-header can be used to set an HTTP header for all responses, will overriding existing values. The flag may be repeated to add multiple headers. Use the format Header-Name: value.

--allow-origin enables cross-origin (CORS) requests from browsers by setting the Access-Control-Allow-Origin header on every response to the origin given, e.g. --allow-origin https://app.example.com. Browser CORS preflight OPTIONS requests are answered by rclone without needing authentication as browsers never send credentials with them. Setting this to * allows any origin which is a security risk on an authenticated server.

--baseurl controls the URL prefix that rclone serves from. By default rclone will serve from the root. If you used --baseurl "/rclone" then rclone would serve from a URL starting with "/rclone/". This is useful if you wish to proxy rclone serve. Rclone automatically inserts leading and trailing "/" on --baseurl, so --baseurl "rclone", --baseurl "/rclone" and --baseurl "/rclone/" are all treated identically.

--disable-zip may be set to disable the zipping download option.

Protocol

The server supports HTTP/1.1 and HTTP/2. HTTP/2 is used automatically for TLS connections. For non-TLS connections, HTTP/2 cleartext (h2c) is supported, allowing HTTP/2 without encryption.

TLS (SSL)

By default this will serve over http. If you want you can serve over https. You will need to supply the --cert and --key flags. If you wish to do client side certificate validation then you will need to supply --client-ca also.

When TLS is configured every listener given with --addr serves TLS. An individual listener can be prefixed with http:// to serve unencrypted HTTP on that address, or with tls:// to state explicitly that it must serve TLS. Using a tls:// address without --cert and --key is an error.

--cert must be set to the path of a file containing either a PEM encoded certificate, or a concatenation of that with the CA certificate. --key must be set to the path of a file with the PEM encoded private key. If setting --client-ca, it should be set to the path of a file with PEM encoded client certificate authority certificates.

--min-tls-version is minimum TLS version that is acceptable. Valid values are "tls1.0", "tls1.1", "tls1.2" and "tls1.3" (default "tls1.0").

Socket activation

Instead of the listening addresses specified above, rclone will listen to all FDs passed by the service manager, if any (and ignore any arguments passed by --addr).

This allows rclone to be a socket-activated service. It can be configured with .socket and .service unit files as described in https://www.freedesktop.org/software/systemd/man/latest/systemd.socket.html.

Socket activation can be tested ad-hoc with the systemd-socket-activatecommand

systemd-socket-activate -l 8000 -- rclone serve

This will socket-activate rclone on the first connection to port 8000 over TCP.

Template

--template allows a user to specify a custom markup template for HTTP and WebDAV serve functions. The server exports the following markup to be used within the template to server pages:

Parameter Subparameter Description
.Name The full path of a file/directory.
.Title Directory listing of '.Name'.
.Sort The current sort used. This is changeable via '?sort=' parameter. Possible values: namedirfirst, name, size, time (default namedirfirst).
.Order The current ordering used. This is changeable via '?order=' parameter. Possible values: asc, desc (default asc).
.Query Currently unused.
.Path The path of the directory from the root with leading and trailing slashes, e.g. '/dir/'.
.IsRoot Boolean for if this is the root directory.
.UpLink The link to the parent directory.
.Static Boolean for if the listing is being written as a static page by rclone index.
.NumDirs The number of directories in the listing.
.NumFiles The number of files in the listing.
.TotalSize The total size in bytes of the files in the listing.
.Breadcrumb Allows for creating a relative navigation.
.Link The link of the Text relative to the root.
.Text The Name of the directory.
.Entries Information about a specific file/directory.
.URL The url of an entry.
.Leaf Currently same as '.URL' but intended to be just the name.
.IsDir Boolean for if an entry is a directory or not.
.Size Size in bytes of the entry.
.ModTime The UTC timestamp of an entry.
.MimeType The MIME type of an entry guessed from its name, or 'inode/directory'.

The server also makes the following functions available so that they can be used within the template. These functions help extend the options for dynamic rendering of HTML. They can be used to render HTML based on specific conditions.

Function Description
afterEpoch Returns the time since the epoch for the given time.
contains Checks whether a given substring is present or not in a given string.
hasPrefix Checks whether the given string begins with the specified prefix.
hasSuffix Checks whether the given string end with the specified suffix.

Authentication

By default this will serve files without needing a login.

You can either use an htpasswd file which can take lots of users, or set a single username and password with the --user and --pass flags.

Alternatively, you can have the reverse proxy manage authentication and use the username provided in the configured header with --user-from-header (e.g., --user-from-header=x-remote-user). Ensure the proxy is trusted and headers cannot be spoofed, as misconfiguration may lead to unauthorized access.

If either of the above authentication methods is not configured and client certificates are required by the --client-ca flag passed to the server, the client certificate common name will be considered as the username.

Use --htpasswd /path/to/htpasswd to provide an htpasswd file. This is in standard apache format and supports MD5, SHA1 and BCrypt for basic authentication. Bcrypt is recommended.

To create an htpasswd file:

touch htpasswd
htpasswd -B htpasswd user
htpasswd -B htpasswd anotherUser

Use the same hash format and cost for every entry in the file. If the entries differ, the time taken to reject a wrong password can reveal which usernames exist.

The password file can be updated while rclone is running.

When authentication is enabled every request must be authenticated, including HTTP OPTIONS requests, as the answer to an OPTIONS request can reveal whether a path exists. Health checks or monitoring which use unauthenticated OPTIONS requests will need to send credentials. The exception is browser CORS preflight requests when --allow-origin is set, which are answered without authentication as browsers never send credentials with them.

Use --realm to set the authentication realm.

Use --salt to change the password hashing salt from the default.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

Auth Proxy

If you supply the parameter --auth-proxy /path/to/program then rclone will use that program to generate backends on the fly which then are used to authenticate incoming requests. This uses a simple JSON based protocol with input on STDIN and output on STDOUT.

PLEASE NOTE: --auth-proxy and --authorized-keys cannot be used together, if --auth-proxy is set the authorized keys option will be ignored.

There is an example program bin/test_proxy.py in the rclone source code.

The program's job is to take a user and pass on the input and turn those into the config for a backend on STDOUT in JSON format. This config will have any default parameters for the backend added, but it won't use configuration from environment variables or command line options - it is the job of the proxy program to make a complete config.

This config generated must have this extra parameter

  • _root - root to use for the backend

And it may have these parameters

  • _obscure - comma separated strings for parameters to obscure
  • _secret_access_key - the secret for S3 access key auth (see below)

If password authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "pass": "mypassword",
  "client_ip": "192.168.1.1"
}

If public-key authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "public_key": "AAAAB3NzaC1yc2EAAAADAQABAAABAQDuwESFdAe14hVS6omeyX7edc...JQdf",
  "client_ip": "192.168.1.1"
}

If the client authenticated with an S3 access key (rclone serve s3), the client never sends its secret, only a signature made with it, so the input contains just the access key ID as the user with no pass or public_key:

{
  "user": "AKIAIOSFODNN7EXAMPLE",
  "client_ip": "192.168.1.1"
}

In this case the program must look up the secret access key for that access key ID and return it in the _secret_access_key field of the output. Rclone then uses that secret to verify the signature on the request, refusing the request if it does not match. This means the proxy program is the source of truth for both the credentials and the backend they map to. If the program does not return _secret_access_key or returns it empty the request is refused.

The program's answer for an access key ID is cached (see below) but is checked with the program again after 5 minutes even if the access key ID is in constant use, so revoking an access key ID in the program takes effect within 5 minutes. A rotated secret takes effect on the first request signed with it.

The client_ip key holds the IP address the client connected from, without a port number. It can be used to restrict logins to certain networks, or to log authentication attempts centrally. It is omitted if the client has no IP address, for example when connecting over a unix socket. Note that if rclone is behind a reverse proxy this will be the address of the reverse proxy and not the original client.

And as an example return this on STDOUT

{
  "type": "sftp",
  "_root": "",
  "_obscure": "pass",
  "user": "me",
  "pass": "mypassword",
  "host": "sftp.example.com"
}

This would mean that an SFTP backend would be created on the fly for the user and pass/public_key returned in the output to the host given. Note that since _obscure is set to pass, rclone will obscure the pass parameter before creating the backend (which is required for sftp backends).

The program can manipulate the supplied user in any way, for example to make proxy to many different sftp backends, you could make the user be [email protected] and then set the host to example.com in the output and the user to user. For security you'd probably want to restrict the host to a limited list.

An internal cache of backends is keyed on the user, the kind of credential presented, a hash of the pass or public_key, and the client_ip. This means that if a user's password or public-key changes, the client connects from a new IP address, or the proxy returns different config parameters (eg a rotated api_key), a fresh backend will be created on the next request rather than the cached one being reused. A backend created for a public key login is never reused for a password login, so a public key cannot be sent as a password to skip the program.

A cached answer is checked with the program again after 5 minutes even if the user is in constant use, so revoking a user in the program takes effect within 5 minutes. For serve sftp this applies to new logins - an existing connection is not closed.

This can be used to build general purpose proxies to any kind of backend that rclone supports.

rclone serve http remote:path [flags]

Options

      --addr stringArray                       IPaddress:Port or :Port to bind server to (default 127.0.0.1:8080)
      --allow-origin string                    Origin which cross-domain request (CORS) can be executed from
      --auth-proxy string                      A program to use to create the backend from the auth
      --baseurl string                         Prefix for URLs - leave blank for root
      --cert string                            TLS PEM key (concatenation of certificate and CA certificate)
      --client-ca string                       Client certificate authority to verify clients with
      --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                     Directory permissions (default 777)
      --disable-dir-list                       Disable HTML directory list on GET request for a directory
      --disable-zip                            Disable zip download of directories
      --file-perms FileMode                    File permissions (default 666)
      --gid uint32                             Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                   help for http
      --htpasswd string                        A htpasswd file - if not provided no authentication is done
      --key string                             TLS PEM Private key
      --link-perms FileMode                    Link permissions (default 666)
      --max-header-bytes int                   Maximum size of request header (default 4096)
      --min-tls-version string                 Minimum TLS version that is acceptable (default "tls1.0")
      --no-checksum                            Don't compare checksums on up/download
      --no-modtime                             Don't read/write the modification time (can speed things up)
      --no-seek                                Don't allow seeking in files
      --pass string                            Password for authentication
      --poll-interval Duration                 Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --read-only                              Only allow read-only access
      --realm string                           Realm for authentication
      --response-header stringArray            Set HTTP header for all responses, overriding existing values
      --salt string                            Password hashing salt (default "dlPL2MqE")
      --server-read-timeout Duration           Timeout for server reading data (default 1h0m0s)
      --server-write-timeout Duration          Timeout for server writing data (default 1h0m0s)
      --template string                        User-specified template
      --uid uint32                             Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                         Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --user string                            User name for authentication
      --user-from-header string                User name from a defined HTTP header
      --vfs-block-norm-dupes                   If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration             Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix          Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix    Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode               Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration       Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                   If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix   Specify the total space of disk (default off)
      --vfs-fast-fingerprint                   Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration            Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                              Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string          Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix              Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix         Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix   If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int             The number of parallel streams to read at once
      --vfs-read-wait Duration                 Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                            Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size           Use the rclone size algorithm for Used size
      --vfs-write-back Duration                Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                Time to wait for in-sequence write before giving error (default 1s)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

rclone serve nfs

Serve the remote as an NFS mount

Synopsis

Create an NFS server that serves the given remote over the network.

This implements an NFSv3 server to serve any rclone remote via NFS.

The primary purpose for this command is to enable the mount command on recent macOS versions where installing FUSE is very cumbersome.

This server does not implement any authentication so any client will be able to access the data. To limit access, you can use serve nfs on the loopback address or rely on secure tunnels (such as SSH) or use firewalling.

For this reason, by default, a random TCP port is chosen and the loopback interface is used for the listening address by default; meaning that it is only available to the local machine. If you want other machines to access the NFS mount over local network, you need to specify the listening address and port using the --addr flag.

Modifying files through the NFS protocol requires VFS caching. Usually you will need to specify --vfs-cache-mode in order to be able to write to the mountpoint (full is recommended). If you don't specify VFS cache mode, the mount will be read-only.

--nfs-cache-type controls the type of the NFS handle cache. By default this is memory where new handles will be randomly allocated when needed. These are stored in memory. If the server is restarted the handle cache will be lost and connected NFS clients will get stale handle errors.

--nfs-cache-type disk uses an on disk NFS handle cache. Rclone hashes the path of the object and stores it in a file named after the hash. These hashes are stored on disk the directory controlled by --cache-dir or the exact directory may be specified with --nfs-cache-dir. Using this means that the NFS server can be restarted at will without affecting the connected clients.

--nfs-cache-type symlink is similar to --nfs-cache-type disk in that it uses an on disk cache, but the cache entries are held as symlinks. Rclone will use the handle of the underlying file as the NFS handle which improves performance. This sort of cache can't be backed up and restored as the underlying handles will change. This is Linux only. It requires running rclone as root or with CAP_DAC_READ_SEARCH. You can run rclone with this extra permission by doing this to the rclone binary sudo setcap cap_dac_read_search+ep /path/to/rclone.

--nfs-cache-handle-limit controls the maximum number of cached NFS handles stored by the caching handler. This should not be set too low or you may experience errors when trying to access files. The default is 1000000, but consider lowering this limit if the server's system resource usage causes problems. This is only used by the memory type cache.

To serve NFS over the network use following command:

rclone serve nfs remote: --addr 0.0.0.0:$PORT --vfs-cache-mode=full

This specifies a port that can be used in the mount command. To mount the server under Linux/macOS, use the following command:

mount -t nfs -o port=$PORT,mountport=$PORT,tcp $HOSTNAME:/ path/to/mountpoint

Where $PORT is the same port number used in the serve nfs command and $HOSTNAME is the network address of the machine that serve nfs was run on.

NFS clients can also mount a subdirectory of the served remote by including it in the mount path. For example to mount only the photos/2024 subdirectory:

mount -t nfs -o port=$PORT,mountport=$PORT,tcp $HOSTNAME:/photos/2024 path/to/mountpoint

The subpath is resolved within the served remote and must refer to an existing directory (not a file or a symlink). Subpath mounts are a convenience equivalent to mounting / and changing directory: they share access to the same underlying VFS and the same file handles, so they do not isolate the client from siblings or parents of the mounted subdirectory.

If --vfs-metadata-extension is in use then for the --nfs-cache-type disk and --nfs-cache-type cache the metadata files will have the file handle of their parent file suffixed with 0x00, 0x00, 0x00, 0x01. This means they can be looked up directly from the parent file handle is desired.

This command is only available on Unix platforms.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

rclone serve nfs remote:path [flags]

Options

      --addr string                            IPaddress:Port or :Port to bind server to
      --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                     Directory permissions (default 777)
      --file-perms FileMode                    File permissions (default 666)
      --gid uint32                             Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                   help for nfs
      --link-perms FileMode                    Link permissions (default 666)
      --nfs-cache-dir string                   The directory the NFS handle cache will use if set
      --nfs-cache-handle-limit int             max file handles cached simultaneously (min 5) (default 1000000)
      --nfs-cache-type memory|disk|symlink     Type of NFS handle cache to use (default memory)
      --no-checksum                            Don't compare checksums on up/download
      --no-modtime                             Don't read/write the modification time (can speed things up)
      --no-seek                                Don't allow seeking in files
      --poll-interval Duration                 Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --read-only                              Only allow read-only access
      --uid uint32                             Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                         Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --vfs-block-norm-dupes                   If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration             Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix          Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix    Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode               Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration       Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                   If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix   Specify the total space of disk (default off)
      --vfs-fast-fingerprint                   Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration            Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                              Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string          Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix              Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix         Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix   If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int             The number of parallel streams to read at once
      --vfs-read-wait Duration                 Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                            Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size           Use the rclone size algorithm for Used size
      --vfs-write-back Duration                Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                Time to wait for in-sequence write before giving error (default 1s)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

rclone serve restic

Serve the remote for restic's REST API.

Synopsis

Run a basic web server to serve a remote over restic's REST backend API over HTTP. This allows restic to use rclone as a data storage mechanism for cloud providers that restic does not support directly.

Restic is a command-line program for doing backups.

The server will log errors. Use -v to see access logs.

--bwlimit will be respected for file transfers. Use --stats to control the stats printing.

Setting up rclone for use by restic

First set up a remote for your chosen cloud provider.

Once you have set up the remote, check it is working with, for example "rclone lsd remote:". You may have called the remote something other than "remote:" - just substitute whatever you called it in the following instructions.

Now start the rclone restic server

rclone serve restic -v remote:backup

Where you can replace "backup" in the above by whatever path in the remote you wish to use.

By default this will serve on "localhost:8080" you can change this with use of the --addr flag.

You might wish to start this server on boot.

Adding --cache-objects=false will cause rclone to stop caching objects returned from the List call. Caching is normally desirable as it speeds up downloading objects, saves transactions and uses very little memory.

Setting up restic to use rclone

Now you can follow the restic instructions on setting up restic.

Note that you will need restic 0.8.2 or later to interoperate with rclone.

For the example above you will want to use "http://localhost:8080/" as the URL for the REST server.

For example:

$ export RESTIC_REPOSITORY=rest:http://localhost:8080/
$ export RESTIC_PASSWORD=yourpassword
$ restic init
created restic backend 8b1a4b56ae at rest:http://localhost:8080/

Please note that knowledge of your password is required to access
the repository. Losing your password means that your data is
irrecoverably lost.
$ restic backup /path/to/files/to/backup
scan [/path/to/files/to/backup]
scanned 189 directories, 312 files in 0:00
[0:00] 100.00%  38.128 MiB / 38.128 MiB  501 / 501 items  0 errors  ETA 0:00
duration: 0:00
snapshot 45c8fdd8 saved

Multiple repositories

Note that you can use the endpoint to host multiple repositories. Do this by adding a directory name or path after the URL. Note that these must end with /. Eg

$ export RESTIC_REPOSITORY=rest:http://localhost:8080/user1repo/
# backup user1 stuff
$ export RESTIC_REPOSITORY=rest:http://localhost:8080/user2repo/
# backup user2 stuff

Private repositories

The--private-repos flag can be used to limit users to repositories starting with a path of /<username>/.

Server options

Use --addr to specify which IP address and port the server should listen on, eg --addr 1.2.3.4:8000 or --addr :8080 to listen to all IPs. By default it only listens on localhost. You can use port :0 to let the OS choose an available port.

If you set --addr to listen on a public or LAN accessible IP address then using Authentication is advised - see the next section for info.

You can use a unix socket by setting the url to unix:///path/to/socket or just by using an absolute path name.

--addr may be repeated to listen on multiple IPs/ports/sockets. Socket activation, described further below, can also be used to accomplish the same.

--server-read-timeout and --server-write-timeout can be used to control the timeouts on the server. Note that this is the total time for a transfer.

--max-header-bytes controls the maximum number of bytes the server will accept in the HTTP header.

--response-header can be used to set an HTTP header for all responses, will overriding existing values. The flag may be repeated to add multiple headers. Use the format Header-Name: value.

--allow-origin enables cross-origin (CORS) requests from browsers by setting the Access-Control-Allow-Origin header on every response to the origin given, e.g. --allow-origin https://app.example.com. Browser CORS preflight OPTIONS requests are answered by rclone without needing authentication as browsers never send credentials with them. Setting this to * allows any origin which is a security risk on an authenticated server.

--baseurl controls the URL prefix that rclone serves from. By default rclone will serve from the root. If you used --baseurl "/rclone" then rclone would serve from a URL starting with "/rclone/". This is useful if you wish to proxy rclone serve. Rclone automatically inserts leading and trailing "/" on --baseurl, so --baseurl "rclone", --baseurl "/rclone" and --baseurl "/rclone/" are all treated identically.

--disable-zip may be set to disable the zipping download option.

Protocol

The server supports HTTP/1.1 and HTTP/2. HTTP/2 is used automatically for TLS connections. For non-TLS connections, HTTP/2 cleartext (h2c) is supported, allowing HTTP/2 without encryption.

TLS (SSL)

By default this will serve over http. If you want you can serve over https. You will need to supply the --cert and --key flags. If you wish to do client side certificate validation then you will need to supply --client-ca also.

When TLS is configured every listener given with --addr serves TLS. An individual listener can be prefixed with http:// to serve unencrypted HTTP on that address, or with tls:// to state explicitly that it must serve TLS. Using a tls:// address without --cert and --key is an error.

--cert must be set to the path of a file containing either a PEM encoded certificate, or a concatenation of that with the CA certificate. --key must be set to the path of a file with the PEM encoded private key. If setting --client-ca, it should be set to the path of a file with PEM encoded client certificate authority certificates.

--min-tls-version is minimum TLS version that is acceptable. Valid values are "tls1.0", "tls1.1", "tls1.2" and "tls1.3" (default "tls1.0").

Socket activation

Instead of the listening addresses specified above, rclone will listen to all FDs passed by the service manager, if any (and ignore any arguments passed by --addr).

This allows rclone to be a socket-activated service. It can be configured with .socket and .service unit files as described in https://www.freedesktop.org/software/systemd/man/latest/systemd.socket.html.

Socket activation can be tested ad-hoc with the systemd-socket-activatecommand

systemd-socket-activate -l 8000 -- rclone serve

This will socket-activate rclone on the first connection to port 8000 over TCP.

Authentication

By default this will serve files without needing a login.

You can either use an htpasswd file which can take lots of users, or set a single username and password with the --user and --pass flags.

Alternatively, you can have the reverse proxy manage authentication and use the username provided in the configured header with --user-from-header (e.g., --user-from-header=x-remote-user). Ensure the proxy is trusted and headers cannot be spoofed, as misconfiguration may lead to unauthorized access.

If either of the above authentication methods is not configured and client certificates are required by the --client-ca flag passed to the server, the client certificate common name will be considered as the username.

Use --htpasswd /path/to/htpasswd to provide an htpasswd file. This is in standard apache format and supports MD5, SHA1 and BCrypt for basic authentication. Bcrypt is recommended.

To create an htpasswd file:

touch htpasswd
htpasswd -B htpasswd user
htpasswd -B htpasswd anotherUser

Use the same hash format and cost for every entry in the file. If the entries differ, the time taken to reject a wrong password can reveal which usernames exist.

The password file can be updated while rclone is running.

When authentication is enabled every request must be authenticated, including HTTP OPTIONS requests, as the answer to an OPTIONS request can reveal whether a path exists. Health checks or monitoring which use unauthenticated OPTIONS requests will need to send credentials. The exception is browser CORS preflight requests when --allow-origin is set, which are answered without authentication as browsers never send credentials with them.

Use --realm to set the authentication realm.

Use --salt to change the password hashing salt from the default.

rclone serve restic remote:path [flags]

Options

      --addr stringArray                IPaddress:Port or :Port to bind server to (default 127.0.0.1:8080)
      --allow-origin string             Origin which cross-domain request (CORS) can be executed from
      --append-only                     Disallow deletion of repository data
      --baseurl string                  Prefix for URLs - leave blank for root
      --cache-objects                   Cache listed objects (default true)
      --cert string                     TLS PEM key (concatenation of certificate and CA certificate)
      --client-ca string                Client certificate authority to verify clients with
  -h, --help                            help for restic
      --htpasswd string                 A htpasswd file - if not provided no authentication is done
      --key string                      TLS PEM Private key
      --max-header-bytes int            Maximum size of request header (default 4096)
      --min-tls-version string          Minimum TLS version that is acceptable (default "tls1.0")
      --pass string                     Password for authentication
      --private-repos                   Users can only access their private repo
      --realm string                    Realm for authentication
      --response-header stringArray     Set HTTP header for all responses, overriding existing values
      --salt string                     Password hashing salt (default "dlPL2MqE")
      --server-read-timeout Duration    Timeout for server reading data (default 1h0m0s)
      --server-write-timeout Duration   Timeout for server writing data (default 1h0m0s)
      --stdio                           Run an HTTP2 server on stdin/stdout
      --user string                     User name for authentication
      --user-from-header string         User name from a defined HTTP header

See the global flags page for global options not listed here.

See Also

rclone serve s3

Serve remote:path over s3.

Synopsis

serve s3 implements a basic s3 server that serves a remote via s3. This can be viewed with an s3 client, or you can make an s3 type remote to read and write to it with rclone.

serve s3 is considered Experimental so use with care.

S3 server supports Signature Version 4 authentication. Just use --auth-key accessKey,secretKey and set the Authorization header correctly in the request. (See the AWS docs).

--auth-key can be repeated for multiple auth pairs. If --auth-key is not provided then serve s3 will allow anonymous access.

Alternatively --auth-proxy can be used to look up the secret for each access key ID and choose the backend it maps to (see Auth Proxy below). When an auth proxy is in use --auth-key is ignored and every request must be signed with the secret the proxy returns for its access key ID. Each access key ID is a separate user, whose multipart uploads and object metadata are private to it, even if the proxy maps several access key IDs to the same backend.

As on AWS, every x-amz-* header in an authenticated request must be in its signed headers list; a request with an unsigned x-amz-* header is refused with AccessDenied. S3 clients sign these headers as a matter of course, but a header added after signing (for example by a proxy in front of the server, or x-amz-meta-* headers sent with a presigned URL which only signs the host) is refused.

Like all rclone flags --auth-key can be set via environment variables, in this case RCLONE_AUTH_KEY. Since this flag can be repeated, the input to RCLONE_AUTH_KEY is CSV encoded. Because the accessKey,secretKey has a comma in, this means it needs to be in quotes.

export RCLONE_AUTH_KEY='"user,pass"'
rclone serve s3 ...

Or to supply multiple identities:

export RCLONE_AUTH_KEY='"user1,pass1","user2,pass2"'
rclone serve s3 ...

Setting this variable without quotes will produce an error.

Please note that some clients may require HTTPS endpoints. See the SSL docs for more information.

This command uses the VFS directory cache. All the functionality will work with --vfs-cache-mode off. Using --vfs-cache-mode full (or writes) can be used to cache objects locally to improve performance.

Use --force-path-style=false if you want to use the bucket name as a part of the hostname (such as mybucket.local)

Use --etag-hash if you want to change the hash uses for the ETag. Note that using anything other than MD5 (the default) is likely to cause problems for S3 clients which rely on the Etag being the MD5. Use --etag-hash auto to use the best hash the backend supports - with --auth-proxy that of each user's backend.

The ETag of an object made by a multipart upload is its hash, like any other object, rather than the -N suffixed ETag AWS S3 gives these. If the remote doesn't have the MD5 of objects, HeadObject and listings only return the MD5 of objects uploaded through serve s3 while the object's metadata is kept (see below).

Quickstart

For a simple set up, to serve remote:path over s3, run the server like this:

rclone serve s3 --auth-key ACCESS_KEY_ID,SECRET_ACCESS_KEY remote:path

For example, to use a simple folder in the filesystem, run the server with a command like this:

rclone serve s3 --auth-key ACCESS_KEY_ID,SECRET_ACCESS_KEY local:/path/to/folder

The rclone.conf for the server could look like this:

[local]
type = local

The local configuration is optional though. If you run the server with a remote:path like /path/to/folder (without the local: prefix and without an rclone.conf file), rclone will fall back to a default configuration, which will be visible as a warning in the logs. But it will run nonetheless.

This will be compatible with an rclone (client) remote configuration which is defined like this:

[serves3]
type = s3
provider = Rclone
endpoint = http://127.0.0.1:8080/
access_key_id = ACCESS_KEY_ID
secret_access_key = SECRET_ACCESS_KEY

Object uploads (PUT)

A PutObject upload only ever changes the object at its key atomically, on success, a failed or interrupted PUT neither removes nor overwrites the object already stored at the key, and never leaves a partial object visible at it.

Remotes that upload atomically (e.g. object stores such as s3) are streamed straight to the destination. On remotes where a partial upload would otherwise be visible (e.g. local), and whenever --vfs-cache-mode is writes or above, the upload is written to a temporary object that is renamed into place on success; these remotes need to support a server-side move or copy for this (nearly all do - without move or copy the upload is written directly and a failed PUT may leave a partial object at the key). If serve s3 is killed part-way through an upload the temporary object (named with a leading .rclone_temp_put_) may be left behind; it is hidden from S3 listings but must be removed manually.

Multipart uploads

Multipart uploads are written, in part-number order, to a temporary object which is renamed into place, server-side, on completion, so the upload is atomic. The object at the key only ever changes on a successful completion. A failed or aborted upload never affects any object already stored under that name and a partly-uploaded object never becomes visible under it.

With the default --vfs-cache-mode off serve s3 streams each multipart upload, in part-number order, into a single streaming upload to the underlying remote, so the whole file is never buffered in memory: only parts which arrive out of order are buffered, within the limits described below. The remote then performs its own internal upload (for example its own multipart upload, still with bounded memory). Remotes that don't support streaming uploads (those that must know the file size before the upload starts, such as onedrive, pcloud, jottacloud, mailru, opendrive, putio, protondrive and zoho) have the parts spooled to a temporary file on local disk instead, and uploaded with the size then known on completion, so they need local disk space for the largest objects in flight rather than memory.

With --vfs-cache-mode writes (or full) the parts are written to a temporary file in the VFS cache and uploaded by the VFS write-back - see Multipart uploads and the VFS cache below.

The rename into place needs the remote to support a server-side move or copy, which nearly all do. It is a cheap rename on most remotes, but on object stores without a real rename (such as s3 itself) the move is performed as a server-side copy and delete of the whole object, which can take time and API calls for large objects. Concurrent multipart uploads of the same key (which S3 permits) are safe. Each writes its own temporary object and the last to complete wins.

On the few remotes that support neither server side move nor copy, the parts are written straight to the destination object instead and never buffered in memory. This is at some cost in atomicity - the incomplete object is visible under its final name while the upload is in flight, as it also is for a plain object PUT on such remotes, and concurrent multipart uploads of the same key write to the same object and can interleave. A failed or aborted upload still leaves any pre-existing object untouched provided the remote uploads atomically and the VFS cache is off; on a remote where partial uploads are visible it may leave partial data at the key (like a plain PUT there), and with --vfs-cache-mode writes (or full) a write to the cache cannot be abandoned, so an aborted upload's partial data is written back to the remote as if it had completed.

Features

  • The whole object is never buffered in memory; only parts arriving out of order are, within fixed limits, whatever the upload size.
  • Parts can be any size. Clients that don't produce uniform-sized parts work fine - for example PostgreSQL backup tools such as pgBarman and pgBackRest, which flush an upload buffer once it grows past the chunk size, so each part is the chunk size plus a variable overshoot.
  • Works through crypt for any part size, since the object is encrypted as one continuous stream.
  • The destination object only ever changes atomically, on completion: an aborted or failed upload leaves any pre-existing object of the same name untouched, and a partly-uploaded object never becomes visible (except on the few remotes with no server-side move or copy, as above).
  • Multipart uploads go through the VFS like any other upload, so they show in rclone's transfer stats and obey --bwlimit.
  • Backend-agnostic - it only needs the remote to support a server-side move or copy for the rename into place, which nearly all do; a remote without streaming upload support spools to local disk as above.

Limitations

  • Parts must arrive in ascending, contiguous part-number order (1, 2, 3, ...). The part the stream needs next is written straight through as it arrives, so it is never buffered, whatever its size. Parts the client uploads concurrently or out of order are buffered until their turn. The memory used for this buffering is capped, per upload, by --multipart-streaming-buffer-limit (default 256M, 0 for no limit), and across all the uploads of each user (each access key) by --multipart-streaming-buffer-total (default 1G, 0 for no limit), counting each buffered part in the whole 1 MiB memory pages it occupies: a part that would take the buffer over either limit is stalled until the stream drains (or, if it is bigger than the limits, until it is the next part), so a client that uploads faster than the remote can accept sees backpressure rather than unbounded server memory use. A stalled part holds its HTTP request open, and one stalled for more than a minute is failed with a SlowDown error, which S3 clients retry. Clients whose upload concurrency times part size exceeds the limits will see these stalls and retries when the remote is slower than the client. Non-contiguous part numbers are rejected on completion. Configure the client to upload in part order, ideally with low concurrency, for the lowest memory use.
  • A part uploaded again before completion - typically a client retrying after a timeout - is accepted: if the earlier copy is still buffered it is replaced, and if it has already been streamed an identical re-upload is a no-op (a re-upload which arrives while the part is being streamed waits for it to finish first). If a part fails part way through being streamed (for example the client's connection drops) the client's retry of it carries on where it stopped, provided it has the same content. What isn't possible is replacing a part that has already been streamed, in whole or in part, with different content - that is rejected. A part which arrives in full but is then rejected, for example because it doesn't match the Content-MD5 the client declared, has already been streamed, so the whole upload is failed and the client must start it again. So is a failure in the stream to the remote itself. (The remote's own upload still retries its internal chunks.)
  • A CompleteMultipartUpload which fails because the parts it lists don't match those uploaded, or there is a gap in the part numbers, leaves the upload open, as on S3, so the client can upload any missing parts and complete it again. If writing the completed object to the remote fails, the upload is failed and must be started again.
  • Parts are serialised into one stream, so ingest from the client is effectively single-threaded. When streaming, the remote's own upload runs concurrently with the parts arriving; with the local disk spool or the VFS cache the upload to the remote only starts on completion.
  • If serve s3 is killed part-way through an upload the temporary object (named with a leading .rclone_temp_multipart_) may be left behind; it is hidden from S3 listings but must be removed manually.

Multipart uploads and the VFS cache

With --vfs-cache-mode writes (or full) multipart uploads do not stream to the remote at all. The parts are written, in part-number order, to a temporary file in the VFS cache. On completion the file is renamed into place and uploaded by the VFS write-back, exactly like a plain object PUT. This needs no streaming upload support from the remote. The rename normally happens in the cache before the upload has started, but the VFS requires the remote to support a server-side move or copy to rename files at all (and uses one if the temporary file has already been written back, e.g. with --vfs-write-back 0). On remotes without either, the parts are written to the cache directly under the final key instead: the upload still never touches memory, but it loses its atomicity - the in-flight upload is visible at the key, and an aborted upload cannot be abandoned once in the cache, so its partial data is written back to the remote as if it were a completed object.

Remotes that benefit from --vfs-cache-mode writes:

  • Remotes over slow or unreliable links. A failure in a streamed upload aborts the whole multipart upload and the client must start again from the first part; a failed write-back upload is retried by the VFS (see --vfs-cache-max-age and friends) without the client being involved. Ingest from the client also runs at local disk speed rather than being throttled to the remote's pace.
  • Workloads that read back or overwrite what they just wrote. The completed object stays in the cache, so subsequent GET/HEAD requests are served locally, and plain PUTs and multipart uploads to the same key go through the same cache entry so the last write wins regardless of upload style.

The trade-offs of the VFS cache:

  • The whole object lands on local disk, so the cache (--cache-dir) needs space for the largest objects in flight; --vfs-cache-max-size cannot evict files which are still being uploaded.
  • The 200 OK for CompleteMultipartUpload means the data is safely in the local cache, not yet on the remote - the same durability the cache gives plain PUTs. If an acknowledgement must mean the data has reached the remote (for example WAL archiving), use the default --vfs-cache-mode off.
  • The upload to the remote only starts on completion, rather than overlapping with the parts arriving, so the data reaches the remote later than with streaming.
  • If serve s3 is killed part-way through an upload, the temporary file survives in the cache and the VFS cache recovery uploads it to the remote on restart as a temporary object (named with a leading .rclone_temp_multipart_); as with the streaming path, it is hidden from S3 listings but must be removed manually.

Cleaning up temporary objects

If serve s3 is killed part-way through an upload it can leave a temporary object behind, named with a leading .rclone_temp_. This whole prefix is reserved: any object whose name (the last /-separated segment of its key) starts with .rclone_temp_ is hidden from S3 listings, so don't give real objects such names - an existing object with such a name disappears from listings (though it stays accessible directly by its key: only listings hide reserved names, GET, HEAD and DELETE of the exact key still work). A temporary object never holds acknowledged data - uploads whose temporary object survived were never confirmed to the client - so old ones are safe to delete:

rclone delete --min-age 24h --include ".rclone_temp_*" remote:path

The --min-age protects uploads which are still in progress: make sure it is longer than your longest upload, especially if several serve s3 instances share the same remote.

rclone v1.75 named its temporary multipart objects .rclone_multipart_upload_*; leftovers from an older server are also hidden from listings and can be cleaned up the same way.

Abandoned uploads

A client which starts a multipart upload and vanishes without either completing or aborting it would otherwise hold on to its resources forever.

An incomplete multipart upload which has had no activity for --multipart-expiry (default 24h) is therefore aborted and cleaned up, exactly as if the client had called AbortMultipartUpload, and a NOTICE is logged.

An upload with a part still being received is never expired, however slowly the part is arriving, and each completed part restarts the clock, so the expiry only needs to outlast the client's pauses between parts, not the whole upload.

Late operations on an expired upload fail with NoSuchUpload, as they do on real S3 when a lifecycle rule has aborted the upload. Set --multipart-expiry 0 to keep incomplete uploads forever - but note that abandoned uploads are then never cleaned up and go on counting towards --multipart-max-uploads below, so enough of them will stop the user starting any new multipart upload until the server is restarted.

Each multipart upload in progress holds server resources - for example the buffers the remote's own upload uses once the first part has arrived - so each user (each access key) can have no more than --multipart-max-uploads (default 1000, 0 for no limit) in progress at once. Starting another fails with a SlowDown error, which S3 clients retry, and a one-off NOTICE is logged.

Disabling streaming

If you pass --disable-multipart-streaming, multipart uploads are instead buffered in memory by the underlying S3 library: every part is held in memory and the whole object is written out in one go when the upload completes. This removes the in-order/contiguous-part restriction above, so parts can be uploaded in any order, but memory use grows with the size of the upload, so it is only suitable for small objects. A one-off NOTICE is logged the first time this happens. This flag is the only thing that makes multipart uploads buffer in memory - it is never done because of missing remote capabilities. The limits on multipart uploads above (--multipart-streaming-buffer-limit, --multipart-streaming-buffer-total and --multipart-max-uploads) do not apply to uploads buffered in memory, so don't use this flag if the server is open to clients you don't trust. Consider --vfs-cache-mode writes instead, which buffers the upload in the VFS cache on disk and takes precedence over --disable-multipart-streaming.

Memory use

serve s3 never buffers whole objects in memory (except with --disable-multipart-streaming, above), and limits the memory each kind of request can use, so that a client, even a hostile one, can't make it use unbounded memory.

  • Object data. PutObject bodies and multipart parts arriving in order are streamed straight through to the remote. Parts arriving out of order are buffered, up to --multipart-streaming-buffer-limit per upload and --multipart-streaming-buffer-total per user.
  • Uploads in progress. Each user can have at most --multipart-max-uploads multipart uploads in progress. An upload holds little memory until its first part arrives. After that, with --vfs-cache-mode off, it holds a --streaming-upload-cutoff sized buffer plus whatever the remote's own upload buffers, which for many remotes is up to their chunk size times their upload concurrency (for example --s3-chunk-size times --s3-upload-concurrency). Plain PutObject uploads use the same while they are in progress.
  • Metadata. The metadata of at most --metadata-max-objects objects of each user is kept, each limited to 2 KB.
  • Request bodies. The XML bodies of requests such as CompleteMultipartUpload and DeleteObjects are limited to 10 MiB, and DeleteObjects to 1000 keys. A browser form (POST) upload holds at most 1 MiB in memory, spooling the rest to a temporary file, and is limited to 5 GB.
  • Listings. Each page of a ListObjects listing reads only the directories needed to fill it, and returns at most 1000 keys. Directory listings are cached by the VFS directory cache.

The largest of these is usually the remote's upload buffers, which can be up to --multipart-max-uploads times the buffers of one upload for each user if clients start many uploads and send part of each. The limits apply to each user separately, so that one user can't stop others uploading, which means the memory use grows with the number of users. To limit it, lower --multipart-max-uploads, the remote's chunk size or upload concurrency, or set --max-buffer-memory to limit the memory used by the buffers of all uploads to remotes which use rclone's buffer pool (including s3, azureblob and b2). Buffered multipart parts come from the same pool, so if you set --max-buffer-memory make it comfortably bigger than --multipart-streaming-buffer-total times the number of users uploading at once, otherwise buffered parts can use up all the memory the remote needs to upload the parts which would free them.

The number of connections is not limited, so if the server is open to clients you don't trust, use --auth-key or --auth-proxy and consider putting it behind a reverse proxy which limits connections.

Design goals

These goals explain many of the limitations below.

  • Objects are plain files on the remote. A bucket is a directory in the root and a key is a path in it, so the remote stays usable by other tools, such as rclone mount or rclone sync. So keys must be paths (see Object names).
  • No database. serve s3 stores nothing but the objects. What the remote can't hold, such as user metadata, tags, the MD5 ETag of an object on a remote without MD5 and the directory markers put, is kept in memory as a best effort, and lost on restart.
  • Works with any rclone remote. What serve s3 can do depends on what the remote supports, such as the hashes used for ETags and modification times, rather than needing a particular backend.
  • Bounded memory. Uploads, including multipart uploads, are streamed to the remote and the memory each request can use is limited (see Memory use). Nothing is written to local disk unless the VFS cache is used or the remote can't stream uploads (see Multipart uploads).
  • Refuse rather than ignore. Unsupported features are refused, or answered as S3 answers when the feature isn't configured, so a client is never told something was done which wasn't (see Compatibility).
  • A storage gateway, not a full S3. serve s3 implements the parts of the S3 API clients use to store and fetch objects. Versioning, ACLs, bucket policies, lifecycle rules and the like are out of scope.
  • Compatible with rclone's s3 backend. This includes its directory markers, which the tests check.

Bugs

For a current list of serve s3 bugs see the serve s3 bug category on GitHub.

Limitations

serve s3 will treat all directories in the root as buckets and ignore all files in the root. You can use CreateBucket to create folders under the root, and put a directory marker to create an empty folder below them (see Object names).

When using PutObject or DeleteObject, rclone will automatically create or clean up empty folders. If you don't want to clean up empty folders automatically, use --directory-markers (see Object names). The deprecated --no-cleanup does the same.

When using ListObjects with the delimiter /, rclone lists the directory the prefix is in, showing its subdirectories as common prefixes, including empty ones. Without a delimiter the listing is recursive: rclone walks the directory tree below the prefix, reading only as many directories as it needs to fill each page, and lists empty directories as their directory markers (see below). Any other delimiter is supported too, but needs the whole directory tree below the prefix walking, unless the delimiter is in the directory names.

Versioning is not supported (see Design goals).

Conditional writes, PutObject and CompleteMultipartUpload with If-None-Match or If-Match, are supported. The conditions are checked before the object is written, and again as it is moved into place when it is written to a temporary object first. This happens for multipart uploads and, for PutObject, with remotes which can upload partial files (such as local disk) or with --vfs-cache-mode minimal or above. The second check is atomic against other writes through serve s3, but neither is atomic against changes made to the remote some other way. If the remote doesn't support the --etag-hash, objects have no ETag so If-Match with an ETag always fails.

Metadata will only be saved in memory (see Design goals) other than the rclone mtime metadata which will be set as the modification time of the file. The metadata of at most --metadata-max-objects objects (default 100000, 0 for no limit) of each user (each access key) is kept; after that the metadata of the user's least recently used object is forgotten. The metadata of an object is ignored once the object has been changed some other way, for example by another --auth-proxy user or directly on the backend. Object tags are kept with the metadata in the same way.

Object names

serve s3 stores objects as files in the backend, so object keys are mapped to file paths rather than treated as the opaque strings AWS S3 allows. Keys must be in canonical path form: keys that contain .. or . path segments, repeated slashes (//), or a leading slash are rejected with a 400 Bad Request (InvalidArgument) instead of being normalised, since normalising them could alias two distinct keys to the same file or resolve a key outside its bucket. This matches the behaviour of other S3 servers such as MinIO.

A key ending in /, such as dir/, is a directory marker, standing for the directory dir. Putting an empty directory marker creates the directory (one with data is rejected with 400 Bad Request), and deleting it removes the directory if it is empty. Directory markers are found by HeadObject, GetObject and CopyObject for any directory, but only the markers of empty directories, and of the directories a marker was put for, are listed as objects, as other directories may not have been made by putting a marker. The markers put are remembered in memory with the metadata (see Limitations), so after a restart, or once forgotten, the marker of a directory holding other keys is no longer listed. While a marker is remembered its directory isn't removed when the last key below it is deleted.

With --directory-markers every directory is listed as a directory marker, so the markers are listed after a restart too, as needed by clients which rely on them, such as rclone's s3 backend with --s3-directory-markers. As every directory has a marker, a directory isn't removed when the last key below it is deleted. Deleting the marker of a directory which holds other keys stops it being listed, and the directory is then removed with the last key below it, but this is only remembered in memory, like the markers put without --directory-markers.

As keys are file paths, a key can't be both an object and a directory holding other objects, for example a/b and a/b/c. Writing an object which would need this fails with 400 Bad Request (InvalidRequest).

Compatibility

serve s3 implements the parts of the S3 API that S3 clients commonly use to store and fetch objects. Features it doesn't support are refused with 501 NotImplemented, or answered as AWS S3 answers for a bucket without that feature configured, rather than being silently ignored.

Feature Supported Notes
ListBuckets, CreateBucket, DeleteBucket, HeadBucket, GetBucketLocation Yes Buckets are the directories in the root of the remote.
ListObjects, ListObjectsV2 Yes Prefixes, delimiters, markers, start-after, continuation tokens, fetch-owner and encoding-type=url all work. Delimiters other than / need the whole directory tree walking.
HeadObject, GetObject Yes Including Range, partNumber, the response-* header overrides and the If-Match, If-None-Match, If-Modified-Since and If-Unmodified-Since conditions.
PutObject Yes Including Content-MD5, aws-chunked uploads and uploads with no Content-Length.
CopyObject Yes Including the metadata and tagging directives and the x-amz-copy-source-if-* conditions.
DeleteObject, DeleteObjects Yes
Multipart uploads Yes CreateMultipartUpload, UploadPart, UploadPartCopy, CompleteMultipartUpload, AbortMultipartUpload, ListMultipartUploads and ListParts. The ETag of the object is its hash, not the -N suffixed one AWS gives. The 5 MiB minimum part size isn't enforced.
Conditional writes Partial If-Match and If-None-Match on PutObject and CompleteMultipartUpload are checked, but not atomically against changes made to the remote some other way (see Limitations).
Authentication Partial AWS Signature Version 4, in headers, presigned URLs and browser form (POST) uploads. Signature Version 2 isn't supported.
Anonymous access Partial All requests are allowed without --auth-key or --auth-proxy, and none are with them, as buckets and objects can't be made public.
Metadata Partial x-amz-meta-*, the storage class and the standard headers such as Content-Type are kept in memory only (see Limitations), except the modification time. The storage class is only recorded, so doesn't change how the object is stored.
Directory markers Partial Keys ending in / are stored as directories, so can't hold data. A marker put for a directory holding other keys is only listed while it is remembered in memory, unless --directory-markers is set (see Object names).
Checksums Partial The x-amz-checksum-* of a PutObject (CRC32, CRC32C, CRC64NVME, SHA1 or SHA256) is checked and returned with the metadata. The Content-MD5 and x-amz-checksum-* of the bodies of other requests, such as DeleteObjects, are checked. The checksums of multipart uploads aren't supported.
GetObjectAttributes Partial Everything but the checksums of multipart uploads.
Object tagging Partial GetObjectTagging, PutObjectTagging and DeleteObjectTagging work, but tags are kept in memory with the metadata.
Bucket owner No serve s3 has no AWS account IDs, so x-amz-expected-bucket-owner and x-amz-source-expected-bucket-owner are ignored.
ACLs Partial GetBucketAcl and GetObjectAcl give the owner full control. Only the private ACL can be set.
CORS Partial Preflight (OPTIONS) requests are answered, allowing any origin. CORS configuration (PutBucketCors) isn't supported.
Server side encryption No SSE-S3 (AES256) is accepted but the data isn't encrypted. SSE-KMS and SSE-C are refused.
Versioning No Buckets are reported as unversioned and ListObjectVersions lists each object as version null. Enabling versioning is refused.
Object lock, retention and legal hold No
Bucket configuration No Bucket policies, tagging, lifecycle, logging, encryption, website, replication, notifications and the like. Getting them fails as for a bucket without them configured.
SelectObjectContent, RestoreObject, torrents No

The ceph s3-tests can be run against serve s3 to check its compatibility. See cmd/serve/s3/s3tests for how.

Authentication

By default this will serve files without needing a login.

You can either use an htpasswd file which can take lots of users, or set a single username and password with the --user and --pass flags.

Alternatively, you can have the reverse proxy manage authentication and use the username provided in the configured header with --user-from-header (e.g., --user-from-header=x-remote-user). Ensure the proxy is trusted and headers cannot be spoofed, as misconfiguration may lead to unauthorized access.

If either of the above authentication methods is not configured and client certificates are required by the --client-ca flag passed to the server, the client certificate common name will be considered as the username.

Use --htpasswd /path/to/htpasswd to provide an htpasswd file. This is in standard apache format and supports MD5, SHA1 and BCrypt for basic authentication. Bcrypt is recommended.

To create an htpasswd file:

touch htpasswd
htpasswd -B htpasswd user
htpasswd -B htpasswd anotherUser

Use the same hash format and cost for every entry in the file. If the entries differ, the time taken to reject a wrong password can reveal which usernames exist.

The password file can be updated while rclone is running.

When authentication is enabled every request must be authenticated, including HTTP OPTIONS requests, as the answer to an OPTIONS request can reveal whether a path exists. Health checks or monitoring which use unauthenticated OPTIONS requests will need to send credentials. The exception is browser CORS preflight requests when --allow-origin is set, which are answered without authentication as browsers never send credentials with them.

Use --realm to set the authentication realm.

Use --salt to change the password hashing salt from the default.

Server options

Use --addr to specify which IP address and port the server should listen on, eg --addr 1.2.3.4:8000 or --addr :8080 to listen to all IPs. By default it only listens on localhost. You can use port :0 to let the OS choose an available port.

If you set --addr to listen on a public or LAN accessible IP address then using Authentication is advised - see the next section for info.

You can use a unix socket by setting the url to unix:///path/to/socket or just by using an absolute path name.

--addr may be repeated to listen on multiple IPs/ports/sockets. Socket activation, described further below, can also be used to accomplish the same.

--server-read-timeout and --server-write-timeout can be used to control the timeouts on the server. Note that this is the total time for a transfer.

--max-header-bytes controls the maximum number of bytes the server will accept in the HTTP header.

--response-header can be used to set an HTTP header for all responses, will overriding existing values. The flag may be repeated to add multiple headers. Use the format Header-Name: value.

--allow-origin enables cross-origin (CORS) requests from browsers by setting the Access-Control-Allow-Origin header on every response to the origin given, e.g. --allow-origin https://app.example.com. Browser CORS preflight OPTIONS requests are answered by rclone without needing authentication as browsers never send credentials with them. Setting this to * allows any origin which is a security risk on an authenticated server.

--baseurl controls the URL prefix that rclone serves from. By default rclone will serve from the root. If you used --baseurl "/rclone" then rclone would serve from a URL starting with "/rclone/". This is useful if you wish to proxy rclone serve. Rclone automatically inserts leading and trailing "/" on --baseurl, so --baseurl "rclone", --baseurl "/rclone" and --baseurl "/rclone/" are all treated identically.

--disable-zip may be set to disable the zipping download option.

Protocol

The server supports HTTP/1.1 and HTTP/2. HTTP/2 is used automatically for TLS connections. For non-TLS connections, HTTP/2 cleartext (h2c) is supported, allowing HTTP/2 without encryption.

TLS (SSL)

By default this will serve over http. If you want you can serve over https. You will need to supply the --cert and --key flags. If you wish to do client side certificate validation then you will need to supply --client-ca also.

When TLS is configured every listener given with --addr serves TLS. An individual listener can be prefixed with http:// to serve unencrypted HTTP on that address, or with tls:// to state explicitly that it must serve TLS. Using a tls:// address without --cert and --key is an error.

--cert must be set to the path of a file containing either a PEM encoded certificate, or a concatenation of that with the CA certificate. --key must be set to the path of a file with the PEM encoded private key. If setting --client-ca, it should be set to the path of a file with PEM encoded client certificate authority certificates.

--min-tls-version is minimum TLS version that is acceptable. Valid values are "tls1.0", "tls1.1", "tls1.2" and "tls1.3" (default "tls1.0").

Socket activation

Instead of the listening addresses specified above, rclone will listen to all FDs passed by the service manager, if any (and ignore any arguments passed by --addr).

This allows rclone to be a socket-activated service. It can be configured with .socket and .service unit files as described in https://www.freedesktop.org/software/systemd/man/latest/systemd.socket.html.

Socket activation can be tested ad-hoc with the systemd-socket-activatecommand

systemd-socket-activate -l 8000 -- rclone serve

This will socket-activate rclone on the first connection to port 8000 over TCP.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

Auth Proxy

If you supply the parameter --auth-proxy /path/to/program then rclone will use that program to generate backends on the fly which then are used to authenticate incoming requests. This uses a simple JSON based protocol with input on STDIN and output on STDOUT.

PLEASE NOTE: --auth-proxy and --authorized-keys cannot be used together, if --auth-proxy is set the authorized keys option will be ignored.

There is an example program bin/test_proxy.py in the rclone source code.

The program's job is to take a user and pass on the input and turn those into the config for a backend on STDOUT in JSON format. This config will have any default parameters for the backend added, but it won't use configuration from environment variables or command line options - it is the job of the proxy program to make a complete config.

This config generated must have this extra parameter

  • _root - root to use for the backend

And it may have these parameters

  • _obscure - comma separated strings for parameters to obscure
  • _secret_access_key - the secret for S3 access key auth (see below)

If password authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "pass": "mypassword",
  "client_ip": "192.168.1.1"
}

If public-key authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "public_key": "AAAAB3NzaC1yc2EAAAADAQABAAABAQDuwESFdAe14hVS6omeyX7edc...JQdf",
  "client_ip": "192.168.1.1"
}

If the client authenticated with an S3 access key (rclone serve s3), the client never sends its secret, only a signature made with it, so the input contains just the access key ID as the user with no pass or public_key:

{
  "user": "AKIAIOSFODNN7EXAMPLE",
  "client_ip": "192.168.1.1"
}

In this case the program must look up the secret access key for that access key ID and return it in the _secret_access_key field of the output. Rclone then uses that secret to verify the signature on the request, refusing the request if it does not match. This means the proxy program is the source of truth for both the credentials and the backend they map to. If the program does not return _secret_access_key or returns it empty the request is refused.

The program's answer for an access key ID is cached (see below) but is checked with the program again after 5 minutes even if the access key ID is in constant use, so revoking an access key ID in the program takes effect within 5 minutes. A rotated secret takes effect on the first request signed with it.

The client_ip key holds the IP address the client connected from, without a port number. It can be used to restrict logins to certain networks, or to log authentication attempts centrally. It is omitted if the client has no IP address, for example when connecting over a unix socket. Note that if rclone is behind a reverse proxy this will be the address of the reverse proxy and not the original client.

And as an example return this on STDOUT

{
  "type": "sftp",
  "_root": "",
  "_obscure": "pass",
  "user": "me",
  "pass": "mypassword",
  "host": "sftp.example.com"
}

This would mean that an SFTP backend would be created on the fly for the user and pass/public_key returned in the output to the host given. Note that since _obscure is set to pass, rclone will obscure the pass parameter before creating the backend (which is required for sftp backends).

The program can manipulate the supplied user in any way, for example to make proxy to many different sftp backends, you could make the user be [email protected] and then set the host to example.com in the output and the user to user. For security you'd probably want to restrict the host to a limited list.

An internal cache of backends is keyed on the user, the kind of credential presented, a hash of the pass or public_key, and the client_ip. This means that if a user's password or public-key changes, the client connects from a new IP address, or the proxy returns different config parameters (eg a rotated api_key), a fresh backend will be created on the next request rather than the cached one being reused. A backend created for a public key login is never reused for a password login, so a public key cannot be sent as a password to skip the program.

A cached answer is checked with the program again after 5 minutes even if the user is in constant use, so revoking a user in the program takes effect within 5 minutes. For serve sftp this applies to new logins - an existing connection is not closed.

This can be used to build general purpose proxies to any kind of backend that rclone supports.

rclone serve s3 remote:path [flags]

Options

      --addr stringArray                              IPaddress:Port or :Port to bind server to (default 127.0.0.1:8080)
      --allow-origin string                           Origin which cross-domain request (CORS) can be executed from
      --auth-key stringArray                          Set key pair for v4 authorization: access_key_id,secret_access_key
      --auth-proxy string                             A program to use to create the backend from the auth
      --baseurl string                                Prefix for URLs - leave blank for root
      --cert string                                   TLS PEM key (concatenation of certificate and CA certificate)
      --client-ca string                              Client certificate authority to verify clients with
      --dir-cache-time Duration                       Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                            Directory permissions (default 777)
      --directory-markers                             List every directory as a directory marker object
      --disable-multipart-streaming                   Buffer multipart uploads in memory instead of streaming them to the backend
      --etag-hash string                              Which hash to use for the ETag, or auto or blank for off (default "MD5")
      --file-perms FileMode                           File permissions (default 666)
      --force-path-style                              If true use path style access if false use virtual hosted style (default true)
      --gid uint32                                    Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                          help for s3
      --htpasswd string                               A htpasswd file - if not provided no authentication is done
      --key string                                    TLS PEM Private key
      --link-perms FileMode                           Link permissions (default 666)
      --max-header-bytes int                          Maximum size of request header (default 4096)
      --metadata-max-objects int                      Maximum number of objects of a user whose metadata is kept in memory, 0 for unlimited (default 100000)
      --min-tls-version string                        Minimum TLS version that is acceptable (default "tls1.0")
      --multipart-expiry Duration                     Abort incomplete multipart uploads idle for longer than this, 0 to keep forever (default 1d)
      --multipart-max-uploads int                     Maximum number of streamed multipart uploads a user can have in progress at once, 0 for unlimited (default 1000)
      --multipart-streaming-buffer-limit SizeSuffix   Maximum memory buffered per streamed multipart upload for parts arriving out of order, 0 for unlimited (default 256Mi)
      --multipart-streaming-buffer-total SizeSuffix   Maximum memory buffered by all the streamed multipart uploads of a user for parts arriving out of order, 0 for unlimited (default 1Gi)
      --no-checksum                                   Don't compare checksums on up/download
      --no-cleanup                                    Deprecated: use --directory-markers instead
      --no-modtime                                    Don't read/write the modification time (can speed things up)
      --no-seek                                       Don't allow seeking in files
      --pass string                                   Password for authentication
      --poll-interval Duration                        Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --read-only                                     Only allow read-only access
      --realm string                                  Realm for authentication
      --response-header stringArray                   Set HTTP header for all responses, overriding existing values
      --salt string                                   Password hashing salt (default "dlPL2MqE")
      --server-read-timeout Duration                  Timeout for server reading data (default 1h0m0s)
      --server-write-timeout Duration                 Timeout for server writing data (default 1h0m0s)
      --uid uint32                                    Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                                Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --user string                                   User name for authentication
      --user-from-header string                       User name from a defined HTTP header
      --vfs-block-norm-dupes                          If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration                    Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix                 Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix           Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode                      Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration              Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                          If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix          Specify the total space of disk (default off)
      --vfs-fast-fingerprint                          Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration                   Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                                     Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string                 Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix                     Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix                Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix          If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int                    The number of parallel streams to read at once
      --vfs-read-wait Duration                        Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                                   Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size                  Use the rclone size algorithm for Used size
      --vfs-write-back Duration                       Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                       Time to wait for in-sequence write before giving error (default 1s)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

rclone serve sftp

Serve the remote over SFTP.

Synopsis

Run an SFTP server to serve a remote over SFTP. This can be used with an SFTP client or you can make a remote of type sftp to use with it.

You can use the filter flags (e.g. --include, --exclude) to control what is served.

The server will respond to a small number of shell commands, mainly md5sum, sha1sum and df, which enable it to provide support for checksums and the about feature when accessed from an sftp remote.

Note that this server uses standard 32 KiB packet payload size, which means you must not configure the client to expect anything else, e.g. with the chunk_size option on an sftp remote.

The server will log errors. Use -v to see access logs.

--bwlimit will be respected for file transfers. Use --stats to control the stats printing.

You must provide some means of authentication, either with --user/--pass, an authorized keys file (specify location with --authorized-keys - the default is the same as ssh), an --auth-proxy, or set the --no-auth flag for no authentication when logging in.

If you don't supply a host --key then rclone will generate rsa, ecdsa and ed25519 variants, and cache them for later use in rclone's cache directory (see rclone help flags cache-dir) in the "serve-sftp" directory.

By default the server binds to localhost:2022 - if you want it to be reachable externally then supply --addr :2022 for example.

This also supports being run with socket activation, in which case it will listen on the first passed FD. It can be configured with .socket and .service unit files as described in https://www.freedesktop.org/software/systemd/man/latest/systemd.socket.html.

Socket activation can be tested ad-hoc with the systemd-socket-activatecommand:

systemd-socket-activate -l 2222 -- rclone serve sftp :local:vfs/

This will socket-activate rclone on the first connection to port 2222 over TCP.

Note that the default of --vfs-cache-mode off is fine for the rclone sftp backend, but it may not be with other SFTP clients.

If --stdio is specified, rclone will serve SFTP over stdio, which can be used with sshd via ~/.ssh/authorized_keys, for example:

restrict,command="rclone serve sftp --stdio ./photos" ssh-rsa ...

On the client you need to set --transfers 1 when using --stdio. Otherwise multiple instances of the rclone server are started by OpenSSH which can lead to "corrupted on transfer" errors. This is the case because the client chooses indiscriminately which server to send commands to while the servers all have different views of the state of the filing system.

The "restrict" in authorized_keys prevents SHA1SUMs and MD5SUMs from being used. Omitting "restrict" and using --sftp-path-override to enable checksumming is possible but less secure and you could use the SFTP server provided by OpenSSH in this case.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

Auth Proxy

If you supply the parameter --auth-proxy /path/to/program then rclone will use that program to generate backends on the fly which then are used to authenticate incoming requests. This uses a simple JSON based protocol with input on STDIN and output on STDOUT.

PLEASE NOTE: --auth-proxy and --authorized-keys cannot be used together, if --auth-proxy is set the authorized keys option will be ignored.

There is an example program bin/test_proxy.py in the rclone source code.

The program's job is to take a user and pass on the input and turn those into the config for a backend on STDOUT in JSON format. This config will have any default parameters for the backend added, but it won't use configuration from environment variables or command line options - it is the job of the proxy program to make a complete config.

This config generated must have this extra parameter

  • _root - root to use for the backend

And it may have these parameters

  • _obscure - comma separated strings for parameters to obscure
  • _secret_access_key - the secret for S3 access key auth (see below)

If password authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "pass": "mypassword",
  "client_ip": "192.168.1.1"
}

If public-key authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "public_key": "AAAAB3NzaC1yc2EAAAADAQABAAABAQDuwESFdAe14hVS6omeyX7edc...JQdf",
  "client_ip": "192.168.1.1"
}

If the client authenticated with an S3 access key (rclone serve s3), the client never sends its secret, only a signature made with it, so the input contains just the access key ID as the user with no pass or public_key:

{
  "user": "AKIAIOSFODNN7EXAMPLE",
  "client_ip": "192.168.1.1"
}

In this case the program must look up the secret access key for that access key ID and return it in the _secret_access_key field of the output. Rclone then uses that secret to verify the signature on the request, refusing the request if it does not match. This means the proxy program is the source of truth for both the credentials and the backend they map to. If the program does not return _secret_access_key or returns it empty the request is refused.

The program's answer for an access key ID is cached (see below) but is checked with the program again after 5 minutes even if the access key ID is in constant use, so revoking an access key ID in the program takes effect within 5 minutes. A rotated secret takes effect on the first request signed with it.

The client_ip key holds the IP address the client connected from, without a port number. It can be used to restrict logins to certain networks, or to log authentication attempts centrally. It is omitted if the client has no IP address, for example when connecting over a unix socket. Note that if rclone is behind a reverse proxy this will be the address of the reverse proxy and not the original client.

And as an example return this on STDOUT

{
  "type": "sftp",
  "_root": "",
  "_obscure": "pass",
  "user": "me",
  "pass": "mypassword",
  "host": "sftp.example.com"
}

This would mean that an SFTP backend would be created on the fly for the user and pass/public_key returned in the output to the host given. Note that since _obscure is set to pass, rclone will obscure the pass parameter before creating the backend (which is required for sftp backends).

The program can manipulate the supplied user in any way, for example to make proxy to many different sftp backends, you could make the user be [email protected] and then set the host to example.com in the output and the user to user. For security you'd probably want to restrict the host to a limited list.

An internal cache of backends is keyed on the user, the kind of credential presented, a hash of the pass or public_key, and the client_ip. This means that if a user's password or public-key changes, the client connects from a new IP address, or the proxy returns different config parameters (eg a rotated api_key), a fresh backend will be created on the next request rather than the cached one being reused. A backend created for a public key login is never reused for a password login, so a public key cannot be sent as a password to skip the program.

A cached answer is checked with the program again after 5 minutes even if the user is in constant use, so revoking a user in the program takes effect within 5 minutes. For serve sftp this applies to new logins - an existing connection is not closed.

This can be used to build general purpose proxies to any kind of backend that rclone supports.

rclone serve sftp remote:path [flags]

Options

      --addr string                            IPaddress:Port or :Port to bind server to (default "localhost:2022")
      --auth-proxy string                      A program to use to create the backend from the auth
      --authorized-keys string                 Authorized keys file (default "~/.ssh/authorized_keys")
      --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                     Directory permissions (default 777)
      --file-perms FileMode                    File permissions (default 666)
      --gid uint32                             Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                   help for sftp
      --key stringArray                        SSH private host key file (Can be multi-valued, leave blank to auto generate)
      --link-perms FileMode                    Link permissions (default 666)
      --no-auth                                Allow connections with no authentication if set
      --no-checksum                            Don't compare checksums on up/download
      --no-modtime                             Don't read/write the modification time (can speed things up)
      --no-seek                                Don't allow seeking in files
      --pass string                            Password for authentication
      --poll-interval Duration                 Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --read-only                              Only allow read-only access
      --stdio                                  Run an sftp server on stdin/stdout
      --uid uint32                             Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                         Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --user string                            User name for authentication
      --vfs-block-norm-dupes                   If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration             Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix          Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix    Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode               Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration       Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                   If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix   Specify the total space of disk (default off)
      --vfs-fast-fingerprint                   Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration            Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                              Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string          Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix              Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix         Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix   If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int             The number of parallel streams to read at once
      --vfs-read-wait Duration                 Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                            Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size           Use the rclone size algorithm for Used size
      --vfs-write-back Duration                Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                Time to wait for in-sequence write before giving error (default 1s)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

rclone serve webdav

Serve remote:path over WebDAV.

Synopsis

Run a basic WebDAV server to serve a remote over HTTP via the WebDAV protocol. This can be viewed with a WebDAV client, through a web browser, or you can make a remote of type WebDAV to read and write it.

WebDAV options

--etag-hash

This controls the ETag header. Without this flag the ETag will be based on the ModTime and Size of the object.

If this flag is set to "auto" then rclone will choose the first supported hash on the backend or you can use a named hash such as "MD5" or "SHA-1". Use the hashsum command to see the full list.

Gzip compression

The server will compress certain response bodies (text and XML, including WebDAV PROPFIND responses) using gzip when the client advertises gzip support via the Accept-Encoding: gzip request header. This reduces bandwidth usage.

Access WebDAV on Windows

WebDAV shared folder can be mapped as a drive on Windows, however the default settings prevent it. Windows will fail to connect to the server using insecure Basic authentication. It will not even display any login dialog. Windows requires SSL / HTTPS connection to be used with Basic. If you try to connect via Add Network Location Wizard you will get the following error: "The folder you entered does not appear to be valid. Please choose another". However, you still can connect if you set the following registry key on a client machine: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\WebClient\Parameters\BasicAuthLevel to 2. The BasicAuthLevel can be set to the following values:

0 - Basic authentication disabled
1 - Basic authentication enabled for SSL connections only
2 - Basic authentication enabled for SSL connections and for non-SSL connections

If required, increase the FileSizeLimitInBytes to a higher value. Navigate to the Services interface, then restart the WebClient service.

Access Office applications on WebDAV

Navigate to following registry HKEY_CURRENT_USER\Software\Microsoft\Office\[14.0/15.0/16.0]\Common\Internet Create a new DWORD BasicAuthLevel with value 2.

0 - Basic authentication disabled
1 - Basic authentication enabled for SSL connections only
2 - Basic authentication enabled for SSL and for non-SSL connections

https://learn.microsoft.com/en-us/office/troubleshoot/powerpoint/office-opens-blank-from-sharepoint

Serving over a unix socket

You can serve the webdav on a unix socket like this:

rclone serve webdav --addr unix:///tmp/my.socket remote:path

and connect to it like this using rclone and the webdav backend:

rclone --webdav-unix-socket /tmp/my.socket --webdav-url http://localhost lsf :webdav:

Note that there is no authentication on http protocol - this is expected to be done by the permissions on the socket.

Symlinks / Junction points

The webdav protocol does not support symlinks or junction points and by default rclone will skip them completely.

You can use -L to get rclone to follow symlinks or you can use --local-links to make rclone show .rclonelink files in place of the symlinks.

NB Do not use --links as since v1.69 this applies to the VFS layer too, use --local-links which only applies to the local backend only.

Server options

Use --addr to specify which IP address and port the server should listen on, eg --addr 1.2.3.4:8000 or --addr :8080 to listen to all IPs. By default it only listens on localhost. You can use port :0 to let the OS choose an available port.

If you set --addr to listen on a public or LAN accessible IP address then using Authentication is advised - see the next section for info.

You can use a unix socket by setting the url to unix:///path/to/socket or just by using an absolute path name.

--addr may be repeated to listen on multiple IPs/ports/sockets. Socket activation, described further below, can also be used to accomplish the same.

--server-read-timeout and --server-write-timeout can be used to control the timeouts on the server. Note that this is the total time for a transfer.

--max-header-bytes controls the maximum number of bytes the server will accept in the HTTP header.

--response-header can be used to set an HTTP header for all responses, will overriding existing values. The flag may be repeated to add multiple headers. Use the format Header-Name: value.

--allow-origin enables cross-origin (CORS) requests from browsers by setting the Access-Control-Allow-Origin header on every response to the origin given, e.g. --allow-origin https://app.example.com. Browser CORS preflight OPTIONS requests are answered by rclone without needing authentication as browsers never send credentials with them. Setting this to * allows any origin which is a security risk on an authenticated server.

--baseurl controls the URL prefix that rclone serves from. By default rclone will serve from the root. If you used --baseurl "/rclone" then rclone would serve from a URL starting with "/rclone/". This is useful if you wish to proxy rclone serve. Rclone automatically inserts leading and trailing "/" on --baseurl, so --baseurl "rclone", --baseurl "/rclone" and --baseurl "/rclone/" are all treated identically.

--disable-zip may be set to disable the zipping download option.

Protocol

The server supports HTTP/1.1 and HTTP/2. HTTP/2 is used automatically for TLS connections. For non-TLS connections, HTTP/2 cleartext (h2c) is supported, allowing HTTP/2 without encryption.

TLS (SSL)

By default this will serve over http. If you want you can serve over https. You will need to supply the --cert and --key flags. If you wish to do client side certificate validation then you will need to supply --client-ca also.

When TLS is configured every listener given with --addr serves TLS. An individual listener can be prefixed with http:// to serve unencrypted HTTP on that address, or with tls:// to state explicitly that it must serve TLS. Using a tls:// address without --cert and --key is an error.

--cert must be set to the path of a file containing either a PEM encoded certificate, or a concatenation of that with the CA certificate. --key must be set to the path of a file with the PEM encoded private key. If setting --client-ca, it should be set to the path of a file with PEM encoded client certificate authority certificates.

--min-tls-version is minimum TLS version that is acceptable. Valid values are "tls1.0", "tls1.1", "tls1.2" and "tls1.3" (default "tls1.0").

Socket activation

Instead of the listening addresses specified above, rclone will listen to all FDs passed by the service manager, if any (and ignore any arguments passed by --addr).

This allows rclone to be a socket-activated service. It can be configured with .socket and .service unit files as described in https://www.freedesktop.org/software/systemd/man/latest/systemd.socket.html.

Socket activation can be tested ad-hoc with the systemd-socket-activatecommand

systemd-socket-activate -l 8000 -- rclone serve

This will socket-activate rclone on the first connection to port 8000 over TCP.

Template

--template allows a user to specify a custom markup template for HTTP and WebDAV serve functions. The server exports the following markup to be used within the template to server pages:

Parameter Subparameter Description
.Name The full path of a file/directory.
.Title Directory listing of '.Name'.
.Sort The current sort used. This is changeable via '?sort=' parameter. Possible values: namedirfirst, name, size, time (default namedirfirst).
.Order The current ordering used. This is changeable via '?order=' parameter. Possible values: asc, desc (default asc).
.Query Currently unused.
.Path The path of the directory from the root with leading and trailing slashes, e.g. '/dir/'.
.IsRoot Boolean for if this is the root directory.
.UpLink The link to the parent directory.
.Static Boolean for if the listing is being written as a static page by rclone index.
.NumDirs The number of directories in the listing.
.NumFiles The number of files in the listing.
.TotalSize The total size in bytes of the files in the listing.
.Breadcrumb Allows for creating a relative navigation.
.Link The link of the Text relative to the root.
.Text The Name of the directory.
.Entries Information about a specific file/directory.
.URL The url of an entry.
.Leaf Currently same as '.URL' but intended to be just the name.
.IsDir Boolean for if an entry is a directory or not.
.Size Size in bytes of the entry.
.ModTime The UTC timestamp of an entry.
.MimeType The MIME type of an entry guessed from its name, or 'inode/directory'.

The server also makes the following functions available so that they can be used within the template. These functions help extend the options for dynamic rendering of HTML. They can be used to render HTML based on specific conditions.

Function Description
afterEpoch Returns the time since the epoch for the given time.
contains Checks whether a given substring is present or not in a given string.
hasPrefix Checks whether the given string begins with the specified prefix.
hasSuffix Checks whether the given string end with the specified suffix.

Authentication

By default this will serve files without needing a login.

You can either use an htpasswd file which can take lots of users, or set a single username and password with the --user and --pass flags.

Alternatively, you can have the reverse proxy manage authentication and use the username provided in the configured header with --user-from-header (e.g., --user-from-header=x-remote-user). Ensure the proxy is trusted and headers cannot be spoofed, as misconfiguration may lead to unauthorized access.

If either of the above authentication methods is not configured and client certificates are required by the --client-ca flag passed to the server, the client certificate common name will be considered as the username.

Use --htpasswd /path/to/htpasswd to provide an htpasswd file. This is in standard apache format and supports MD5, SHA1 and BCrypt for basic authentication. Bcrypt is recommended.

To create an htpasswd file:

touch htpasswd
htpasswd -B htpasswd user
htpasswd -B htpasswd anotherUser

Use the same hash format and cost for every entry in the file. If the entries differ, the time taken to reject a wrong password can reveal which usernames exist.

The password file can be updated while rclone is running.

When authentication is enabled every request must be authenticated, including HTTP OPTIONS requests, as the answer to an OPTIONS request can reveal whether a path exists. Health checks or monitoring which use unauthenticated OPTIONS requests will need to send credentials. The exception is browser CORS preflight requests when --allow-origin is set, which are answered without authentication as browsers never send credentials with them.

Use --realm to set the authentication realm.

Use --salt to change the password hashing salt from the default.

VFS - Virtual File System

This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system.

Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below.

The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory.

VFS Directory Cache

Using the --dir-cache-time flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache.

    --dir-cache-time duration   Time to cache directory entries for (default 5m0s)
    --poll-interval duration    Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s)

However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval.

You can send a SIGHUP signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this:

kill -SIGHUP $(pidof rclone)

If you configure rclone with a remote control then you can use rclone rc to flush the whole directory cache:

rclone rc vfs/forget

Or individual files or directories:

rclone rc vfs/forget file=path/to/file dir=path/to/dir

VFS File Buffering

The --buffer-size flag determines the amount of memory, that will be used to buffer data in advance.

Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared.

This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not yet read. If the buffer is empty, only a small amount of memory will be used.

The maximum memory used by rclone for buffering can be up to --buffer-size * open files.

VFS File Caching

These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility.

For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details.

Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both.

    --cache-dir string                     Directory rclone will use for caching.
    --vfs-cache-mode CacheMode             Cache mode off|minimal|writes|full (default off)
    --vfs-cache-max-age duration           Max time since last access of objects in the cache (default 1h0m0s)
    --vfs-cache-max-size SizeSuffix        Max total size of objects in the cache (default off)
    --vfs-cache-min-free-space SizeSuffix  Target minimum free space on the disk containing the cache (default off)
    --vfs-cache-poll-interval duration     Interval to poll the cache for stale objects (default 1m0s)
    --vfs-write-back duration              Time to writeback files after last use when using cache (default 5s)

If run with -vv rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with --cache-dir or setting the appropriate environment variable.

The cache has 4 different modes selected by --vfs-cache-mode. The higher the cache mode the more compatible rclone becomes at the cost of using disk space.

Note that files are written back to the remote only when they are closed and if they haven't been accessed for --vfs-write-back seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags.

If using --vfs-cache-max-size or --vfs-cache-min-free-space note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every --vfs-cache-poll-interval. Secondly because open files cannot be evicted from the cache. When --vfs-cache-max-size or --vfs-cache-min-free-space is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached.

The --vfs-cache-max-age will evict files from the cache after the set time since last access has passed; it is based on access time, not on when the file was first added to the cache. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w .

You should not run two copies of rclone using the same VFS cache with the same or overlapping remotes if using --vfs-cache-mode > off. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with --cache-dir. You don't need to worry about this if the remotes in use don't overlap.

--vfs-cache-mode off

In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk.

This will mean some operations are not possible

  • Files can't be opened for both read AND write
  • Files opened for write can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files open for read with O_TRUNC will be opened write only
  • Files open for write only will behave as if O_TRUNC was supplied
  • Open modes O_APPEND, O_TRUNC are ignored
  • If an upload fails it can't be retried

--vfs-cache-mode minimal

This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space.

These operations are not possible

  • Files opened for write only can't be seeked
  • Existing files opened for write must have O_TRUNC set
  • Files opened for write only will ignore O_APPEND, O_TRUNC
  • If an upload fails it can't be retried

--vfs-cache-mode writes

In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first.

This mode should support all normal file system operations.

If an upload fails it will be retried at exponentially increasing intervals up to 1 minute.

--vfs-cache-mode full

In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well.

In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded.

So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them.

This mode should support all normal file system operations and is otherwise identical to --vfs-cache-mode writes.

When reading a file rclone will read --buffer-size plus --vfs-read-ahead bytes ahead. The --buffer-size is buffered in memory whereas the --vfs-read-ahead is buffered on disk.

When using this mode it is recommended that --buffer-size is not set too large and --vfs-read-ahead is set large if required.

IMPORTANT not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected.

Fingerprinting

Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from:

  • size
  • modification time
  • hash

where available on an object.

On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object).

For example hash is slow with the local and sftp backends as they have to read the entire file and hash it, and modtime is slow with the s3, swift, ftp and qinqstor backends because they need to do an extra API call to fetch it.

If you use the --vfs-fast-fingerprint flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files.

If you are running a vfs cache over local, s3 or swift backends then using this flag is recommended.

Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again.

VFS Chunked Reading

When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests.

These flags control the chunking:

    --vfs-read-chunk-size SizeSuffix        Read the source objects in chunks (default 128M)
    --vfs-read-chunk-size-limit SizeSuffix  Max chunk doubling size (default off)
    --vfs-read-chunk-streams int            The number of parallel streams to read at once

The chunking behaves differently depending on the --vfs-read-chunk-streams parameter.

--vfs-read-chunk-streams == 0

Rclone will start reading a chunk of size --vfs-read-chunk-size, and then double the size for each read. When --vfs-read-chunk-size-limit is specified, and greater than --vfs-read-chunk-size, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely.

With --vfs-read-chunk-size 100M and --vfs-read-chunk-size-limit 0 the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When --vfs-read-chunk-size-limit 500M is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on.

Setting --vfs-read-chunk-size to 0 or "off" disables chunked reading.

The chunks will not be buffered in memory.

--vfs-read-chunk-streams > 0

Rclone reads --vfs-read-chunk-streams chunks of size --vfs-read-chunk-size concurrently. The size for each read will stay constant.

This improves performance massively on high latency links or very high bandwidth links to high performance object stores.

Some experimentation will be needed to find the optimum values of --vfs-read-chunk-size and --vfs-read-chunk-streams as these will depend on the backend in use and the latency to the backend.

For high performance object stores (eg AWS S3) a reasonable place to start might be --vfs-read-chunk-streams 16 and --vfs-read-chunk-size 4M. In testing with AWS S3 the performance scaled roughly as the --vfs-read-chunk-streams setting.

Similar settings should work for high latency links, but depending on the latency they may need more --vfs-read-chunk-streams in order to get the throughput.

VFS Performance

These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the chunked reading feature.

In particular S3 and Swift benefit hugely from the --no-modtime flag (or use --use-server-modtime for a slightly different effect) as each read of the modification time takes a transaction.

    --no-checksum     Don't compare checksums on up/download.
    --no-modtime      Don't read/write the modification time (can speed things up).
    --no-seek         Don't allow seeking in files.
    --read-only       Only allow read-only access.

Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file.

    --vfs-read-wait duration   Time to wait for in-sequence read before seeking (default 20ms)
    --vfs-write-wait duration  Time to wait for in-sequence write before giving error (default 1s)

When using VFS write caching (--vfs-cache-mode with value writes or full), the global flag --transfers can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag --checkers has no effect on the VFS).

    --transfers int  Number of file transfers to run in parallel (default 4)

Symlinks

By default the VFS does not support symlinks. However this may be enabled with either of the following flags:

    --links      Translate symlinks to/from regular files with a '.rclonelink' extension.
    --vfs-links  Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS

As most cloud storage systems do not support symlinks directly, rclone stores the symlink as a normal file with a special extension. So a file which appears as a symlink link-to-file.txt would be stored on cloud storage as link-to-file.txt.rclonelink and the contents would be the path to the symlink destination.

Note that --links enables symlink translation globally in rclone - this includes any backend which supports the concept (for example the local backend). --vfs-links just enables it for the VFS layer.

This scheme is compatible with that used by the local backend with the --local-links flag.

The --vfs-links flag has been designed for rclone mount, rclone nfsmount and rclone serve nfs.

It hasn't been tested with the other rclone serve commands yet.

A limitation of the current implementation is that it expects the caller to resolve sub-symlinks. For example given this directory tree

.
├── dir
│   └── file.txt
└── linked-dir -> dir

The VFS will correctly resolve linked-dir but not linked-dir/file.txt. This is not a problem for the tested commands but may be for other commands.

Note that there is an outstanding issue with symlink support issue #8245 with duplicate files being created when symlinks are moved into directories where there is a file of the same name (or vice versa).

VFS Case Sensitivity

Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file.

File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case.

Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default.

The --vfs-case-insensitive VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below.

The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote.

Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target.

If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true".

The --no-unicode-normalization flag controls whether a similar "fixup" is performed for filenames that differ but are canonically equivalent with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of false on macOS, to avoid encoding compatibility issues.

In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the --vfs-block-norm-dupes flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. --vfs-block- norm-dupes prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in rclone sync.

VFS Disk Options

This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically.

    --vfs-disk-space-total-size    Manually set the total disk space size (example: 256G, default: -1)

Alternate report of used bytes

Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running df on the filesystem, then pass the flag --vfs-used-is-size to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to rclone size and compute the total used space itself.

WARNING: Contrary to rclone size, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching.

VFS Metadata

If you use the --vfs-metadata-extension flag you can get the VFS to expose files which contain the metadata as a JSON blob. These files will not appear in the directory listing, but can be stat-ed and opened and once they have been they will appear in directory listings until the directory cache expires.

Note that some backends won't create metadata unless you pass in the --metadata flag.

For example, using rclone mount with --metadata --vfs-metadata-extension .metadata we get

$ ls -l /mnt/
total 1048577
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G

$ cat /mnt/1G.metadata
{
        "atime": "2025-03-04T17:34:22.317069787Z",
        "btime": "2025-03-03T16:03:37.708253808Z",
        "gid": "1000",
        "mode": "100664",
        "mtime": "2025-03-03T16:03:39.640238323Z",
        "uid": "1000"
}

$ ls -l /mnt/
total 1048578
-rw-rw-r-- 1 user user 1073741824 Mar  3 16:03 1G
-rw-rw-r-- 1 user user        185 Mar  3 16:03 1G.metadata

If the file has no metadata it will be returned as {} and if there is an error reading the metadata the error will be returned as {"error":"error string"}.

Auth Proxy

If you supply the parameter --auth-proxy /path/to/program then rclone will use that program to generate backends on the fly which then are used to authenticate incoming requests. This uses a simple JSON based protocol with input on STDIN and output on STDOUT.

PLEASE NOTE: --auth-proxy and --authorized-keys cannot be used together, if --auth-proxy is set the authorized keys option will be ignored.

There is an example program bin/test_proxy.py in the rclone source code.

The program's job is to take a user and pass on the input and turn those into the config for a backend on STDOUT in JSON format. This config will have any default parameters for the backend added, but it won't use configuration from environment variables or command line options - it is the job of the proxy program to make a complete config.

This config generated must have this extra parameter

  • _root - root to use for the backend

And it may have these parameters

  • _obscure - comma separated strings for parameters to obscure
  • _secret_access_key - the secret for S3 access key auth (see below)

If password authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "pass": "mypassword",
  "client_ip": "192.168.1.1"
}

If public-key authentication was used by the client, input to the proxy process (on STDIN) would look similar to this:

{
  "user": "me",
  "public_key": "AAAAB3NzaC1yc2EAAAADAQABAAABAQDuwESFdAe14hVS6omeyX7edc...JQdf",
  "client_ip": "192.168.1.1"
}

If the client authenticated with an S3 access key (rclone serve s3), the client never sends its secret, only a signature made with it, so the input contains just the access key ID as the user with no pass or public_key:

{
  "user": "AKIAIOSFODNN7EXAMPLE",
  "client_ip": "192.168.1.1"
}

In this case the program must look up the secret access key for that access key ID and return it in the _secret_access_key field of the output. Rclone then uses that secret to verify the signature on the request, refusing the request if it does not match. This means the proxy program is the source of truth for both the credentials and the backend they map to. If the program does not return _secret_access_key or returns it empty the request is refused.

The program's answer for an access key ID is cached (see below) but is checked with the program again after 5 minutes even if the access key ID is in constant use, so revoking an access key ID in the program takes effect within 5 minutes. A rotated secret takes effect on the first request signed with it.

The client_ip key holds the IP address the client connected from, without a port number. It can be used to restrict logins to certain networks, or to log authentication attempts centrally. It is omitted if the client has no IP address, for example when connecting over a unix socket. Note that if rclone is behind a reverse proxy this will be the address of the reverse proxy and not the original client.

And as an example return this on STDOUT

{
  "type": "sftp",
  "_root": "",
  "_obscure": "pass",
  "user": "me",
  "pass": "mypassword",
  "host": "sftp.example.com"
}

This would mean that an SFTP backend would be created on the fly for the user and pass/public_key returned in the output to the host given. Note that since _obscure is set to pass, rclone will obscure the pass parameter before creating the backend (which is required for sftp backends).

The program can manipulate the supplied user in any way, for example to make proxy to many different sftp backends, you could make the user be [email protected] and then set the host to example.com in the output and the user to user. For security you'd probably want to restrict the host to a limited list.

An internal cache of backends is keyed on the user, the kind of credential presented, a hash of the pass or public_key, and the client_ip. This means that if a user's password or public-key changes, the client connects from a new IP address, or the proxy returns different config parameters (eg a rotated api_key), a fresh backend will be created on the next request rather than the cached one being reused. A backend created for a public key login is never reused for a password login, so a public key cannot be sent as a password to skip the program.

A cached answer is checked with the program again after 5 minutes even if the user is in constant use, so revoking a user in the program takes effect within 5 minutes. For serve sftp this applies to new logins - an existing connection is not closed.

This can be used to build general purpose proxies to any kind of backend that rclone supports.

rclone serve webdav remote:path [flags]

Options

      --addr stringArray                       IPaddress:Port or :Port to bind server to (default 127.0.0.1:8080)
      --allow-origin string                    Origin which cross-domain request (CORS) can be executed from
      --auth-proxy string                      A program to use to create the backend from the auth
      --baseurl string                         Prefix for URLs - leave blank for root
      --cert string                            TLS PEM key (concatenation of certificate and CA certificate)
      --client-ca string                       Client certificate authority to verify clients with
      --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)
      --dir-perms FileMode                     Directory permissions (default 777)
      --disable-dir-list                       Disable HTML directory list on GET request for a directory
      --disable-zip                            Disable zip download of directories
      --etag-hash string                       Which hash to use for the ETag, or auto or blank for off
      --file-perms FileMode                    File permissions (default 666)
      --gid uint32                             Override the gid field set by the filesystem (not supported on Windows) (default 1000)
  -h, --help                                   help for webdav
      --htpasswd string                        A htpasswd file - if not provided no authentication is done
      --key string                             TLS PEM Private key
      --link-perms FileMode                    Link permissions (default 666)
      --max-header-bytes int                   Maximum size of request header (default 4096)
      --min-tls-version string                 Minimum TLS version that is acceptable (default "tls1.0")
      --no-checksum                            Don't compare checksums on up/download
      --no-modtime                             Don't read/write the modification time (can speed things up)
      --no-seek                                Don't allow seeking in files
      --pass string                            Password for authentication
      --poll-interval Duration                 Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s)
      --read-only                              Only allow read-only access
      --realm string                           Realm for authentication
      --response-header stringArray            Set HTTP header for all responses, overriding existing values
      --salt string                            Password hashing salt (default "dlPL2MqE")
      --server-read-timeout Duration           Timeout for server reading data (default 1h0m0s)
      --server-write-timeout Duration          Timeout for server writing data (default 1h0m0s)
      --template string                        User-specified template
      --uid uint32                             Override the uid field set by the filesystem (not supported on Windows) (default 1000)
      --umask FileMode                         Override the permission bits set by the filesystem (not supported on Windows) (default 002)
      --user string                            User name for authentication
      --user-from-header string                User name from a defined HTTP header
      --vfs-block-norm-dupes                   If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost)
      --vfs-cache-max-age Duration             Max time since last access of objects in the cache (default 1h0m0s)
      --vfs-cache-max-size SizeSuffix          Max total size of objects in the cache (default off)
      --vfs-cache-min-free-space SizeSuffix    Target minimum free space on the disk containing the cache (default off)
      --vfs-cache-mode CacheMode               Cache mode off|minimal|writes|full (default off)
      --vfs-cache-poll-interval Duration       Interval to poll the cache for stale objects (default 1m0s)
      --vfs-case-insensitive                   If a file name not found, find a case insensitive match
      --vfs-disk-space-total-size SizeSuffix   Specify the total space of disk (default off)
      --vfs-fast-fingerprint                   Use fast (less accurate) fingerprints for change detection
      --vfs-handle-caching Duration            Time to keep file handle and downloaders alive after last close (default 5s)
      --vfs-links                              Translate symlinks to/from regular files with a '.rclonelink' extension for the VFS
      --vfs-metadata-extension string          Set the extension to read metadata from
      --vfs-read-ahead SizeSuffix              Extra read ahead over --buffer-size when using cache-mode full
      --vfs-read-chunk-size SizeSuffix         Read the source objects in chunks (default 128Mi)
      --vfs-read-chunk-size-limit SizeSuffix   If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off)
      --vfs-read-chunk-streams int             The number of parallel streams to read at once
      --vfs-read-wait Duration                 Time to wait for in-sequence read before seeking (default 20ms)
      --vfs-refresh                            Refreshes the directory cache recursively in the background on start
      --vfs-used-is-size rclone size           Use the rclone size algorithm for Used size
      --vfs-write-back Duration                Time to writeback files after last use when using cache (default 5s)
      --vfs-write-wait Duration                Time to wait for in-sequence write before giving error (default 1s)

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

See Also

rclone settier

Changes storage class/tier of objects in remote.

Synopsis

Changes storage tier or class at remote if supported. Few cloud storage services provides different storage classes on objects, for example AWS S3 and Glacier, Azure Blob storage - Hot, Cool and Archive, Google Cloud Storage, Regional Storage, Nearline, Coldline etc.

Note that, certain tier changes make objects not available to access immediately. For example tiering to archive in azure blob storage makes objects in frozen state, user can restore by setting tier to Hot/Cool, similarly S3 to Glacier makes object inaccessible.true

You can use it to tier single object

rclone settier Cool remote:path/file

Or use rclone filters to set tier on only specific files

rclone --include "*.txt" settier Hot remote:path/dir

Or just provide remote directory and all files in directory will be tiered

rclone settier tier remote:path/dir
rclone settier tier remote:path [flags]

Options

  -h, --help   help for settier

See the global flags page for global options not listed here.

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone test

Run a test command

Synopsis

Rclone test is used to run test commands.

Select which test command you want with the subcommand, eg

rclone test memory remote:

Each subcommand has its own options which you can see in their help.

NB Be careful running these commands, they may do strange things so reading their documentation first is recommended.

Options

  -h, --help   help for test

See the global flags page for global options not listed here.

See Also

rclone test changenotify

Log any change notify requests for the remote passed in.

rclone test changenotify remote: [flags]

Options

  -h, --help                     help for changenotify
      --poll-interval Duration   Time to wait between polling for changes (default 10s)

See the global flags page for global options not listed here.

See Also

rclone test histogram

Makes a histogram of file name characters.

Synopsis

This command outputs JSON which shows the histogram of characters used in filenames in the remote:path specified.

The data doesn't contain any identifying information but is useful for the rclone developers when developing filename compression.

rclone test histogram [remote:path] [flags]

Options

  -h, --help   help for histogram

See the global flags page for global options not listed here.

See Also

rclone test info

Discovers file name or other limitations for paths.

Synopsis

Discovers what filenames and upload methods are possible to write to the paths passed in and how long they can be. It can take some time. It will write test files into the remote:path passed in. It outputs a bit of go code for each one.

NB this can create undeletable files and other hazards - use with care!

rclone test info [remote:path]+ [flags]

Options

      --all                    Run all tests
      --check-base32768        Check can store all possible base32768 characters
      --check-control          Check control characters
      --check-length           Check max filename length
      --check-normalization    Check UTF-8 Normalization
      --check-streaming        Check uploads with indeterminate file size
  -h, --help                   help for info
      --keep-test-files        Keep test files after execution
      --upload-wait Duration   Wait after writing a file (default 0s)
      --write-json string      Write results to file

See the global flags page for global options not listed here.

See Also

rclone test makefile

Make files with random contents of the size given

rclone test makefile <size> [<file>]+ [flags]

Options

      --ascii      Fill files with random ASCII printable bytes only
      --chargen    Fill files with a ASCII chargen pattern
  -h, --help       help for makefile
      --pattern    Fill files with a periodic pattern
      --seed int   Seed for the random number generator (0 for random) (default 1)
      --sparse     Make the files sparse (appear to be filled with ASCII 0x00)
      --zero       Fill files with ASCII 0x00

See the global flags page for global options not listed here.

See Also

rclone test makefiles

Make a random file hierarchy in a directory

rclone test makefiles <dir> [flags]

Options

      --ascii                      Fill files with random ASCII printable bytes only
      --chargen                    Fill files with a ASCII chargen pattern
      --files int                  Number of files to create (default 1000)
      --files-per-directory int    Average number of files per directory (default 10)
      --flat                       If set create all files in the root directory
  -h, --help                       help for makefiles
      --max-depth int              Maximum depth of directory hierarchy (default 10)
      --max-file-size SizeSuffix   Maximum size of files to create (default 100)
      --max-name-length int        Maximum size of file names (default 12)
      --min-file-size SizeSuffix   Minimum size of file to create
      --min-name-length int        Minimum size of file names (default 4)
      --pattern                    Fill files with a periodic pattern
      --seed int                   Seed for the random number generator (0 for random) (default 1)
      --sparse                     Make the files sparse (appear to be filled with ASCII 0x00)
      --zero                       Fill files with ASCII 0x00

See the global flags page for global options not listed here.

See Also

rclone test memory

Load all the objects at remote:path into memory and report memory stats.

rclone test memory remote:path [flags]

Options

  -h, --help   help for memory

See the global flags page for global options not listed here.

See Also

rclone test speed

Run a speed test to the remote

Synopsis

Run a speed test to the remote.

This command runs a series of uploads and downloads to the remote, measuring and printing the speed of each test using varying file sizes and numbers of files.

Test time can be inaccurate with small file caps and large files. As it uses the results of an initial test to determine how many files to use in each subsequent test.

It is recommended to use -q flag for a simpler output. e.g.:

rclone test speed remote: -q

NB This command will create and delete files on the remote in a randomly named directory which will be automatically removed on a clean exit.

You can use the --json flag to only print the results in JSON format.

rclone test speed <remote> [flags]

Options

      --ascii                Fill files with random ASCII printable bytes only
      --chargen              Fill files with a ASCII chargen pattern
      --file-cap int         Maximum number of files to use in each test (default 100)
  -h, --help                 help for speed
      --json                 Output only results in JSON format
      --large SizeSuffix     Size of large files (default 1Gi)
      --medium SizeSuffix    Size of medium files (default 10Mi)
      --pattern              Fill files with a periodic pattern
      --seed int             Seed for the random number generator (0 for random) (default 1)
      --small SizeSuffix     Size of small files (default 1Ki)
      --sparse               Make the files sparse (appear to be filled with ASCII 0x00)
      --test-time Duration   Length for each test to run (default 15s)
      --zero                 Fill files with ASCII 0x00

See the global flags page for global options not listed here.

See Also

rclone touch

Create new file or change file modification time.

Synopsis

Set the modification time on file(s) as specified by remote:path to have the current time.

If remote:path does not exist then a zero sized file will be created, unless --no-create or --recursive is provided.

If --recursive is used then recursively sets the modification time on all existing files that is found under the path. Filters are supported, and you can test with the --dry-run or the --interactive/-i flag. This will touch --transfers files concurrently.

If --timestamp is used then sets the modification time to that time instead of the current time. Times may be specified as one of:

  • 'YYMMDD' - e.g. 171030
  • 'YYYY-MM-DD' - e.g. 2006-01-02
  • 'YYYY-MM-DDTHH:MM:SS' - e.g. 2006-01-02T15:04:05
  • 'YYYY-MM-DDTHH:MM:SS.SSS' - e.g. 2006-01-02T15:04:05.123456789
  • 'YYYY-MM-DDTHH:MM:SSZ' - e.g. 2006-01-02T15:04:05Z

Note that value of --timestamp is in UTC. If you want local time then add the --localtime flag.

Metadata can be added when creating a new file with --metadata-set. For example: rclone touch remote:path -M --metadata-set key=value

rclone touch remote:path [flags]

Options

  -h, --help               help for touch
      --localtime          Use localtime for timestamp, not UTC
  -C, --no-create          Do not create the file if it does not exist (implied with --recursive)
  -R, --recursive          Recursively touch all files
  -t, --timestamp string   Use specified time instead of the current time of day

Options shared with other commands are described next. See the global flags page for global options not listed here.

Important Options

Important flags useful for most commands

  -n, --dry-run         Do a trial run with no permanent changes
  -i, --interactive     Enable interactive mode
  -v, --verbose count   Print lots more stuff (repeat for more)

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

rclone tree

List the contents of the remote in a tree like fashion.

Synopsis

Lists the contents of a remote in a similar way to the unix tree command.

For example

$ rclone tree remote:path
/
├── file1
├── file2
├── file3
└── subdir
    ├── file4
    └── file5

1 directories, 5 files

You can use any of the filtering options with the tree command (e.g. --include and --exclude. You can also use --fast-list.

The tree command has many options for controlling the listing which are compatible with the tree command, for example you can include file sizes with --size. Note that not all of them have short options as they conflict with rclone's short options.

For a more interactive navigation of the remote see the ncdu command.

rclone tree remote:path [flags]

Options

  -a, --all             All files are listed (list . files too)
  -d, --dirs-only       List directories only
      --dirsfirst       List directories before files (-U disables)
      --full-path       Print the full path prefix for each file
  -h, --help            help for tree
      --level int       Descend only level directories deep
  -D, --modtime         Print the date of last modification.
      --noindent        Don't print indentation lines
      --noreport        Turn off file/directory count at end of tree listing
  -o, --output string   Output to file instead of stdout
  -p, --protections     Print the protections for each file.
  -Q, --quote           Quote filenames with double quotes.
  -s, --size            Print the size in bytes of each file.
      --sort string     Select sort: name,version,size,mtime,ctime
      --sort-ctime      Sort files by last status change time
  -t, --sort-modtime    Sort files by last modification time
  -r, --sort-reverse    Reverse the order of the sort
  -U, --unsorted        Leave files unsorted
      --version         Sort files alphanumerically by version

Options shared with other commands are described next. See the global flags page for global options not listed here.

Filter Options

Flags for filtering directory listings

      --delete-excluded                     Delete files on dest excluded from sync
      --exclude stringArray                 Exclude files matching pattern
      --exclude-from stringArray            Read file exclude patterns from file (use - to read from stdin)
      --exclude-if-present stringArray      Exclude directories if filename is present
      --files-from stringArray              Read list of source-file names from file (use - to read from stdin)
      --files-from-raw stringArray          Read list of source-file names from file without any processing of lines (use - to read from stdin)
      --files-from0 stringArray             Read list of source-file names from file using NUL as separator (use - to read from stdin)
  -f, --filter stringArray                  Add a file filtering rule
      --filter-from stringArray             Read file filtering patterns from a file (use - to read from stdin)
      --hash-filter string                  Partition filenames by hash k/n or randomly @/n
      --ignore-case                         Ignore case in filters (case insensitive)
      --include stringArray                 Include files matching pattern
      --include-from stringArray            Read file include patterns from file (use - to read from stdin)
      --max-age Duration                    Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --max-depth int                       If set limits the recursion depth to this (default -1)
      --max-size SizeSuffix                 Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off)
      --metadata-exclude stringArray        Exclude metadatas matching pattern
      --metadata-exclude-from stringArray   Read metadata exclude patterns from file (use - to read from stdin)
      --metadata-filter stringArray         Add a metadata filtering rule
      --metadata-filter-from stringArray    Read metadata filtering patterns from a file (use - to read from stdin)
      --metadata-include stringArray        Include metadatas matching pattern
      --metadata-include-from stringArray   Read metadata include patterns from file (use - to read from stdin)
      --min-age Duration                    Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off)
      --min-size SizeSuffix                 Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off)

Listing Options

Flags for listing directories

      --default-time Time   Time to show if modtime is unknown for files and directories (default 2000-01-01T00:00:00Z)
      --fast-list           Use recursive list if available; uses more memory but fewer transactions

See Also

  • rclone - Show help for rclone commands, flags and backends.

Copying single files

rclone normally syncs or copies directories. However, if the source remote points to a file, rclone will just copy that file. The destination remote must point to a directory - rclone will give the error Failed to create file system for "remote:file": is a file not a directory if it isn't.

For example, suppose you have a remote with a file in called test.jpg, then you could copy just that file like this

rclone copy remote:test.jpg /tmp/download

The file test.jpg will be placed inside /tmp/download.

This is equivalent to specifying

rclone copy --files-from /tmp/files remote: /tmp/download

Where /tmp/files contains the single line

test.jpg

It is recommended to use copy when copying individual files, not sync. They have pretty much the same effect but copy will use a lot less memory.

Syntax of remote paths

The syntax of the paths passed to the rclone command are as follows.

/path/to/dir

This refers to the local file system.

On Windows \ may be used instead of / in local paths **only**, non local paths must use /. See local filesystem documentation for more about Windows-specific paths.

These paths needn't start with a leading / - if they don't then they will be relative to the current directory.

remote:path/to/dir

This refers to a directory path/to/dir on remote: as defined in the config file (configured with rclone config).

remote:/path/to/dir

On most backends this is refers to the same directory as remote:path/to/dir and that format should be preferred. On a very small number of remotes (FTP, SFTP, Dropbox for business) this will refer to a different directory. On these, paths without a leading / will refer to your "home" directory and paths with a leading / will refer to the root.

:backend:path/to/dir

This is an advanced form for creating remotes on the fly. backend should be the name or prefix of a backend (the type in the config file) and all the configuration for the backend should be provided on the command line (or in environment variables).

Here are some examples:

rclone lsd --http-url https://pub.rclone.org :http:

To list all the directories in the root of https://pub.rclone.org/.

rclone lsf --http-url https://example.com :http:path/to/dir

To list files and directories in https://example.com/path/to/dir/

rclone copy --http-url https://example.com :http:path/to/dir /tmp/dir

To copy files and directories in https://example.com/path/to/dir to /tmp/dir.

rclone copy --sftp-host example.com :sftp:path/to/dir /tmp/dir

To copy files and directories from example.com in the relative directory path/to/dir to /tmp/dir using sftp.

Connection strings {#connection-strings}

The above examples can also be written using a connection string syntax, so instead of providing the arguments as command line parameters --http-url https://pub.rclone.org they are provided as part of the remote specification as a kind of connection string.

rclone lsd ":http,url='https://pub.rclone.org':"
rclone lsf ":http,url='https://example.com':path/to/dir"
rclone copy ":http,url='https://example.com':path/to/dir" /tmp/dir
rclone copy :sftp,host=example.com:path/to/dir /tmp/dir

These can apply to modify existing remotes as well as create new remotes with the on the fly syntax. This example is equivalent to adding the --drive-shared-with-me parameter to the remote gdrive:.

rclone lsf "gdrive,shared_with_me:path/to/dir"

The major advantage to using the connection string style syntax is that it only applies to the remote, not to all the remotes of that type of the command line. A common confusion is this attempt to copy a file shared on google drive to the normal drive which does not work because the --drive-shared-with-me flag applies to both the source and the destination.

rclone copy --drive-shared-with-me gdrive:shared-file.txt gdrive:

However using the connection string syntax, this does work.

rclone copy "gdrive,shared_with_me:shared-file.txt" gdrive:

Note that the connection string only affects the options of the immediate backend. If for example gdriveCrypt is a crypt based on gdrive, then the following command will not work as intended, because shared_with_me is ignored by the crypt backend:

rclone copy "gdriveCrypt,shared_with_me:shared-file.txt" gdriveCrypt:

The connection strings have the following syntax

remote,parameter=value,parameter2=value2:path/to/dir
:backend,parameter=value,parameter2=value2:path/to/dir

If the parameter has a : or , then it must be placed in quotes " or ', so

remote,parameter="colon:value",parameter2="comma,value":path/to/dir
:backend,parameter='colon:value',parameter2='comma,value':path/to/dir

If a quoted value needs to include that quote, then it should be doubled, so

remote,parameter="with""quote",parameter2='with''quote':path/to/dir

This will make parameter be with"quote and parameter2 be with'quote.

If you leave off the =parameter then rclone will substitute =true which works very well with flags. For example, to use s3 configured in the environment you could use:

rclone lsd :s3,env_auth:

Which is equivalent to

rclone lsd :s3,env_auth=true:

Note that on the command line you might need to surround these connection strings with " or ' to stop the shell interpreting any special characters within them.

If you are a shell master then you'll know which strings are OK and which aren't, but if you aren't sure then enclose them in " and use ' as the inside quote. This syntax works on all OSes.

rclone copy ":http,url='https://example.com':path/to/dir" /tmp/dir

On Linux/macOS some characters are still interpreted inside " strings in the shell (notably \ and $ and ") so if your strings contain those you can swap the roles of " and ' thus. (This syntax does not work on Windows.)

rclone copy ':http,url="https://example.com":path/to/dir' /tmp/dir

You can use rclone config string to convert a remote into a connection string.

Connection strings, config and logging

If you supply extra configuration to a backend by command line flag, environment variable or connection string then rclone will add a suffix based on the hash of the config to the name of the remote, eg

rclone -vv lsf --s3-chunk-size 20M s3:

Has the log message

DEBUG : s3: detected overridden config - adding "{Srj1p}" suffix to name

This is so rclone can tell the modified remote apart from the unmodified remote when caching the backends.

This should only be noticeable in the logs.

This means that on the fly backends such as

rclone -vv lsf :s3,env_auth:

Will get their own names

DEBUG : :s3: detected overridden config - adding "{YTu53}" suffix to name

Valid remote names

Remote names are case sensitive, and must adhere to the following rules:

  • May contain number, letter, _, -, ., +, @ and space.
  • May not start with - or space.
  • May not end with space.

Starting with rclone version 1.61, any Unicode numbers and letters are allowed, while in older versions it was limited to plain ASCII (0-9, A-Z, a-z). If you use the same rclone configuration from different shells, which may be configured with different character encoding, you must be cautious to use characters that are possible to write in all of them. This is mostly a problem on Windows, where the console traditionally uses a non-Unicode character set - defined by the so-called "code page".

If you want to configure a remote using environment variables then its name becomes part of the environment variable name, which in practice restricts the characters you can use in it. See the Config file section under Environment variables for details.

Do not use single character names on Windows as it creates ambiguity with Windows drives' names, e.g.: remote called C is indistinguishable from C drive. Rclone will always assume that single letter name refers to a drive.

Adding global configuration to a remote {#globalconfig}

It is possible to add global configuration to the remote configuration which will be applied just before the remote is created.

This can be done in two ways. The first is to use override.var = value in the config file or the connection string for a temporary change, and the second is to use global.var = value in the config file or connection string for a permanent change.

This is explained fully below.

override.var

This is used to override a global variable just for the duration of the remote creation. It won't affect other remotes even if they are created at the same time.

This is very useful for overriding networking config needed for just for that remote. For example, say you have a remote which needs --no-check-certificate as it is running on test infrastructure without a proper certificate. You could supply the --no-check-certificate flag to rclone, but this will affect all the remotes. To make it just affect this remote you use an override. You could put this in the config file:

[remote]
type = XXX
...
override.no_check_certificate = true

or use it in the connection string remote,override.no_check_certificate=true: (or just remote,override.no_check_certificate:).

Note how the global flag name loses its initial -- and gets - replaced with _ and gets an override. prefix.

Not all global variables make sense to be overridden like this as the config is only applied during the remote creation. Here is a non exhaustive list of ones which might be useful:

  • allow_https_downgrade
  • bind_addr
  • ca_cert
  • client_cert
  • client_key
  • connect_timeout
  • disable_http2
  • disable_http_keep_alives
  • dump
  • expect_continue_timeout
  • headers
  • http_proxy
  • low_level_retries
  • max_connections
  • no_check_certificate
  • no_gzip
  • timeout
  • traffic_class
  • use_cookies
  • use_server_modtime
  • user_agent

An override.var will override all other config methods, but just for the duration of the creation of the remote.

global.var

This is used to set a global variable for everything. The global variable is set just before the remote is created.

This is useful for parameters (eg sync parameters) which can't be set as an override. For example, say you have a remote where you would always like to use the --checksum flag. You could supply the --checksum flag to rclone on every command line, but instead you could put this in the config file:

[remote]
type = XXX
...
global.checksum = true

or use it in the connection string remote,global.checksum=true: (or just remote,global.checksum:). This is equivalent to using the --checksum flag.

Note how the global flag name loses its initial -- and gets - replaced with _ and gets a global. prefix.

Any global variable can be set like this and it is exactly equivalent to using the equivalent flag on the command line. This means it will affect all uses of rclone.

If two remotes set the same global variable then the first one instantiated will be overridden by the second one. A global.var will override all other config methods when the remote is created.

Quoting and the shell

When you are typing commands to your computer you are using something called the command line shell. This interprets various characters in an OS specific way.

Here are some gotchas which may help users unfamiliar with the shell rules

Linux / macOS

If your names have spaces or shell metacharacters (e.g. *, ?, $, ', ", etc.) then you must quote them. Use single quotes ' by default.

rclone copy 'Important files?' remote:backup

If you want to send a ' you will need to use ", e.g.

rclone copy "O'Reilly Reviews" remote:backup

The rules for quoting metacharacters are complicated and if you want the full details you'll have to consult the manual page for your shell.

Windows

If your names have spaces in you need to put them in ", e.g.

rclone copy "E:\folder name\folder name\folder name" remote:backup

If you are using the root directory on its own then don't quote it (see #464 for why), e.g.

rclone copy E:\ remote:backup

Copying files or directories with : in the names

rclone uses : to mark a remote name. This is, however, a valid filename component in non-Windows OSes. The remote name parser will only search for a : up to the first / so if you need to act on a file or directory like this then use the full path starting with a /, or use ./ as a current directory prefix.

So to sync a directory called sync:me to a remote called remote: use

rclone sync --interactive ./sync:me remote:path

or

rclone sync --interactive /full/path/to/sync:me remote:path

Server-side copy

Most remotes (but not all - see the overview) support server-side copy.

This means if you want to copy one folder to another then rclone won't download all the files and re-upload them; it will instruct the server to copy them in place.

Eg

rclone copy s3:oldbucket s3:newbucket

Will copy the contents of oldbucket to newbucket without downloading and re-uploading.

Remotes which don't support server-side copy will download and re-upload in this case.

Server-side copies are used with sync and copy and will be identified in the log when using the -v flag. The move command may also use them if remote doesn't support server-side move directly. This is done by issuing a server-side copy then a delete which is much quicker than a download and re-upload.

Server-side copies will only be attempted if the remote names are the same.

This can be used when scripting to make aged backups efficiently, e.g.

rclone sync --interactive remote:current-backup remote:previous-backup
rclone sync --interactive /path/to/files remote:current-backup

Metadata support {#metadata}

Metadata is data about a file (or directory) which isn't the contents of the file (or directory). Normally rclone only preserves the modification time and the content (MIME) type where possible.

Rclone supports preserving all the available metadata on files and directories when using the --metadata or -M flag.

Exactly what metadata is supported and what that support means depends on the backend. Backends that support metadata have a metadata section in their docs and are listed in the features table (Eg local, s3)

Some backends don't support metadata, some only support metadata on files and some support metadata on both files and directories.

Rclone only supports a one-time sync of metadata. This means that metadata will be synced from the source object to the destination object only when the source object has changed and needs to be re-uploaded. If the metadata subsequently changes on the source object without changing the object itself then it won't be synced to the destination object. This is in line with the way rclone syncs Content-Type without the --metadata flag.

Using --metadata when syncing from local to local will preserve file attributes such as file mode, owner, extended attributes (not Windows).

Note that arbitrary metadata may be added to objects using the --metadata-set key=value flag when the object is first uploaded. This flag can be repeated as many times as necessary.

The --metadata-mapper flag can be used to pass the name of a program in which can transform metadata when it is being copied from source to destination.

Rclone supports --metadata-set and --metadata-mapper when doing server-side Move and server-side Copy, but not when doing server side DirMove (renaming a directory) as this would involve recursing into the directory. Note that you can disable DirMove with --disable DirMove and rclone will revert back to using Move for each individual object where --metadata-set and --metadata-mapper are supported.

Types of metadata

Metadata is divided into two type. System metadata and User metadata.

Metadata which the backend uses itself is called system metadata. For example on the local backend the system metadata uid will store the user ID of the file when used on a unix based platform.

Arbitrary metadata is called user metadata and this can be set however is desired.

When objects are copied from backend to backend, they will attempt to interpret system metadata if it is supplied. Metadata may change from being user metadata to system metadata as objects are copied between different backends. For example copying an object from s3 sets the content-type metadata. In a backend which understands this (like azureblob) this will become the Content-Type of the object. In a backend which doesn't understand this (like the local backend) this will become user metadata. However should the local object be copied back to s3, the Content-Type will be set correctly.

Metadata framework

Rclone implements a metadata framework which can read metadata from an object and write it to the object when (and only when) it is being uploaded.

This metadata is stored as a dictionary with string keys and string values.

There are some limits on the names of the keys (these may be clarified further in the future).

  • must be lower case
  • may be a-z 0-9 containing . - or _
  • length is backend dependent

Each backend can provide system metadata that it understands. Some backends can also store arbitrary user metadata.

Where possible the key names are standardized, so, for example, it is possible to copy object metadata from s3 to azureblob for example and metadata will be translated appropriately.

Some backends have limits on the size of the metadata and rclone will give errors on upload if they are exceeded.

Metadata preservation

The goal of the implementation is to

  1. Preserve metadata if at all possible
  2. Interpret metadata if at all possible

The consequences of 1 is that you can copy an S3 object to a local disk then back to S3 losslessly. Likewise you can copy a local file with file attributes and xattrs from local disk to s3 and back again losslessly.

The consequence of 2 is that you can copy an S3 object with metadata to Azureblob (say) and have the metadata appear on the Azureblob object also.

Standard system metadata

Here is a table of standard system metadata which, if appropriate, a backend may implement.

key description example
mode File type and mode: octal, unix style 0100664
uid User ID of owner: decimal number 500
gid Group ID of owner: decimal number 500
rdev Device ID (if special file) => hexadecimal 0
atime Time of last access: RFC 3339 2006-01-02T15:04:05.999999999Z07:00
mtime Time of last modification: RFC 3339 2006-01-02T15:04:05.999999999Z07:00
btime Time of file creation (birth): RFC 3339 2006-01-02T15:04:05.999999999Z07:00
utime Time of file upload: RFC 3339 2006-01-02T15:04:05.999999999Z07:00
cache-control Cache-Control header no-cache
content-disposition Content-Disposition header inline
content-encoding Content-Encoding header gzip
content-language Content-Language header en-US
content-type Content-Type header text/plain

The metadata keys mtime and content-type will take precedence if supplied in the metadata over reading the Content-Type or modification time of the source object.

Hashes are not included in system metadata as there is a well defined way of reading those already.

Performance {#performance}

Rclone's defaults are fairly conservative. They are chosen to work on small machines, slow links and with cloud providers which rate limit heavily. If you have plenty of bandwidth, memory and a fast storage system then rclone can usually be made to run a great deal faster by adjusting a few flags.

The right values depend on the size of your files, your network and the storage systems at each end, so treat the numbers below as starting points. Change one thing at a time and measure the result.

Measuring

Use rclone copy with -P/--progress to watch the transfer rate while you are experimenting. For unattended runs log the stats instead, e.g. --stats 1m --stats-one-line --stats-log-level NOTICE.

A transfer is always limited by something: the network bandwidth, the rate at which the storage system will accept requests, disk IO, CPU or memory. Tuning is working out which limit you are hitting and raising it until you hit the next one. Watch the network, disk, CPU and memory use of the machine rclone is running on as well as rclone's stats.

Lots of files

The time taken to transfer small files is dominated by the round trip time of the requests rather than by the bandwidth, so the way to make them go faster is to do more of them at once.

--transfers sets the number of files transferred in parallel. The default is 4. Double this until the transfer rate stops increasing or the storage system starts returning errors. Values of 32 to 64 are common and values in the hundreds work well with the large object storage systems (e.g. s3, azureblob, googlecloudstorage, b2) when transferring lots of small files.

--checkers sets the number of directory listings and file comparisons done in parallel. The default is 8. Raising this speeds up syncs where most of the files are already present on the destination and stops the checking being the bottleneck when --transfers is large. 16 to 64 is a reasonable range.

Not all storage systems like lots of parallel requests. If you see rate limiting errors (e.g. HTTP 429 or 503 errors) or lots of low level retries in the log then you have gone too far, so reduce --transfers and --checkers or use --tpslimit to cap the request rate. Local disks, especially spinning ones, and network file systems may get slower rather than faster with too much parallelism.

Large files

A single stream often won't fill a fast link, so rclone can transfer a large file as several chunks in parallel. These are called multi-thread transfers and are used for files bigger than --multi-thread-cutoff on backends which support them.

The number of chunks transferred in parallel is set by --multi-thread-streams (default 4) and this works for all backends so increase this first to make large files transfer faster.

The size of the chunks is set by the backend, usually with a --backend-chunk-size flag, e.g. --s3-chunk-size. Backends without a natural chunk size (e.g. local, smb) use --multi-thread-chunk-size. Larger chunks mean fewer requests which increases throughput on big files. See the docs for the backend you are using for details.

Chunked uploads are usually buffered in memory, so in the worst case rclone will use

--transfers * upload concurrency * chunk size

of memory for buffers. For example --transfers 256 with an upload concurrency of 8 and a chunk size of 16 MiB could use up to 32 GiB. When you raise --transfers to speed up small files you may therefore need to lower the upload concurrency, or set --max-buffer-memory which limits the total memory used for buffers and makes transfers wait when the limit is reached.

--use-mmap makes rclone return buffer memory to the OS as soon as it has finished with it which is helpful when using lots of transfers or large chunks.

A mixture of large and small files

By default rclone transfers files in roughly the order it finds them. With a mixture of file sizes this means that at times all the transfers are large files, which fill the bandwidth but make few requests, and at other times all the transfers are small files, which make lots of requests but use little of the bandwidth.

Using --order-by size,mixed,50 makes half of the transfers take the smallest files in the backlog and half of them take the largest. This keeps both large and small files moving at once so the bandwidth and the request rate are both kept full for the whole of the run.

This works best if rclone has lots of files to choose from so use it with a big --max-backlog, e.g. --max-backlog 2000000. Each item in the backlog uses in the order of 1 KiB of memory.

Doing less work

Fewer requests to the storage system means a faster transfer.

  • If the source and destination are on the same remote then rclone will use server-side copy where it is supported so the data doesn't need to be downloaded and uploaded again.
  • --fast-list lists bucket-based remotes with fewer transactions at the cost of using more memory. This may be faster or slower than the default so it is worth testing.
  • --no-traverse stops rclone listing the destination which helps when copying a small number of files into a large destination.
  • --size-only or --checksum avoid reading the modification time. On some object storage systems (e.g. s3, swift) this needs an extra request for each file. Alternatively use --use-server-modtime with --update.
  • --no-check-dest skips checking the destination entirely. Only use this if you know that none of the files are there already.

Options

Rclone has a number of options to control its behaviour. These are documented below, and in the flags page.

Options that take parameters can have the values passed in two ways, --option=value or --option value. However boolean (true/false) options behave slightly differently to the other options in that --boolean sets the option to true and the absence of the flag sets it to false. It is also possible to specify --boolean=false or --boolean=true. Note that --boolean false is not valid - this is parsed as --boolean and the false is parsed as an extra command line argument for rclone.

String values that are recognized as special identifiers, e.g. the name of the log level to set with option --log-level, are case insensitive, e.g. --log-level ERROR and --log-level error are identical.

Options documented to take a stringArray parameter accept multiple values. To pass more than one value, repeat the option; for example: --include value1 --include value2. Other options may only accept a single value, and should only be specified once, but where the specified parameter may indicate a list of values separated by space or comma. Such options are documented to take a CommaSepList parameter, if comma separated, and SpaceSepList if space separated, although some may also have different parameters such as DumpFlags or just string and the help text explains that this will be interpreted as a list.

Floating-point values with fractional part must use period (.) as decimal separator, common in English-speaking countries, regardless of your configured system locale. Parameter type float accepts decimal and hexadecimal floating-point numbers as defined by the Go syntax for floating-point literals.

Time and duration options {#time-options}

Options that take a Time or Duration parameter must be specified as a formatted string describing an absolute or relative time. Note that both Time and Duration parameter types can be expressed as either absolute or relative time, just with different interpretations, e.g. a relative time will be treated as an offset from the current time when passed as a Time value.

An absolute time can be specified as a string in one of the following formats:

  • RFC3339 - e.g. 2006-01-02T15:04:05Z or 2006-01-02T15:04:05+07:00
  • ISO8601 Date and time, local timezone - 2006-01-02T15:04:05
  • ISO8601 Date and time, local timezone - 2006-01-02 15:04:05
  • ISO8601 Date - 2006-01-02 (YYYY-MM-DD)

A relative time is a string with a, possibly signed, sequence of decimal numbers, each with optional fraction, and each with a unit suffix. If the string only contains a single number, then the unit suffix is optional and will default to seconds, i.e. a plain decimal value will be treated as a number of seconds. The following suffixes are valid:

  • ms - Milliseconds
  • s - Seconds
  • m - Minutes
  • h - Hours
  • d - Days
  • w - Weeks
  • M - Months
  • y - Years

Examples: "10", "300ms", "-1.5h", "2h45m" or "1d12h".

Size options

Options that take a SizeSuffix parameter can be specified as an integer value, which will then be assumed to represent a KiB value (multiples of 1024 bytes) by default. The interpretation can be changed by appending a suffix: B for Byte, K for KiB, M for MiB, G for GiB, T for TiB and P for PiB. These are the binary units, e.g. 1, 2**10, 2**20, 2**30 respectively.

See also --human-readable.

Main options

--allow-https-downgrade

By default rclone refuses to follow a redirect from an https:// URL to a plain http:// URL, as any credentials or other headers in the request would then be sent unencrypted, and the response could be altered by anyone on the network path. The request fails with an error which mentions this flag.

Set --allow-https-downgrade to follow such redirects, for example when downloading from a mirror which redirects to plain HTTP. Some backends still refuse to follow them, as doing so would send their credentials in cleartext.

This option defaults to false.

--backup-dir string

When using sync, copy or move, any files which would have been overwritten or deleted are moved in their original hierarchy into this directory.

If --suffix is set, then the moved files will have the suffix added to them. If there is a file with the same path (after the suffix has been added) in the directory, then it will be overwritten.

The remote in use must support server-side move or copy and you must use the same remote as the destination of the sync. The backup directory must not overlap the destination directory without it being excluded by a filter rule.

For example

rclone sync --interactive /path/to/local remote:current --backup-dir remote:old

will sync /path/to/local to remote:current, but for any files which would have been updated or deleted will be stored in remote:old.

If running rclone from a script you might want to use today's date as the directory name passed to --backup-dir to store the old files, or you might want to pass --suffix with today's date. This can be done with --suffix $(date +%F) in bash, and --suffix $(Get-Date -Format 'yyyy-MM-dd') in PowerShell.

See --compare-dest and --copy-dest.

--bind string

Local address to bind to for outgoing connections. This can be an IPv4 address (1.2.3.4), an IPv6 address (1234::789A) or host name. If the host name doesn't resolve or resolves to more than one IP address it will give an error.

You can use --bind 0.0.0.0 to force rclone to use IPv4 addresses and --bind ::0 to force rclone to use IPv6 addresses.

--bwlimit BwTimetable

This option controls the bandwidth limit. For example

--bwlimit 10M

would mean limit the upload and download bandwidth to 10 MiB/s. NB this is bytes per second not bits per second. To use a single limit, specify the desired bandwidth in KiB/s, or use a suffix B|K|M|G|T|P. The default is 0 which means to not limit bandwidth.

The upload and download bandwidth can be specified separately, as --bwlimit UP:DOWN, so

--bwlimit 10M:100k

would mean limit the upload bandwidth to 10 MiB/s and the download bandwidth to 100 KiB/s. Either limit can be "off" meaning no limit, so to just limit the upload bandwidth you would use

--bwlimit 10M:off

this would limit the upload bandwidth to 10 MiB/s but the download bandwidth would be unlimited.

When specified as above the bandwidth limits last for the duration of run of the rclone binary.

It is also possible to specify a "timetable" of limits, which will cause certain limits to be applied at certain times. To specify a timetable, format your entries as WEEKDAY-HH:MM,BANDWIDTH WEEKDAY-HH:MM,BANDWIDTH... where: WEEKDAY is optional element.

  • BANDWIDTH can be a single number, e.g.100k or a pair of numbers for upload:download, e.g.10M:1M.
  • WEEKDAY can be written as the whole word or only using the first 3 characters. It is optional.
  • HH:MM is an hour from 00:00 to 23:59.

Entries can be separated by spaces or semicolons.

Note: Semicolons can be used as separators instead of spaces to avoid parsing issues in environments like Docker.

An example of a typical timetable to avoid link saturation during daytime working hours could be:

Using spaces as separators: --bwlimit "08:00,512k 12:00,10M 13:00,512k 18:00,30M 23:00,off"

Using semicolons as separators: --bwlimit "08:00,512k;12:00,10M;13:00,512k;18:00,30M;23:00,off"

In these examples, the transfer bandwidth will be set to 512 KiB/s at 8am every day. At noon, it will rise to 10 MiB/s, and drop back to 512 KiB/sec at 1pm. At 6pm, the bandwidth limit will be set to 30 MiB/s, and at 11pm it will be completely disabled (full speed). Anything between 11pm and 8am will remain unlimited.

An example of timetable with WEEKDAY could be:

Using spaces as separators: --bwlimit "Mon-00:00,512 Fri-23:59,10M Sat-10:00,1M Sun-20:00,off"

Using semicolons as separators: --bwlimit "Mon-00:00,512;Fri-23:59,10M;Sat-10:00,1M;Sun-20:00,off"

It means that, the transfer bandwidth will be set to 512 KiB/s on Monday. It will rise to 10 MiB/s before the end of Friday. At 10:00 on Saturday it will be set to 1 MiB/s. From 20:00 on Sunday it will be unlimited.

Timeslots without WEEKDAY are extended to the whole week. So this example:

--bwlimit "Mon-00:00,512 12:00,1M Sun-20:00,off"

Is equivalent to this:

--bwlimit "Mon-00:00,512 Mon-12:00,1M Tue-12:00,1M Wed-12:00,1M Thu-12:00,1M Fri-12:00,1M Sat-12:00,1M Sun-12:00,1M Sun-20:00,off"

Bandwidth limit apply to the data transfer for all backends. For most backends the directory listing bandwidth is also included (exceptions being the non HTTP backends, ftp, sftp and storj).

Note that the units are Byte/s, not bit/s. Typically connections are measured in bit/s - to convert divide by 8. For example, let's say you have a 10 Mbit/s connection and you wish rclone to use half of it - 5 Mbit/s. This is 5/8 = 0.625 MiB/s so you would use a --bwlimit 0.625M parameter for rclone.

On Unix systems (Linux, macOS, …) the bandwidth limiter can be toggled by sending a SIGUSR2 signal to rclone. This allows to remove the limitations of a long running rclone transfer and to restore it back to the value specified with --bwlimit quickly when needed. Assuming there is only one rclone instance running, you can toggle the limiter like this:

kill -SIGUSR2 $(pidof rclone)

If you configure rclone with a remote control then you can use change the bwlimit dynamically:

rclone rc core/bwlimit rate=1M

--bwlimit-file BwTimetable

This option controls per file bandwidth limit. For the options see the --bwlimit flag.

For example use this to allow no transfers to be faster than 1 MiB/s

--bwlimit-file 1M

This can be used in conjunction with --bwlimit.

Note that if a schedule is provided the file will use the schedule in effect at the start of the transfer.

--buffer-size SizeSuffix

Use this sized buffer to speed up file transfers. Each --transfer will use this much memory for buffering.

When using mount or cmount each open file descriptor will use this much memory for buffering. See the mount documentation for more details.

Set to 0 to disable the buffering for the minimum memory usage.

Note that the memory allocation of the buffers is influenced by the --use-mmap flag.

--cache-dir string

Specify the directory rclone will use for caching, to override the default.

Default value is depending on operating system:

  • Windows %LocalAppData%\rclone, if LocalAppData is defined.
  • macOS $HOME/Library/Caches/rclone if HOME is defined.
  • Unix $XDG_CACHE_HOME/rclone if XDG_CACHE_HOME is defined, else $HOME/.cache/rclone if HOME is defined.
  • Fallback (on all OS) to $TMPDIR/rclone, where TMPDIR is the value from --temp-dir.

You can use the config paths command to see the current value.

Cache directory is heavily used by the VFS File Caching mount feature, but also by serve, GUI and other parts of rclone.

--check-first

If this flag is set then in a sync, copy or move, rclone will do all the checks to see whether files need to be transferred before doing any of the transfers. Normally rclone would start running transfers as soon as possible.

This flag can be useful on IO limited systems where transfers interfere with checking.

It can also be useful to ensure perfect ordering when using --order-by.

If both --check-first and --order-by are set when doing rclone move then rclone will use the transfer thread to delete source files which don't need transferring. This will enable perfect ordering of the transfers and deletes but will cause the transfer stats to have more items in than expected.

Using this flag can use more memory as it effectively sets --max-backlog to infinite. This means that all the info on the objects to transfer is held in memory before the transfers start.

--checkers int

Originally controlling just the number of file checkers to run in parallel, e.g. by rclone copy. Now a fairly universal parallelism control used by rclone in several places.

Note: checkers do the equality checking of files during a sync. For some storage systems (e.g. S3, Swift, Dropbox) this can take a significant amount of time so they are run in parallel.

The default is to run 8 checkers in parallel. However, in case of slow-reacting backends you may need to lower (rather than increase) this default by setting --checkers to 4 or less threads. This is especially advised if you are experiencing backend server crashes during file checking phase (e.g. on subsequent or top-up backups where little or no file copying is done and checking takes up most of the time). Increase this setting only with utmost care, while monitoring your server health and file checking throughput.

-c, --checksum

Normally rclone will look at modification time and size of files to see if they are equal. If you set this flag then rclone will check the file hash and size to determine if files are equal.

This is useful when the remote doesn't support setting modified time and a more accurate sync is desired than just checking the file size.

This is very useful when transferring between remotes which store the same hash type on the object, e.g. Drive and Swift. For details of which remotes support which hash type see the table in the overview section.

Eg rclone --checksum sync s3:/bucket swift:/bucket would run much quicker than without the --checksum flag.

When using this flag, rclone won't update mtimes of remote files if they are incorrect as it would normally.

--color AUTO|NEVER|ALWAYS

Specify when colors (and other ANSI codes) should be added to the output.

AUTO only allows ANSI codes when the output is a terminal. This is the default.

NEVER never allow ANSI codes.

ALWAYS always add ANSI codes, regardless of the output format (terminal or file).

--compare-dest stringArray

When using sync, copy or move, the specified paths are checked in addition to the destination for files. If a file identical to the source is found, that file is not copied from source. This is useful to copy just files that have changed since the last backup.

You must use the same remote as the destination of the sync. The compare directory must not overlap the destination directory.

See --copy-dest and --backup-dir.

--config string

Specify the location of the rclone configuration file, to override the default. E.g. rclone config --config="rclone.conf".

The exact default is a bit complex to describe, due to changes introduced through different versions of rclone while preserving backwards compatibility, but in most cases it is as simple as:

  • %APPDATA%/rclone/rclone.conf on Windows
  • ~/.config/rclone/rclone.conf on other

The complete logic is as follows: Rclone will look for an existing configuration file in any of the following locations, in priority order:

  1. rclone.conf (in program directory, where rclone executable is)
  2. %APPDATA%/rclone/rclone.conf (only on Windows)
  3. $XDG_CONFIG_HOME/rclone/rclone.conf (on all systems, including Windows)
  4. ~/.config/rclone/rclone.conf (see below for explanation of ~ symbol)
  5. ~/.rclone.conf

If no existing configuration file is found, then a new one will be created in the following location:

  • On Windows: Location 2 listed above, except in the unlikely event that APPDATA is not defined, then location 4 is used instead.
  • On Unix: Location 3 if XDG_CONFIG_HOME is defined, else location 4.
  • Fallback to location 5 (on all OS), when the rclone directory cannot be created, but if also a home directory was not found then path .rclone.conf relative to current working directory will be used as a final resort.

The ~ symbol in paths above represent the home directory of the current user on any OS, and the value is defined as following:

  • On Windows: %HOME% if defined, else %USERPROFILE%, or else %HOMEDRIVE%\%HOMEPATH%.
  • On Unix: $HOME if defined, else by looking up current user in OS-specific user database (e.g. passwd file), or else use the result from shell command cd && pwd.

If you run rclone config file you will see where the default location is for you. Running rclone config touch will ensure a configuration file exists, creating an empty one in the default location if there is none.

The fact that an existing file rclone.conf in the same directory as the rclone executable is always preferred, means that it is easy to run in "portable" mode by downloading rclone executable to a writable directory and then create an empty file rclone.conf in the same directory.

If the location is set to empty string "" or path to a file with name notfound, or the os null device represented by value NUL on Windows and /dev/null on Unix systems, then rclone will keep the configuration file in memory only.

You may see a log message "Config file not found - using defaults" if there is no configuration file. This can be suppressed, e.g. if you are using rclone entirely with on the fly remotes, by using memory-only configuration file or by creating an empty configuration file, as described above.

The file format is basic INI: Sections of text, led by a [section] header and followed by key=value entries on separate lines. In rclone each remote is represented by its own section, where the section name defines the name of the remote. Options are specified as the key=value entries, where the key is the option name without the --backend- prefix, in lowercase and with _ instead of -. E.g. option --mega-hard-delete corresponds to key hard_delete. Only backend options can be specified. A special, and required, key type identifies the storage system, where the value is the internal lowercase name as returned by command rclone help backends. Comments are indicated by ; or # at the beginning of a line.

Example:

[megaremote]
type = mega
user = [email protected]
pass = PDPcQVVjVtzFY-GTdDFozqBhTdsPg3qH

Note that passwords are in obscured form. Also, many storage systems uses token-based authentication instead of passwords, and this requires additional steps. It is easier, and safer, to use the interactive command rclone config instead of manually editing the configuration file.

The configuration file will typically contain login information, and should therefore have restricted permissions so that only the current user can read it. Rclone tries to ensure this when it writes the file. You may also choose to encrypt the file.

When token-based authentication are used, the configuration file must be writable, because rclone needs to update the tokens inside it.

To reduce risk of corrupting an existing configuration file, rclone will not write directly to it when saving changes. Instead it will first write to a new, temporary, file. If a configuration file already existed, it will (on Unix systems) try to mirror its permissions to the new file. Then it will rename the existing file to a temporary name as backup. Next, rclone will rename the new file to the correct name, before finally cleaning up by deleting the backup file.

If the configuration file path used by rclone is a symbolic link, then this will be evaluated and rclone will write to the resolved path, instead of overwriting the symbolic link. Temporary files used in the process (described above) will be written to the same parent directory as that of the resolved configuration file, but if this directory is also a symbolic link it will not be resolved and the temporary files will be written to the location of the directory symbolic link.

--contimeout Duration

Set the connection timeout. This should be in go time format which looks like 5s for 5 seconds, 10m for 10 minutes, or 3h30m.

The connection timeout is the amount of time rclone will wait for a connection to go through to a remote object storage system. It is 1m by default.

--copy-dest stringArray

When using sync, copy or move, the specified paths are checked in addition to the destination for files. This part is the same as --compare-dest, but the difference is that with --copy-dest, if a file identical to the source is found, that file is server-side copied from the specified paths to the destination. This is useful for incremental backup.

The remote in use must support server-side copy and you must use the same remote as the destination of the sync. The compare directory must not overlap the destination directory.

See --compare-dest and --backup-dir.

--dedupe-mode interactive|skip|first|newest|oldest|largest|smallest|rename|list

Mode to run dedupe command in. One of interactive, skip, first, newest, oldest, largest, smallest, rename list. The default is interactive. See the dedupe command for more information as to what these options mean.

--default-time Time

If a file or directory does not have a modification time rclone can read then rclone will display this fixed time instead.

The default is 2000-01-01 00:00:00 UTC. This can be configured in any of the ways shown in time options.

For example --default-time 2020-06-01 to set the default time to the 1st of June 2020 or --default-time 0s to set the default time to the time rclone started up.

--disable string

This disables a comma separated list of optional features. For example to disable server-side move and server-side copy use:

--disable move,copy

The features can be put in any case.

To see a list of which features can be disabled use:

--disable help

The features a remote has can be seen in JSON format with:

rclone backend features remote:

See the overview features and optional features to get an idea of which feature does what.

Note that some features can be set to true if they are true/false feature flag features by prefixing them with !. For example the CaseInsensitive feature can be forced to false with --disable CaseInsensitive and forced to true with --disable '!CaseInsensitive'. In general it isn't a good idea doing this but it may be useful in extremis.

(Note that ! is a shell command which you will need to escape with single quotes or a backslash on unix like platforms.)

This flag can be useful for debugging and in exceptional circumstances (e.g. Google Drive limiting the total volume of server-side copies to 100 GiB/day).

--disable-http2

This stops rclone from trying to use HTTP/2 if available. This can sometimes speed up transfers due to a problem in the Go standard library.

--dscp string

Specify a DSCP value or name to use in connections. This could help QoS system to identify traffic class. BE, EF, DF, LE, CSx and AFxx are allowed.

See the description of differentiated services to get an idea of this field. Setting this to 1 (LE) to identify the flow to SCAVENGER class can avoid occupying too much bandwidth in a network with DiffServ support (RFC 8622).

For example, if you configured QoS on router to handle LE properly. Running:

rclone copy --dscp LE from:/from to:/to

would make the priority lower than usual internet flows.

This option has no effect on Windows (see golang/go#42728).

-n, --dry-run

Do a trial run with no permanent changes. Use this to see what rclone would do without actually doing it. Useful when setting up the sync command which deletes files in the destination.

--expect-continue-timeout Duration

This specifies the amount of time to wait for a server's first response headers after fully writing the request headers if the request has an "Expect: 100-continue" header. Not all backends support using this.

Zero means no timeout and causes the body to be sent immediately, without waiting for the server to approve. This time does not include the time to send the request header.

The default is 1s. Set to 0 to disable.

--error-on-no-transfer

By default, rclone will exit with return code 0 if there were no errors.

This option allows rclone to return exit code 9 if no files were transferred between the source and destination. This allows using rclone in scripts, and triggering follow-on actions if data was copied, or skipping if not.

NB: Enabling this option turns a usually non-fatal error into a potentially fatal one - please check and adjust your scripts accordingly!

--fix-case

Normally, a sync to a case insensitive dest (such as macOS / Windows) will not result in a matching filename if the source and dest filenames have casing differences but are otherwise identical. For example, syncing hello.txt to HELLO.txt will normally result in the dest filename remaining HELLO.txt. If --fix-case is set, then HELLO.txt will be renamed to hello.txt to match the source.

NB:

  • directory names with incorrect casing will also be fixed
  • --fix-case will be ignored if --immutable is set
  • using --local-case-sensitive instead is not advisable; it will cause HELLO.txt to get deleted!
  • the old dest filename must not be excluded by filters. Be especially careful with --files-from, which does not respect --ignore-case!
  • on remotes that do not support server-side move, --fix-case will require downloading the file and re-uploading it. To avoid this, do not use --fix-case.

--fs-cache-expire-duration Duration

When using rclone via the API rclone caches created remotes for 5 minutes by default in the "fs cache". This means that if you do repeated actions on the same remote then rclone won't have to build it again from scratch, which makes it more efficient.

This flag sets the time that the remotes are cached for. If you set it to 0 (or negative) then rclone won't cache the remotes at all.

Note that if you use some flags, eg --backup-dir and if this is set to 0 rclone may build two remotes (one for the source or destination and one for the --backup-dir where it may have only built one before.

--fs-cache-expire-interval Duration

This controls how often rclone checks for cached remotes to expire. See the --fs-cache-expire-duration documentation above for more info. The default is 60s, set to 0 to disable expiry.

--header stringArray

Add an HTTP header for all transactions. The flag can be repeated to add multiple headers.

If you want to add headers only for uploads use --header-upload and if you want to add headers only for downloads use --header-download.

This flag is supported for all HTTP based backends even those not supported by --header-upload and --header-download so may be used as a workaround for those with care.

The headers are only sent to the host of the original request. If the server redirects to another host (including a subdomain or a different port), or redirects an https request to http, the headers are not sent to it, or to any further hop in that redirect chain. Rclone logs this at debug level (-vv).

rclone ls remote:test --header "X-Rclone: Foo" --header "X-LetMeIn: Yes"

--header-download stringArray

Add an HTTP header for all download transactions. The flag can be repeated to add multiple headers.

rclone sync --interactive s3:test/src ~/dst --header-download "X-Amz-Meta-Test: Foo" --header-download "X-Amz-Meta-Test2: Bar"

See GitHub issue #59 for currently supported backends.

--header-upload stringArray

Add an HTTP header for all upload transactions. The flag can be repeated to add multiple headers.

rclone sync --interactive ~/src s3:test/dst --header-upload "Content-Disposition: attachment; filename='cool.html'" --header-upload "X-Amz-Meta-Test: FooBar"

See GitHub issue #59 for currently supported backends.

--http-proxy string

Use this option to set an HTTP proxy for all HTTP based services to use.

Rclone also supports the standard HTTP proxy environment variables which it will pick up automatically. The is the way the HTTP proxy will normally be set but this flag can be used to override it.

--human-readable

Rclone commands output values for sizes (e.g. number of bytes) and counts (e.g. number of files) either as raw numbers, or in human-readable format.

In human-readable format the values are scaled to larger units, indicated with a suffix shown after the value, and rounded to three decimals. Rclone consistently uses binary units (powers of 2) for sizes and decimal units (powers of 10) for counts. The unit prefix for size is according to IEC standard notation, e.g. Ki for kibi. Used with byte unit, 1 KiB means 1024 Byte. In list type of output, only the unit prefix appended to the value (e.g. 9.762Ki), while in more textual output the full unit is shown (e.g. 9.762 KiB). For counts the SI standard notation is used, e.g. prefix k for kilo. Used with file counts, 1k means 1000 files.

The various list commands output raw numbers by default. Option --human-readable will make them output values in human-readable format instead (with the short unit prefix).

The about command outputs human-readable by default, with a command-specific option --full to output the raw numbers instead.

Command size outputs both human-readable and raw numbers in the same output.

The tree command also considers --human-readable, but it will not use the exact same notation as the other commands: It rounds to one decimal, and uses single letter suffix, e.g. K instead of Ki. The reason for this is that it relies on an external library.

The interactive command ncdu shows human-readable by default, and responds to key u for toggling human-readable format.

--ignore-case-sync

Using this option will cause rclone to ignore the case of the files when synchronizing so files will not be copied/synced when the existing filenames are the same, even if the casing is different.

--ignore-checksum

Normally rclone will check that the checksums of transferred files match, and give an error "corrupted on transfer" if they don't.

You can use this option to skip that check. You should only use it if you have had the "corrupted on transfer" error message and you are sure you might want to transfer potentially corrupted data.

--ignore-existing

Using this option will make rclone unconditionally skip all files that exist on the destination, no matter the content of these files.

While this isn't a generally recommended option, it can be useful in cases where your files change due to encryption. However, it cannot correct partial transfers in case a transfer was interrupted.

When performing a move/moveto command, this flag will leave skipped files in the source location unchanged when a file with the same name exists on the destination.

--ignore-size

Normally rclone will look at modification time and size of files to see if they are equal. If you set this flag then rclone will check only the modification time. If --checksum is set then it only checks the checksum.

It will also cause rclone to skip verifying the sizes are the same after transfer.

This can be useful for transferring files to and from OneDrive which occasionally misreports the size of image files (see #399 for more info).

-I, --ignore-times

Using this option will cause rclone to unconditionally upload all files regardless of the state of files on the destination.

Normally rclone would skip any files that have the same modification time and are the same size (or have the same checksum if using --checksum).

--immutable

Treat source and destination files as immutable and disallow modification.

With this option set, files will be created and deleted as requested, but existing files will never be updated. If an existing file does not match between the source and destination, rclone will give the error Source and destination exist but do not match: immutable file modified.

Note that only commands which transfer files (e.g. sync, copy or move) are affected by this behavior, and only modification is disallowed. Files may still be deleted explicitly (e.g. delete, purge) or implicitly (e.g. sync, move). Use copy --immutable if it is desired to avoid deletion as well as modification.

This can be useful as an additional layer of protection for immutable or append-only data sets (notably backup archives), where modification implies corruption and should not be propagated.

--inplace {#inplace}

The --inplace flag changes the behaviour of rclone when uploading files to some backends (backends with the PartialUploads feature flag set) such as:

  • local
  • ftp
  • sftp
  • pcloud

Without --inplace (the default) rclone will first upload to a temporary file with an extension like this, where XXXXXX represents a hash of the source file's fingerprint and .partial is --partial-suffix value (.partial by default).

original-file-name.XXXXXX.partial

(rclone will make sure the final name is no longer than 100 characters by truncating the original-file-name part if necessary).

When the upload is complete, rclone will rename the .partial file to the correct name, overwriting any existing file at that point. If the upload fails then the .partial file will be deleted.

This prevents other users of the backend from seeing partially uploaded files in their new names and prevents overwriting the old file until the new one is completely uploaded.

If the --inplace flag is supplied, rclone will upload directly to the final name without creating a .partial file.

This means that an incomplete file will be visible in the directory listings while the upload is in progress and any existing files will be overwritten as soon as the upload starts. If the transfer fails then the file will be deleted. This can cause data loss of the existing file if the transfer fails.

Note that on the local file system if you don't use --inplace hard links (Unix only) will be broken. And if you do use --inplace you won't be able to update in use executables.

On macOS, if rclone clones a file (see --local-no-clone) and the destination already exists, the existing file is replaced with a new one, even with --inplace, so hard links to it will be broken. Use --local-no-clone to keep them.

Note also that versions of rclone prior to v1.63.0 behave as if the --inplace flag is always supplied.

-i, --interactive {#interactive}

This flag can be used to tell rclone that you wish a manual confirmation before destructive operations.

It is recommended that you use this flag while learning rclone especially with rclone sync.

For example

$ rclone delete --interactive /tmp/dir
rclone: delete "important-file.txt"?
y) Yes, this is OK (default)
n) No, skip this
s) Skip all delete operations with no more questions
!) Do all delete operations with no more questions
q) Exit rclone now.
y/n/s/!/q> n

The options mean

  • y: Yes, this operation should go ahead. You can also press Return for this to happen. You'll be asked every time unless you choose s or !.
  • n: No, do not do this operation. You'll be asked every time unless you choose s or !.
  • s: Skip all the following operations of this type with no more questions. This takes effect until rclone exits. If there are any different kind of operations you'll be prompted for them.
  • !: Do all the following operations with no more questions. Useful if you've decided that you don't mind rclone doing that kind of operation. This takes effect until rclone exits . If there are any different kind of operations you'll be prompted for them.
  • q: Quit rclone now, just in case!

--leave-root

During rmdirs it will not remove root directory, even if it's empty.

-l, --links

Normally rclone will ignore symlinks or junction points (which behave like symlinks under Windows). Ignored files won't be copied, moved or deleted in a sync.

If you supply this flag then rclone will copy symbolic links from any supported backend, and store them as text files, with a .rclonelink suffix in the destination.

The text file will contain the target of the symbolic link.

The --links / -l flag enables this feature for all supported backends and the VFS. There are individual flags for just enabling it for the VFS --vfs-links and the local backend --local-links if required.

--list-cutoff int {#list-cutoff}

When syncing rclone needs to sort directory entries before comparing them. Below this threshold (1,000,000) by default, rclone will store the directory entries in memory. 1,000,000 entries will take approx 1GB of RAM to store. Above this threshold rclone will store directory entries on disk and sort them without using a lot of memory.

Doing this is slightly less efficient then sorting them in memory and will only work well for the bucket based backends (eg s3, b2, azureblob, swift) but these are the only backends likely to have millions of entries in a directory.

--log-file string

Log all of rclone's output to a file. This is not active by default. This can be useful for tracking down problems with syncs in combination with the -v flag. See the logging section for more info.

If the file exists, then rclone will append to it.

Note that if you are using the logrotate program to manage rclone's logs, then you should use the copytruncate option as rclone doesn't have a signal to rotate logs.

Alternatively you can use the options below to manage rclone's built in log rotation.

--log-file-max-size SizeSuffix

Maximum size of the log file before it's rotated (eg 10M). This SizeSuffix is rounded to the nearest MiB or 1 MiB if lower.

If --log-file is not set then this option will be ignored.

If this option is not set, then the other log rotation options will be ignored.

For example if the following flags are in use

rclone --log-file rclone.log --log-file-max-size 1M --log-file-max-backups 3

Then this will create log files which look like this

$ ls -l
-rw-------  1 user user  1048491 Apr 11 17:15 rclone-2025-04-11T17-15-29.998.log
-rw-------  1 user user  1048511 Apr 11 17:15 rclone-2025-04-11T17-15-30.467.log
-rw-------  1 user user  1048559 Apr 11 17:15 rclone-2025-04-11T17-15-30.543.log
-rw-------  1 user user   521602 Apr 11 17:15 rclone.log

The file rclone.log being the current one.

--log-file-compress

If set, compress rotated log files using gzip. This changes the extension of the old log files to .log.gz.

Defaults to false - don't compress log files.

--log-file-max-age Duration

Maximum duration to retain old log files (eg 7d). This is rounded to the dearest day, or 1 day if lower.

The default is to retain all old log files.

--log-file-max-backups int

Maximum number of old log files to retain

The default is to retain all old log files.

--log-format string

Comma separated list of log format options. The accepted options are:

  • date - Add a date in the format YYYY/MM/YY to the log.
  • time - Add a time to the log in format HH:MM:SS.
  • microseconds - Add microseconds to the time in format HH:MM:SS.SSSSSS.
  • UTC - Make the logs in UTC not localtime.
  • longfile - Adds the source file and line number of the log statement.
  • shortfile - Adds the source file and line number of the log statement.
  • pid - Add the process ID to the log - useful with rclone mount --daemon.
  • nolevel - Don't add the level to the log.
  • json - Equivalent to adding --use-json-log

They are added to the log line in the order above.

The default log format is "date,time".

--log-level LogLevel

This sets the log level for rclone. The default log level is NOTICE.

DEBUG is equivalent to -vv. It outputs lots of debug info - useful for bug reports and really finding out what rclone is doing.

INFO is equivalent to -v. It outputs information about each transfer and prints stats once a minute by default.

NOTICE is the default log level if no logging flags are supplied. It outputs very little when things are working normally. It outputs warnings and significant events.

ERROR is equivalent to -q. It only outputs error messages.

See also the logging section.

--windows-event-log-level LogLevel

If this is configured (the default is OFF) then logs of this level and above will be logged to the Windows event log in addition to the normal logs. These will be logged in JSON format as described below regardless of what format the main logs are configured for.

The Windows event log only has 3 levels of severity Info, Warning and Error. If enabled we map rclone levels like this.

  • Error ← ERROR (and above)
  • Warning ← WARNING (note that this level is defined but not currently used).
  • Info ← NOTICE, INFO and DEBUG.

Rclone will declare its log source as "rclone" if it is has enough permissions to create the registry key needed. If not then logs will appear as "Application". You can run rclone version --windows-event-log-level DEBUG once as administrator to create the registry key in advance.

Note that the --windows-event-log-level level must be greater (more severe) than or equal to the --log-level. For example to log DEBUG to a log file but ERRORs to the event log you would use

--log-file rclone.log --log-level DEBUG --windows-event-log-level ERROR

This option is only supported Windows platforms.

--use-json-log

This switches the log format to JSON. The log messages are then streamed as individual JSON objects, with fields: level, msg, source, and time. The resulting format is what is sometimes referred to as newline-delimited JSON (NDJSON), or JSON Lines (JSONL). This is well suited for processing by traditional line-oriented tools and shell pipelines, but a complete log file is not strictly valid JSON and needs a parser that can handle it.

The JSON logs will be printed on a single line, but are shown expanded here for clarity.

{
  "time": "2025-05-13T17:30:51.036237518+01:00",
  "level": "debug",
  "msg": "4 go routines active\n",
  "source": "cmd/cmd.go:298"
}

Completed data transfer logs will have extra size information. Logs which are about a particular object will have object and objectType fields also.

{
  "time": "2025-05-13T17:38:05.540846352+01:00",
  "level": "info",
  "msg": "Copied (new) to: file2.txt",
  "size": 6,
  "object": "file.txt",
  "objectType": "*local.Object",
  "source": "operations/copy.go:368"
}

Stats logs will contain a stats field which is the same as returned from the rc call core/stats.

{
  "time": "2025-05-13T17:38:05.540912847+01:00",
  "level": "info",
  "msg": "...text version of the stats...",
  "stats": {
    "bytes": 6,
    "checks": 0,
    "deletedDirs": 0,
    "deletes": 0,
    "elapsedTime": 0.000904825,
    ...truncated for clarity...
    "totalBytes": 6,
    "totalChecks": 0,
    "totalTransfers": 1,
    "transferTime": 0.000882794,
    "transfers": 1
  },
  "source": "accounting/stats.go:569"
}

--low-level-retries int

This controls the number of low level retries rclone does.

A low level retry is used to retry a failing operation - typically one HTTP request. This might be uploading a chunk of a big file for example. You will see low level retries in the log with the -v flag.

This shouldn't need to be changed from the default in normal operations. However, if you get a lot of low level retries you may wish to reduce the value so rclone moves on to a high level retry (see the --retries flag) quicker.

Disable low level retries with --low-level-retries 1.

--max-backlog int

This is the maximum allowable backlog of files in a sync/copy/move queued for being checked or transferred.

This can be set arbitrarily large. It will only use memory when the queue is in use. Note that it will use in the order of N KiB of memory when the backlog is in use.

Setting this large allows rclone to calculate how many files are pending more accurately, give a more accurate estimated finish time and make --order-by work more accurately.

Setting this small will make rclone more synchronous to the listings of the remote which may be desirable.

Setting this to a negative number will make the backlog as large as possible.

--max-buffer-memory SizeSuffix {#max-buffer-memory}

If set, don't allocate more than given amount of memory as buffers. If not set or set to 0 or off this will not limit the amount of memory in use.

This includes memory used by buffers created by the --buffer flag and buffers used by multi-thread transfers.

Most multi-thread transfers do not take additional memory, but some do depending on the backend (eg the s3 backend for uploads). This means there is a tension between total setting --transfers as high as possible and memory use.

Setting --max-buffer-memory allows the buffer memory to be controlled so that it doesn't overwhelm the machine and allows --transfers to be set large.

--max-connections int

This sets the maximum number of concurrent calls to the backend API. It may not map 1:1 to TCP or HTTP connections depending on the backend in use and the use of HTTP1 vs HTTP2.

When downloading files, backends only limit the initial opening of the stream. The bulk data download is not counted as a connection. This means that the --max--connections flag won't limit the total number of downloads.

Note that it is possible to cause deadlocks with this setting so it should be used with care.

If you are doing a sync or copy then make sure --max-connections is one more than the sum of --transfers and --checkers.

If you use --check-first then --max-connections just needs to be one more than the maximum of --checkers and --transfers.

So for --max-connections 3 you'd use --checkers 2 --transfers 2 --check-first or --checkers 1 --transfers 1.

Setting this flag can be useful for backends which do multipart uploads to limit the number of simultaneous parts being transferred.

--max-delete int

This tells rclone not to delete more than N files. If that limit is exceeded then a fatal error will be generated and rclone will stop the operation in progress.

--max-delete-size SizeSuffix

Rclone will stop deleting files when the total size of deletions has reached the size specified. It defaults to off.

If that limit is exceeded then a fatal error will be generated and rclone will stop the operation in progress.

--max-depth int

This modifies the recursion depth for all the commands except purge.

So if you do rclone --max-depth 1 ls remote:path you will see only the files in the top level directory. Using --max-depth 2 means you will see all the files in first two directory levels and so on.

For historical reasons the lsd command defaults to using a --max-depth of 1 - you can override this with the command line flag.

You can use this command to disable recursion (with --max-depth 1).

Note that if you use this with sync and --delete-excluded the files not recursed through are considered excluded and will be deleted on the destination. Test first with --dry-run if you are not sure what will happen.

--max-duration Duration

Rclone will stop transferring when it has run for the duration specified. Defaults to off.

When the limit is reached all transfers will stop immediately. Use --cutoff-mode to modify this behaviour.

Rclone will exit with exit code 10 if the duration limit is reached.

--max-transfer SizeSuffix

Rclone will stop transferring when it has reached the size specified. Defaults to off.

When the limit is reached all transfers will stop immediately. Use --cutoff-mode to modify this behaviour.

Rclone will exit with exit code 8 if the transfer limit is reached.

--cutoff-mode HARD|SOFT|CAUTIOUS

Configure the behavior of --max-transfer and --max-duration.

HARD will stop transferring immediately when rclone reaches the limit. This is the default.

SOFT will stop starting new transfers when rclone reaches the limit.

CAUTIOUS will try to prevent rclone from reaching the limit. Only applicable for --max-transfer.

-M, --metadata

Setting this flag enables rclone to copy the metadata from the source to the destination. For local backends this is ownership, permissions, xattr etc. See the metadata section for more info.

--metadata-mapper SpaceSepList {#metadata-mapper}

If you supply the parameter --metadata-mapper /path/to/program then rclone will use that program to map metadata from source object to destination object.

The argument to this flag should be a command with an optional space separated list of arguments. If one of the arguments has a space in then enclose it in ", if you want a literal " in an argument then enclose the argument in " and double the ". See CSV encoding for more info.

--metadata-mapper "python bin/test_metadata_mapper.py"
--metadata-mapper 'python bin/test_metadata_mapper.py "argument with a space"'
--metadata-mapper 'python bin/test_metadata_mapper.py "argument with ""two"" quotes"'

This uses a simple JSON based protocol with input on STDIN and output on STDOUT. This will be called for every file and directory copied and may be called concurrently.

The program's job is to take a metadata blob on the input and turn it into a metadata blob on the output suitable for the destination backend.

Input to the program (via STDIN) might look like this. This provides some context for the Metadata which may be important.

  • SrcFs is the config string for the remote that the object is currently on.
  • SrcFsType is the name of the source backend.
  • DstFs is the config string for the remote that the object is being copied to
  • DstFsType is the name of the destination backend.
  • Remote is the path of the object relative to the root.
  • Size, MimeType, ModTime are attributes of the object.
  • IsDir is true if this is a directory (not yet implemented).
  • ID is the source ID of the object if known.
  • Metadata is the backend specific metadata as described in the backend docs.
{
  "SrcFs": "gdrive:",
  "SrcFsType": "drive",
  "DstFs": "newdrive:user",
  "DstFsType": "onedrive",
  "Remote": "test.txt",
  "Size": 6,
  "MimeType": "text/plain; charset=utf-8",
  "ModTime": "2022-10-11T17:53:10.286745272+01:00",
  "IsDir": false,
  "ID": "xyz",
  "Metadata": {
    "btime": "2022-10-11T16:53:11Z",
    "content-type": "text/plain; charset=utf-8",
    "mtime": "2022-10-11T17:53:10.286745272+01:00",
    "owner": "[email protected]",
    "permissions": "...",
    "description": "my nice file",
    "starred": "false"
  }
}

The program should then modify the input as desired and send it to STDOUT. The returned Metadata field will be used in its entirety for the destination object. Any other fields will be ignored. Note in this example we translate user names and permissions and add something to the description:

{
  "Metadata": {
    "btime": "2022-10-11T16:53:11Z",
    "content-type": "text/plain; charset=utf-8",
    "mtime": "2022-10-11T17:53:10.286745272+01:00",
    "owner": "[email protected]",
    "permissions": "...",
    "description": "my nice file [migrated from domain1]",
    "starred": "false"
  }
}

Metadata can be removed here too.

An example python program might look something like this to implement the above transformations.

import sys, json

i = json.load(sys.stdin)
metadata = i["Metadata"]
# Add tag to description
if "description" in metadata:
    metadata["description"] += " [migrated from domain1]"
else:
    metadata["description"] = "[migrated from domain1]"
# Modify owner
if "owner" in metadata:
    metadata["owner"] = metadata["owner"].replace("domain1.com", "domain2.com")
o = { "Metadata": metadata }
json.dump(o, sys.stdout, indent="\t")

You can find this example (slightly expanded) in the rclone source code at bin/test_metadata_mapper.py.

If you want to see the input to the metadata mapper and the output returned from it in the log you can use -vv --dump mapper.

See the metadata section for more info.

--metadata-set stringArray

Specify value as string in format key=value to add metadata key with value value when uploading. This can be repeated as many times as required. See the metadata section for more info.

--modify-window Duration

When checking whether a file has been modified, this is the maximum allowed time difference that a file can have and still be considered equivalent.

The default is 1ns unless this is overridden by a remote. For example OS X only stores modification times to the nearest second so if you are reading and writing to an OS X filing system this will be 1s by default.

This command line flag allows you to override that computed default.

--multi-thread-write-buffer-size SizeSuffix

When transferring with multiple threads, rclone will buffer the specified number of bytes in memory before writing to disk for each thread.

This can improve performance if the underlying filesystem does not deal well with a lot of small writes in different positions of the file, so if you see transfers being limited by disk write speed, you might want to experiment with different values. Specially for magnetic drives and remote file systems a higher value can be useful.

Nevertheless, the default of 128k should be fine for almost all use cases, so before changing it ensure that network is not really your bottleneck.

As a final hint, size is not the only factor: block size (or similar concept) can have an impact. In one case, we observed that exact multiples of 16k performed much better than other values.

--multi-thread-chunk-size SizeSuffix

Normally the chunk size for multi thread transfers is set by the backend. However some backends such as local and smb (which implement OpenWriterAt but not OpenChunkWriter) don't have a natural chunk size.

In this case the value of this option is used (default 64Mi).

--multi-thread-cutoff SizeSuffix {#multi-thread-cutoff}

When transferring files above specified size to capable backends, rclone will use multiple threads to transfer the file (default 256M).

Capable backends are marked in the overview as MultithreadUpload. (They need to implement either the OpenWriterAt or OpenChunkWriter internal interfaces). These include local, s3, azureblob, b2, oracleobjectstorage and smb at the time of writing.

On the local disk, rclone preallocates the file (using fallocate(FALLOC_FL_KEEP_SIZE) on unix or NTSetInformationFile on Windows both of which takes no time) then each thread writes directly into the file at the correct place. This means that rclone won't create fragmented or sparse files and there won't be any assembly time at the end of the transfer.

The number of threads used to transfer is controlled by --multi-thread-streams.

Use -vv if you wish to see info about the threads.

This will work with the sync, copy and move commands, and friends copyto, moveto. Multi thread transfers will be used with rclone mount and rclone serve if --vfs-cache-mode is set to writes or above.

Most multi-thread transfers do not take additional memory, but some do (for example uploading to s3). In the worst case memory usage can be at maximum --transfers * --multi-thread-chunk-size * --multi-thread-streams or specifically for the s3 backend --transfers * --s3-chunk-size * --s3-concurrency. However you can use the --max-buffer-memory flag to control the maximum memory used here.

NB that this only works with supported backends as the destination but will work with any backend as the source.

NB that multi-thread copies are disabled for local to local copies as they are faster without unless --multi-thread-streams is set explicitly.

NB on Windows using multi-thread transfers to the local disk will cause the resulting files to be sparse. Use --local-no-sparse to disable sparse files (which may cause long delays at the start of transfers) or disable multi-thread transfers with --multi-thread-streams 0

--multi-thread-streams int

When using multi thread transfers (see above --multi-thread-cutoff) this sets the number of streams to use. Set to 0 to disable multi thread transfers (Default 4).

If the backend has a --backend-upload-concurrency setting (eg --s3-upload-concurrency) then this setting will be used as the number of transfers instead if it is larger than the value of --multi-thread-streams or --multi-thread-streams isn't set.

--name-transform stringArray

--name-transform introduces path name transformations for rclone copy, rclone sync, and rclone move. These transformations enable modifications to source and destination file names by applying prefixes, suffixes, and other alterations during transfer operations. For detailed docs and examples, see convmv.

--no-check-dest

The --no-check-dest can be used with move or copy and it causes rclone not to check the destination at all when copying files.

This means that:

  • the destination is not listed minimising the API calls
  • files are always transferred
  • this can cause duplicates on remotes which allow it (e.g. Google Drive)
  • --retries 1 is recommended otherwise you'll transfer everything again on a retry

This flag is useful to minimise the transactions if you know that none of the files are on the destination.

This is a specialized flag which should be ignored by most users!

--no-gzip-encoding

Don't set Accept-Encoding: gzip. This means that rclone won't ask the server for compressed files automatically. Useful if you've set the server to return files with Content-Encoding: gzip but you uploaded compressed files.

There is no need to set this in normal operation, and doing so will decrease the network transfer efficiency of rclone.

--no-traverse

The --no-traverse flag controls whether the destination file system is traversed when using the copy or move commands. --no-traverse is not compatible with sync and will be ignored if you supply it with sync.

If you are only copying a small number of files (or are filtering most of the files) and/or have a large number of files on the destination then --no-traverse will stop rclone listing the destination and save time.

However, if you are copying a large number of files, especially if you are doing a copy where lots of the files under consideration haven't changed and won't need copying then you shouldn't use --no-traverse.

See rclone copy for an example of how to use it.

--no-unicode-normalization

Don't normalize unicode characters in filenames during the sync routine.

Sometimes, an operating system will store filenames containing unicode parts in their decomposed form (particularly macOS). Some cloud storage systems will then recompose the unicode, resulting in duplicate files if the data is ever copied back to a local filesystem.

Using this flag will disable that functionality, treating each unicode character as unique. For example, by default é and é will be normalized into the same character. With --no-unicode-normalization they will be treated as unique characters.

--no-update-modtime

When using this flag, rclone won't update modification times of remote files if they are incorrect as it would normally.

This can be used if the remote is being synced with another tool also (e.g. the Google Drive client).

--no-update-dir-modtime

When using this flag, rclone won't update modification times of remote directories if they are incorrect as it would normally.

--order-by string

The --order-by flag controls the order in which files in the backlog are processed in rclone sync, rclone copy and rclone move.

The order by string is constructed like this. The first part describes what aspect is being measured:

  • size - order by the size of the files
  • name - order by the full path of the files
  • modtime - order by the modification date of the files
  • pattern - order by a list of path glob patterns

The size, name and modtime comparisons can have a modifier appended with a comma:

  • ascending or asc - order so that the smallest (or oldest) is processed first
  • descending or desc - order so that the largest (or newest) is processed first
  • mixed - order so that the smallest is processed first for some threads and the largest for others

If the modifier is mixed then it can have an optional percentage (which defaults to 50), e.g. size,mixed,25 which means that 25% of the threads should be taking the smallest items and 75% the largest. The threads which take the smallest first will always take the smallest first and likewise the largest first threads. The mixed mode can be useful to minimise the transfer time when you are transferring a mixture of large and small files - the large files are guaranteed upload threads and bandwidth and the small files will be processed continuously.

If no modifier is supplied then the order is ascending.

For pattern, append one or more comma-separated patterns. Patterns use the filter pattern syntax. Files matching the first pattern are processed first, followed by files matching the second pattern, and so on. Files which do not match a pattern are processed last. Within each group, files are ordered by full path. Patterns cannot contain a comma.

For example

  • --order-by size,desc - send the largest files first
  • --order-by modtime,ascending - send the oldest files first
  • --order-by name - send the files with alphabetically by path first
  • --order-by 'pattern,*.tar.gz,*.md5' - send archives before checksum files

If the --order-by flag is not supplied or it is supplied with an empty string then the default ordering will be used which is as scanned. With --checkers 1 this is mostly alphabetical, however with the default --checkers 8 it is somewhat random.

Limitations

The --order-by flag does not do a separate pass over the data. This means that it may transfer some files out of the order specified if

  • there are no files in the backlog or the source has not been fully scanned yet
  • there are more than --max-backlog files in the backlog

Rclone will do its best to transfer the best file it has so in practice this should not cause a problem. Think of --order-by as being more of a best efforts flag rather than a perfect ordering.

If you want perfect ordering then you will need to specify --check-first which will find all the files which need transferring first before transferring any.

--partial-suffix string {#partial-suffix}

When --inplace is not used, it causes rclone to use the --partial-suffix as suffix for temporary files.

Suffix length limit is 16 characters.

The default is .partial.

--password-command SpaceSepList {#password-command}

This flag supplies a program which should supply the config password when run. This is an alternative to rclone prompting for the password or setting the RCLONE_CONFIG_PASS variable. It is also used when setting the config password for the first time.

The argument to this should be a command with a space separated list of arguments. If one of the arguments has a space in then enclose it in ", if you want a literal " in an argument then enclose the argument in " and double the ". This includes the command itself: if the path to the executable contains a space, it must be quoted too. See CSV encoding for more info.

Eg

--password-command "echo hello"
--password-command 'echo "hello with space"'
--password-command 'echo "hello with ""quotes"" and space"'
--password-command '"/path with a space/get-password.sh"'

Note that when changing the configuration password the environment variable RCLONE_PASSWORD_CHANGE=1 will be set. This can be used to distinguish initial decryption of the config file from the new password.

See the Configuration Encryption for more info.

See a Windows PowerShell example on the Wiki.

-P, --progress

This flag makes rclone update the stats in a static block in the terminal providing a realtime overview of the transfer.

Any log messages will scroll above the static block. Log messages will push the static block down to the bottom of the terminal where it will stay.

Normally this is updated every 500mS but this period can be overridden with the --stats flag.

This can be used with the --stats-one-line flag for a simpler display.

To change the display length of filenames (for different terminal widths), see the --stats-file-name-length option. The default output is formatted for 80 character wide terminals.

Note: On Windows until this bug is fixed all non-ASCII characters will be replaced with . when --progress is in use.

--progress-terminal-title

This flag, when used with -P/--progress, will print the string ETA: %s to the terminal title.

-q, --quiet

This flag will limit rclone's output to error messages only.

--refresh-times

The --refresh-times flag can be used to update modification times of existing files when they are out of sync on backends which don't support hashes.

This is useful if you uploaded files with the incorrect timestamps and you now wish to correct them.

This flag is only useful for destinations which don't support hashes (e.g. crypt).

This can be used any of the sync commands sync, copy or move.

To use this flag you will need to be doing a modification time sync (so not using --size-only or --checksum). The flag will have no effect when using --size-only or --checksum.

If this flag is used when rclone comes to upload a file it will check to see if there is an existing file on the destination. If this file matches the source with size (and checksum if available) but has a differing timestamp then instead of re-uploading it, rclone will update the timestamp on the destination file. If the checksum does not match rclone will upload the new file. If the checksum is absent (e.g. on a crypt backend) then rclone will update the timestamp.

Note that some remotes can't set the modification time without re-uploading the file so this flag is less useful on them.

Normally if you are doing a modification time sync rclone will update modification times without --refresh-times provided that the remote supports checksums and the checksums match on the file. However if the checksums are absent then rclone will upload the file rather than setting the timestamp as this is the safe behaviour.

--retries int

Retry the entire sync if it fails this many times it fails (default 3).

Some remotes can be unreliable and a few retries help pick up the files which didn't get transferred because of errors.

Disable retries with --retries 1.

--retries-sleep Duration

This sets the interval between each retry specified by --retries

The default is 0. Use 0 to disable.

--server-side-across-configs

Allow server-side operations (e.g. copy or move) to work across different configurations.

This can be useful if you wish to do a server-side copy or move between two remotes which use the same backend but are configured differently.

Note that this isn't enabled by default because it isn't easy for rclone to tell if it will work between any two configurations.

--size-only

Normally rclone will look at modification time and size of files to see if they are equal. If you set this flag then rclone will check only the size.

This can be useful transferring files from Dropbox which have been modified by the desktop sync client which doesn't set checksums of modification times in the same way as rclone.

--stats Duration

Commands which transfer data (sync, copy, copyto, move, moveto) will print data transfer stats at regular intervals to show their progress.

This sets the interval.

The default is 1m. Use 0 to disable.

If you set the stats interval then all commands can show stats. This can be useful when running other commands, check or mount for example.

Stats are logged at INFO level by default which means they won't show at default log level NOTICE. Use --stats-log-level NOTICE or -v to make them show. See the logging section for more info on log levels.

Note that on macOS you can send a SIGINFO (which is normally ctrl-T in the terminal) to make the stats print immediately.

--stats-file-name-length int

By default, the --stats output will truncate file names and paths longer than 40 characters. This is equivalent to providing --stats-file-name-length 40. Use --stats-file-name-length 0 to disable any truncation of file names printed by stats.

--stats-log-level LogLevel

Log level to show --stats output at. This can be DEBUG, INFO, NOTICE, or ERROR. The default is INFO. This means at the default level of logging, which is NOTICE, the stats won't show - if you want them to then use --stats-log-level NOTICE. See the logging section for more details on log levels.

--stats-one-line

When this is specified, rclone condenses the stats into a single line showing the most important stats only.

--stats-one-line-date

When this is specified, rclone enables the single-line stats and prepends

the display with a date string. The default is 2006/01/02 15:04:05 -

--stats-one-line-date-format string

When this is specified, rclone enables the single-line stats and prepends the display with a user-supplied date string. The date string MUST be enclosed in quotes. Follow golang specs for date formatting syntax.

--stats-unit string

By default, data transfer rates will be printed in bytes per second, corresponding to --stats-unit bytes.

This option allows the data rate to be printed in bits per second, by specifying --stats-unit bits. Data transfer volume will still be reported in bytes.

The rate is reported as a binary unit, not SI unit. So 1 Mbit/s equals 1,048,576 bit/s and not 1,000,000 bit/s.

--suffix string

When using sync, copy or move any files which would have been overwritten or deleted will have the suffix added to them. If there is a file with the same path (after the suffix has been added), then it will be overwritten.

The remote in use must support server-side move or copy and you must use the same remote as the destination of the sync.

This is for use with files to add the suffix in the current directory or with --backup-dir. See --backup-dir for more info.

For example

rclone copy --interactive /path/to/local/file remote:current --suffix .bak

will copy /path/to/local to remote:current, but for any files which would have been updated or deleted have .bak added.

If using rclone sync with --suffix and without --backup-dir then it is recommended to put a filter rule in excluding the suffix otherwise the sync will delete the backup files.

rclone sync --interactive /path/to/local/file remote:current --suffix .bak --exclude "*.bak"

--suffix-keep-extension

When using --suffix, setting this causes rclone put the SUFFIX before the extension of the files that it backs up rather than after.

So let's say we had --suffix -2019-01-01, without the flag file.txt would be backed up to file.txt-2019-01-01 and with the flag it would be backed up to file-2019-01-01.txt. This can be helpful to make sure the suffixed files can still be opened.

If a file has two (or more) extensions and the second (or subsequent) extension is recognised as a valid mime type, then the suffix will go before that extension. So file.tar.gz would be backed up to file-2019-01-01.tar.gz whereas file.badextension.gz would be backed up to file.badextension-2019-01-01.gz.

--syslog

On capable OSes (not Windows or Plan9) send all log output to syslog.

This can be useful for running rclone in a script or rclone mount.

--syslog-facility string

If using --syslog this sets the syslog facility (e.g. KERN, USER). See man syslog for a list of possible facilities. The default facility is DAEMON.

--temp-dir string

Specify the directory rclone will use for temporary files, to override the default. Make sure the directory exists and have accessible permissions.

By default the operating system's temp directory will be used:

  • On Unix systems, $TMPDIR if non-empty, else /tmp.
  • On Windows, the first non-empty value from %TMP%, %TEMP%, %USERPROFILE%, or the Windows directory.

When overriding the default with this option, the specified path will be set as value of environment variable TMPDIR on Unix systems and TMP and TEMP on Windows.

You can use the config paths command to see the current value.

--tpslimit float

Limit transactions per second to this number. Default is 0 which is used to mean unlimited transactions per second.

A transaction is roughly defined as an API call; its exact meaning will depend on the backend. For HTTP based backends it is an HTTP PUT/GET/POST/etc and its response. For FTP/SFTP it is a round trip transaction over TCP.

For example, to limit rclone to 10 transactions per second use --tpslimit 10, or to 1 transaction every 2 seconds use --tpslimit 0.5.

Use this when the number of transactions per second from rclone is causing a problem with the cloud storage provider (e.g. getting you banned or rate limited).

This can be very useful for rclone mount to control the behaviour of applications using it.

This limit applies to all HTTP based backends and to the FTP and SFTP backends. It does not apply to the local backend or the Storj backend.

See also --tpslimit-burst.

--tpslimit-burst int

Max burst of transactions for --tpslimit (default 1).

Normally --tpslimit will do exactly the number of transaction per second specified. However if you supply --tps-burst then rclone can save up some transactions from when it was idle giving a burst of up to the parameter supplied.

For example if you provide --tpslimit-burst 10 then if rclone has been idle for more than 10*--tpslimit then it can do 10 transactions very quickly before they are limited again.

This may be used to increase performance of --tpslimit without changing the long term average number of transactions per second.

--track-renames

By default, rclone doesn't keep track of renamed files, so if you rename a file locally then sync it to a remote, rclone will delete the old file on the remote and upload a new copy.

An rclone sync with --track-renames runs like a normal sync, but keeps track of objects which exist in the destination but not in the source (which would normally be deleted), and which objects exist in the source but not the destination (which would normally be transferred). These objects are then candidates for renaming.

After the sync, rclone matches up the source only and destination only objects using the --track-renames-strategy specified and either renames the destination object or transfers the source and deletes the destination object. --track-renames is stateless like all of rclone's syncs.

To use this flag the destination must support server-side copy or server-side move, and to use a hash based --track-renames-strategy (the default) the source and the destination must have a compatible hash.

If the destination does not support server-side copy or move, rclone will fall back to the default behaviour and log an error level message to the console.

Encrypted destinations are not currently supported by --track-renames if --track-renames-strategy includes hash.

Note that --track-renames is incompatible with --no-traverse and that it uses extra memory to keep track of all the rename candidates.

Note also that --track-renames is incompatible with --delete-before and will select --delete-after instead of --delete-during.

--track-renames-strategy string

This option changes the file matching criteria for --track-renames.

The matching is controlled by a comma separated selection of these tokens:

  • modtime - the modification time of the file - not supported on all backends
  • hash - the hash of the file contents - not supported on all backends
  • leaf - the name of the file not including its directory name
  • size - the size of the file (this is always enabled)

The default option is hash.

Using --track-renames-strategy modtime,leaf would match files based on modification time, the leaf of the file name and the size only.

Using --track-renames-strategy modtime or leaf can enable --track-renames support for encrypted destinations.

Note that the hash strategy is not supported with encrypted destinations.

--delete-(before,during,after)

This option allows you to specify when files on your destination are deleted when you sync folders.

Specifying the value --delete-before will delete all files present on the destination, but not on the source before starting the transfer of any new or updated files. This uses two passes through the file systems, one for the deletions and one for the copies.

Specifying --delete-during will delete files while checking and uploading files. This is the fastest option and uses the least memory.

Specifying --delete-after (the default value) will delay deletion of files until all new/updated files have been successfully transferred. The files to be deleted are collected in the copy pass then deleted after the copy pass has completed successfully. The files to be deleted are held in memory so this mode may use more memory. This is the safest mode as it will only delete files if there have been no errors subsequent to that. If there have been errors before the deletions start then you will get the message not deleting files as there were IO errors.

--fast-list

When doing anything which involves a directory listing (e.g. sync, copy, ls - in fact nearly every command), rclone has different strategies to choose from.

The basic strategy is to list one directory and processes it before using more directory lists to process any subdirectories. This is a mandatory backend feature, called List, which means it is supported by all backends. This strategy uses small amount of memory, and because it can be parallelised it is fast for operations involving processing of the list results.

Some backends provide the support for an alternative strategy, where all files beneath a directory can be listed in one (or a small number) of transactions. Rclone supports this alternative strategy through an optional backend feature called ListR. You can see in the storage system overview documentation's optional features section which backends it is enabled for (these tend to be the bucket-based ones, e.g. S3, B2, GCS, Swift). This strategy requires fewer transactions for highly recursive operations, which is important on backends where this is charged or heavily rate limited. It may be faster (due to fewer transactions) or slower (because it can't be parallelized) depending on different parameters, and may require more memory if rclone has to keep the whole listing in memory.

Which listing strategy rclone picks for a given operation is complicated, but in general it tries to choose the best possible. It will prefer ListR in situations where it doesn't need to store the listed files in memory, e.g. for unlimited recursive ls command variants. In other situations it will prefer List, e.g. for sync and copy, where it needs to keep the listed files in memory, and is performing operations on them where parallelization may be a huge advantage.

Rclone is not able to take all relevant parameters into account for deciding the best strategy, and therefore allows you to influence the choice in two ways: You can stop rclone from using ListR by disabling the feature, using the --disable option (--disable ListR), or you can allow rclone to use ListR where it would normally choose not to do so due to higher memory usage, using the --fast-list option. Rclone should always produce identical results either way. Using --disable ListR or --fast-list on a remote which doesn't support ListR does nothing, rclone will just ignore it.

A rule of thumb is that if you pay for transactions and can fit your entire sync listing into memory, then --fast-list is recommended. If you have a very big sync to do, then don't use --fast-list, otherwise you will run out of memory. Run some tests and compare before you decide, and if in doubt then just leave the default, let rclone decide, i.e. not use --fast-list.

--timeout Duration

This sets the IO idle timeout. If a transfer has started but then becomes idle for this long it is considered broken and disconnected.

The default is 5m. Set to 0 to disable.

--transfers int

The number of file transfers to run in parallel. It can sometimes be useful to set this to a smaller number if the remote is giving a lot of timeouts or bigger if you have lots of bandwidth and a fast remote.

The default is to run 4 file transfers in parallel.

Look at --multi-thread-streams if you would like to control single file transfers.

-u, --update

This forces rclone to skip any files which exist on the destination and have a modified time that is newer than the source file.

This can be useful in avoiding needless transfers when transferring to a remote which doesn't support modification times directly (or when using --use-server-modtime to avoid extra API calls) as it is more accurate than a --size-only check and faster than using --checksum. On such remotes (or when using --use-server-modtime) the time checked will be the uploaded time.

If an existing destination file has a modification time older than the source file's, it will be updated if the sizes are different. If the sizes are the same, it will be updated if the checksum is different or not available.

If an existing destination file has a modification time equal (within the computed modify window) to the source file's, it will be updated if the sizes are different. The checksum will not be checked in this case unless the --checksum flag is provided.

In all other cases the file will not be updated.

Consider using the --modify-window flag to compensate for time skews between the source and the backend, for backends that do not support mod times, and instead use uploaded times. However, if the backend does not support checksums, note that syncing or copying within the time skew window may still result in additional transfers for safety.

--use-mmap

If this flag is set then rclone will use anonymous memory allocated by mmap on Unix based platforms and VirtualAlloc on Windows for its transfer buffers (size controlled by --buffer-size). Memory allocated like this does not go on the Go heap and can be returned to the OS immediately when it is finished with.

If this flag is not set then rclone will allocate and free the buffers using the Go memory allocator which may use more memory as memory pages are returned less aggressively to the OS.

It is possible this does not work well on all platforms so it is disabled by default; in the future it may be enabled by default.

--use-server-modtime

Some object-store backends (e.g, Swift, S3) do not preserve file modification times (modtime). On these backends, rclone stores the original modtime as additional metadata on the object. By default it will make an API call to retrieve the metadata when the modtime is needed by an operation.

Use this flag to disable the extra API call and rely instead on the server's modified time. In cases such as a local to remote sync using --update, knowing the local file is newer than the time it was last uploaded to the remote is sufficient. In those cases, this flag can speed up the process and reduce the number of API calls necessary.

This flag is only supported on certain backends and will be silently ignored on unsupported backends. Supported backends include azureblob, oracleobjectstorage, s3, swift.

Using this flag on a sync operation without also using --update would cause all files modified at any time other than the last upload time to be uploaded again, which is probably not what you want.

-v, -vv, --verbose

With -v rclone will tell you about each file that is transferred and a small number of significant events.

With -vv rclone will become very verbose telling you about every file it considers and transfers. Please send bug reports with a log with this setting.

When setting verbosity as an environment variable, use RCLONE_VERBOSE=1 or RCLONE_VERBOSE=2 for -v and -vv respectively.

-V, --version

Prints the version number

SSL/TLS options

The outgoing SSL/TLS connections rclone makes can be controlled with these options. For example this can be very useful with the HTTP or WebDAV backends. Rclone HTTP servers have their own set of configuration for SSL/TLS which you can find in their documentation.

--ca-cert stringArray

This loads the PEM encoded certificate authority certificates and uses it to verify the certificates of the servers rclone connects to.

If you have generated certificates signed with a local CA then you will need this flag to connect to servers using those certificates.

--client-cert string

This loads the PEM encoded client side certificate.

This is used for mutual TLS authentication.

The --client-key flag is required too when using this.

--client-key string

This loads the PEM encoded client side private key used for mutual TLS authentication. Used in conjunction with --client-cert.

Supported types are:

  • Unencrypted PKCS#1 ("BEGIN RSA PRIVATE KEY")
  • Unencrypted PKCS#8 ("BEGIN PRIVATE KEY")
  • Encrypted PKCS#8 ("BEGIN ENCRYPTED PRIVATE KEY")
  • Legacy PEM encryption (e.g., DEK-Info headers), which are automatically detected.

--client-pass string

This can be used to supply an optional password to decrypt the client key file.

NB the password should be obscured so it should be the output of rclone obscure YOURPASSWORD.

--no-check-certificate

--no-check-certificate controls whether a client verifies the server's certificate chain and host name. If --no-check-certificate is true, TLS accepts any certificate presented by the server and any host name in that certificate. In this mode, TLS is susceptible to man-in-the-middle attacks.

This option defaults to false.

This should be used only for testing.

Configuration encryption

Your configuration file contains information for logging in to your cloud services. This means that you should keep your rclone.conf file in a secure location.

If you are in an environment where that isn't possible, you can add a password to your configuration. This means that you will have to supply the password every time you start rclone.

To add a password to your rclone configuration, execute rclone config.

$ rclone config
Current remotes:

e) Edit existing remote
n) New remote
d) Delete remote
s) Set configuration password
q) Quit config
e/n/d/s/q>

Go into s, Set configuration password:

e/n/d/s/q> s
Your configuration is not encrypted.
If you add a password, you will protect your login information to cloud services.
a) Add Password
q) Quit to main menu
a/q> a
Enter NEW configuration password:
password:
Confirm NEW password:
password:
Password set
Your configuration is encrypted.
c) Change Password
u) Unencrypt configuration
q) Quit to main menu
c/u/q>

Your configuration is now encrypted, and every time you start rclone you will have to supply the password. See below for details. In the same menu, you can change the password or completely remove encryption from your configuration.

There is no way to recover the configuration if you lose your password.

You can also use

rclone uses nacl secretbox which in turn uses XSalsa20 and Poly1305 to encrypt and authenticate your configuration with secret-key cryptography. The password is SHA-256 hashed, which produces the key for secretbox. The hashed password is not stored.

While this provides very good security, we do not recommend storing your encrypted rclone configuration in public if it contains sensitive information, maybe except if you use a very strong password.

If it is safe in your environment, you can set the RCLONE_CONFIG_PASS environment variable to contain your password, in which case it will be used for decrypting the configuration.

You can set this for a session from a script. For unix like systems save this to a file called set-rclone-password:

#!/bin/echo Source this file don't run it

read -s RCLONE_CONFIG_PASS
export RCLONE_CONFIG_PASS

Then source the file when you want to use it. From the shell you would do source set-rclone-password. It will then ask you for the password and set it in the environment variable.

An alternate means of supplying the password is to provide a script which will retrieve the password and print on standard output. This script should have a fully specified path name and not rely on any environment variables. The script is supplied either via --password-command="..." command line argument or via the RCLONE_PASSWORD_COMMAND environment variable.

One useful example of this is using the passwordstore application to retrieve the password:

export RCLONE_PASSWORD_COMMAND="pass rclone/config"

If the passwordstore password manager holds the password for the rclone configuration, using the script method means the password is primarily protected by the passwordstore system, and is never embedded in the clear in scripts, nor available for examination using the standard commands available. It is quite possible with long running rclone sessions for copies of passwords to be innocently captured in log files or terminal scroll buffers, etc. Using the script method of supplying the password enhances the security of the config password considerably.

If you are running rclone inside a script, unless you are using the --password-command method, you might want to disable password prompts. To do that, pass the parameter --ask-password=false to rclone. This will make rclone fail instead of asking for a password if RCLONE_CONFIG_PASS doesn't contain a valid password, and --password-command has not been supplied.

Whenever running commands that may be affected by options in a configuration file, rclone will look for an existing file according to the rules described above, and load any it finds. If an encrypted file is found, this includes decrypting it, with the possible consequence of a password prompt. When executing a command line that you know are not actually using anything from such a configuration file, you can avoid it being loaded by overriding the location, e.g. with one of the documented special values for memory-only configuration. Since only backend options can be stored in configuration files, this is normally unnecessary for commands that do not operate on backends, e.g. completion. However, it will be relevant for commands that do operate on backends in general, but are used without referencing a stored remote, e.g. listing local filesystem paths, or connection strings: rclone --config="" ls .

Configuration encryption cheatsheet

You can quickly apply a configuration encryption without plain-text at rest or transfer. Detailed instructions for popular OSes:

Mac

  • Generate and store a password

    security add-generic-password -a rclone -s config -w $(openssl rand -base64 40)
    
  • Add the retrieval instruction to your .zprofile / .profile

    export RCLONE_PASSWORD_COMMAND="/usr/bin/security find-generic-password -a rclone -s config -w"
    

Linux

  • Prerequisite: Linux doesn't come with a default password manager. Let's install the "pass" utility using a package manager, e.g. apt install pass, yum install pass, etc.; then initialize a password store: pass init rclone.

  • Generate and store a password

    echo $(openssl rand -base64 40) | pass insert -m rclone/config
    
  • Add the retrieval instruction

    export RCLONE_PASSWORD_COMMAND="/usr/bin/pass rclone/config"
    

Windows

  • Generate and store a password

    New-Object -TypeName PSCredential -ArgumentList "rclone", (ConvertTo-SecureString -String ([System.Web.Security.Membership]::GeneratePassword(40, 10)) -AsPlainText -Force) | Export-Clixml -Path "rclone-credential.xml"
    
  • Add the password retrieval instruction

    [Environment]::SetEnvironmentVariable("RCLONE_PASSWORD_COMMAND", "[System.Runtime.InteropServices.Marshal]::PtrToStringAuto([System.Runtime.InteropServices.Marshal]::SecureStringToBSTR((Import-Clixml -Path "rclone-credential.xml").Password))")
    

Encrypt the config file (all systems)

  • Execute rclone config, and select option s) Set configuration password

  • Add/update the password from previous steps

Developer options

These options are useful when developing or debugging rclone. There are also some more remote specific options which aren't documented here which are used for testing. These start with remote name e.g. --drive-test-option - see the docs for the remote in question.

--cpuprofile string

Write CPU profile to a file. This can be analysed with go tool pprof.

--memprofile string

Write memory profile to a file. This can be analysed with go tool pprof.

--dump DumpFlags

The --dump flag takes a comma separated list of flags to dump info about.

Note that some headers, such as Accept-Encoding, may not be correct as shown in the request, and the response may not show Content-Encoding if the go standard libraries auto gzip encoding was in effect. In this case the body of the request will be gunzipped before showing it.

The available flags are:

  • headers dumps HTTP headers. Any Authorization: headers will be excluded, but output may still contain sensitive information. Can be very verbose. Useful for debugging only. Use auth if you do want the Authorization: headers.
  • auth dumps HTTP headers like headers, but also includes any Authorization: headers. This means the output will probably contain sensitive information. Use headers to dump without Authorization: headers. Can be very verbose. Useful for debugging only. This flag also makes the debug log of the config process (e.g. rclone config -vv) show answers to questions, passwords and tokens which are otherwise redacted.
  • bodies dumps HTTP headers and bodies. May contain sensitive info. Can be very verbose. Useful for debugging only. Note that the bodies are buffered in memory so don't use this for enormous files.
  • requests is similar to bodies, but dumps the request bodies and the response headers. Useful for debugging download problems.
  • responses is similar to bodies, but dumps the response bodies and the request headers. Useful for debugging upload problems.
  • filters dumps the filters. Useful to see exactly what include and exclude options are filtering on.
  • goroutines dumps a list of the running go-routines at the end of the command.
  • openfiles dumps a list of the open files at the end of the command. It uses the lsof Unix command to do that, so you'll need that installed to use it.
  • mapper dumps the JSON blobs being sent to the program supplied with --metadata-mapper and received from it. It can be useful for debugging the metadata mapper interface.
  • curl dumps the HTTP request as a curl command. Can be used with the other HTTP debugging flags (e.g. requests, bodies). By default the auth will be masked - use with auth to have the curl commands with authentication too.
  • errors only dumps the HTTP transactions which fail with a retryable error - that is a transport error, an HTTP 429 (too many requests) or an HTTP 5xx server error. The other dump flags control what is dumped, so for example use --dump errors,bodies to dump the headers and bodies of failed transactions only. On its own --dump errors dumps the headers. This lets you capture first-failure diagnostics without the noise of dumping every transaction. May contain sensitive info.
  • trace logs connection level events for each HTTP transaction using Go's net/http/httptrace - DNS resolution, TCP connect, TLS handshake (including the negotiated TLS version, cipher, ALPN protocol and server certificate), connection reuse, request write and time to first response byte. Each line is tagged with the time elapsed since the start of the transaction so you can see where the time goes. This is complementary to the other dump flags: it shows how the connection behaved rather than what was sent, so it is useful for debugging connectivity, DNS, TLS, proxy and keep-alive problems. It does not dump headers or bodies - combine it with headers or bodies for that.

Filtering

For the filtering options

  • --delete-excluded
  • --filter
  • --filter-from
  • --exclude
  • --exclude-from
  • --exclude-if-present
  • --include
  • --include-from
  • --files-from
  • --files-from-raw
  • --files-from0
  • --min-size
  • --max-size
  • --min-age
  • --max-age
  • --hash-filter
  • --dump filters
  • --metadata-include
  • --metadata-include-from
  • --metadata-exclude
  • --metadata-exclude-from
  • --metadata-filter
  • --metadata-filter-from

See the filtering section.

Remote control

For the remote control options and for instructions on how to remote control rclone:

  • --rc
  • Anything starting with --rc-

See the remote control section.

Logging

rclone has 4 levels of logging, ERROR, NOTICE, INFO and DEBUG.

By default, rclone logs to standard error. This means you can redirect standard error and still see the normal output of rclone commands (e.g. rclone ls).

By default, rclone will produce Error and Notice level messages.

If you use the -q flag, rclone will only produce Error messages.

If you use the -v flag, rclone will produce Error, Notice and Info messages.

If you use the -vv flag, rclone will produce Error, Notice, Info and Debug messages.

You can also control the log levels with the --log-level flag.

If you use the --log-file option, rclone will redirect Error, Info and Debug messages along with standard error to a file.

If you use the --syslog flag then rclone will log to syslog and the --syslog-facility control which facility it uses.

Rclone prefixes all log messages with their level in capitals, e.g. INFO which makes it easy to grep the log file for different kinds of information.

Metrics

Rclone can publish metrics in the OpenMetrics/Prometheus format.

To enable the metrics endpoint, use the --metrics-addr flag. Metrics can also be published on the --rc-addr port if the --rc flag and --rc-enable-metrics flags are supplied or if using rclone rcd --rc-enable-metrics

Rclone provides extensive configuration options for the metrics HTTP endpoint. These settings are grouped under the Metrics section and have a prefix --metrics-*.

When metrics are enabled with --rc-enable-metrics, they will be published on the same port as the rc API. In this case, the --metrics-* flags will be ignored, and the HTTP endpoint configuration will be managed by the --rc-* parameters.

Exit code

If any errors occur during the command execution, rclone will exit with a non-zero exit code. This allows scripts to detect when rclone operations have failed.

During the startup phase, rclone will exit immediately if an error is detected in the configuration. There will always be a log message immediately before exiting.

When rclone is running it will accumulate errors as it goes along, and only exit with a non-zero exit code if (after retries) there were still failed transfers. For every error counted there will be a high priority log message (visible with -q) showing the message and which file caused the problem. A high priority message is also shown when starting a retry so the user can see that any previous error messages may not be valid after the retry. If rclone has done a retry it will log a high priority message if the retry was successful.

List of exit codes

  • 0 - Success
  • 1 - Error not otherwise categorised
  • 2 - Syntax or usage error
  • 3 - Directory not found
  • 4 - File not found
  • 5 - Temporary error (one that more retries might fix) (Retry errors)
  • 6 - Less serious errors (like 461 errors from dropbox) (NoRetry errors)
  • 7 - Fatal error (one that more retries won't fix, like account suspended) (Fatal errors)
  • 8 - Transfer exceeded - limit set by --max-transfer reached
  • 9 - Operation successful, but no files transferred (Requires --error-on-no-transfer)
  • 10 - Duration exceeded - limit set by --max-duration reached

Environment variables

Rclone can be configured entirely using environment variables. These can be used to set defaults for options or config file entries.

Options

Every option in rclone can have its default set by environment variable.

To find the name of the environment variable, first, take the long option name, strip the leading --, change - to _, make upper case and prepend RCLONE_.

For example, to always set --stats 5s, set the environment variable RCLONE_STATS=5s. If you set stats on the command line this will override the environment variable setting.

Or to always use the trash in drive --drive-use-trash, set RCLONE_DRIVE_USE_TRASH=true.

Verbosity is slightly different, the environment variable equivalent of --verbose or -v is RCLONE_VERBOSE=1, or for -vv, RCLONE_VERBOSE=2.

The same parser is used for the options and the environment variables so they take exactly the same form.

The options set by environment variables can be seen with the -vv flag, e.g. rclone version -vv.

Options that can appear multiple times (type stringArray) are treated slightly differently as environment variables can only be defined once. In order to allow a simple mechanism for adding one or many items, the input is treated as a CSV encoded string. For example

Environment variable Equivalent options
RCLONE_EXCLUDE="*.jpg" --exclude "*.jpg"
RCLONE_EXCLUDE="*.jpg,*.png" --exclude "*.jpg" --exclude "*.png"
RCLONE_EXCLUDE='"*.jpg","*.png"' --exclude "*.jpg" --exclude "*.png"
RCLONE_EXCLUDE='"/directory with comma , in it /**"' `--exclude "/directory with comma , in it /**"

If stringArray options are defined as environment variables and options on the command line then all the values will be used.

Config file

You can set defaults for values in the config file on an individual remote basis. The names of the config items are documented in the page for each backend.

To find the name of the environment variable, you need to set, take RCLONE_CONFIG_ + name of remote + _ + name of config file option and make it all uppercase. Note that the name of the remote is only converted to uppercase - characters such as - and . are not changed to _ - so a remote called my-remote needs RCLONE_CONFIG_MY-REMOTE_TYPE. Many shells only allow letters, digits and _ (underscore) in environment variable names, so it is best to use only those characters in the name of a remote you want to configure this way.

For example, to configure an S3 remote named mys3: without a config file (using unix ways of setting environment variables):

$ export RCLONE_CONFIG_MYS3_TYPE=s3
$ export RCLONE_CONFIG_MYS3_ACCESS_KEY_ID=XXX
$ export RCLONE_CONFIG_MYS3_SECRET_ACCESS_KEY=XXX
$ rclone lsd mys3:
          -1 2016-09-21 12:54:21        -1 my-bucket
$ rclone listremotes | grep mys3
mys3:

Note that if you want to create a remote using environment variables you must create the ..._TYPE variable as above.

Note that the name of a remote created using environment variable is case insensitive, in contrast to regular remotes stored in config file as documented above. You must write the name in uppercase in the environment variable, but as seen from example above it will be listed and can be accessed in lowercase, while you can also refer to the same remote in uppercase:

$ rclone lsd mys3:
          -1 2016-09-21 12:54:21        -1 my-bucket
$ rclone lsd MYS3:
          -1 2016-09-21 12:54:21        -1 my-bucket

Note that you can only set the options of the immediate backend, so RCLONE_CONFIG_MYS3CRYPT_ACCESS_KEY_ID has no effect, if myS3Crypt is a crypt remote based on an S3 remote. However RCLONE_S3_ACCESS_KEY_ID will set the access key of all remotes using S3, including myS3Crypt.

Note also that now rclone has connection strings, it is probably easier to use those instead which makes the above example

rclone lsd :s3,access_key_id=XXX,secret_access_key=XXX:

Precedence

The various different methods of backend configuration are read in this order and the first one with a value is used.

  • Parameters in connection strings, e.g. myRemote,skip_links:
  • Flag values as supplied on the command line, e.g. --skip-links
  • Remote specific environment vars, e.g. RCLONE_CONFIG_MYREMOTE_SKIP_LINKS (see above).
  • Backend-specific environment vars, e.g. RCLONE_LOCAL_SKIP_LINKS.
  • Backend generic environment vars, e.g. RCLONE_SKIP_LINKS.
  • Config file, e.g. skip_links = true.
  • Default values, e.g. false - these can't be changed.

So if both --skip-links is supplied on the command line and an environment variable RCLONE_LOCAL_SKIP_LINKS is set, the command line flag will take preference.

The backend configurations set by environment variables can be seen with the -vv flag, e.g. rclone about myRemote: -vv.

For non backend configuration the order is as follows:

  • Flag values as supplied on the command line, e.g. --stats 5s.
  • Environment vars, e.g. RCLONE_STATS=5s.
  • Default values, e.g. 1m - these can't be changed.

Other environment variables

  • RCLONE_CONFIG_PASS set to contain your config file password (see Configuration Encryption section)
  • HTTP_PROXY, HTTPS_PROXY and NO_PROXY (or the lowercase versions thereof).
    • HTTPS_PROXY takes precedence over HTTP_PROXY for https requests.
    • The environment values may be either a complete URL or a "host[:port]" for, in which case the "http" scheme is assumed.
  • USER and LOGNAME values are used as fallbacks for current username. The primary method for looking up username is OS-specific: Windows API on Windows, real user ID in /etc/passwd on Unix systems. In the documentation the current username is simply referred to as $USER.
  • RCLONE_CONFIG_DIR - rclone sets this variable for use in config files and sub processes to point to the directory holding the config file.

The options set by environment variables can be seen with the -vv and --log-level=DEBUG flags, e.g. rclone version -vv.

Configuring rclone on a remote / headless machine

Some of the configurations (those involving oauth2) require an internet-connected web browser.

If you are trying to set rclone up on a remote or headless machine with no browser available on it (e.g. a NAS or a server in a datacenter), then you will need to use an alternative means of configuration. There are three ways of doing it, described below.

Configuring using rclone authorize

On the headless machine run rclone config, but answer N to the question Use web browser to automatically authenticate rclone with remote?.

Use web browser to automatically authenticate rclone with remote?
 * Say Y if the machine running rclone has a web browser you can use
 * Say N if running rclone on a (remote) machine without web browser access
If not sure try Y. If Y failed, try N.

y) Yes (default)
n) No
y/n> n

Option config_token.
For this to work, you will need rclone available on a machine that has
a web browser available.
For more help and alternate methods see: https://rclone.org/remote_setup/
Execute the following on the machine with the web browser (same rclone
version recommended):
        rclone authorize "onedrive"
Then paste the result.
Enter a value.
config_token>

Then on your main desktop machine, run rclone authorize.

rclone authorize "onedrive"
NOTICE: Make sure your Redirect URL is set to "http://localhost:53682/" in your custom config.
NOTICE: If your browser doesn't open automatically go to the following link: http://127.0.0.1:53682/auth?state=xxxxxxxxxxxxxxxxxxxxxx
NOTICE: Log in and authorize rclone for access
NOTICE: Waiting for code...

Got code
Paste the following into your remote machine --->
SECRET_TOKEN
<---End paste

Then back to the headless machine, paste in the code.

config_token> SECRET_TOKEN
--------------------
[acd12]
client_id =
client_secret =
token = SECRET_TOKEN
--------------------
y) Yes this is OK
e) Edit this remote
d) Delete this remote
y/e/d>

Configuring by copying the config file

Rclone stores all of its configuration in a single file. This can easily be copied to configure a remote rclone (although some backends do not support reusing the same configuration, consult your backend documentation to be sure).

Start by running rclone config to create the configuration file on your desktop machine.

rclone config

Then locate the file by running rclone config file.

$ rclone config file
Configuration file is stored at:
/home/user/.rclone.conf

Finally, transfer the file to the remote machine (scp, cut paste, ftp, sftp, etc.) and place it in the correct location (use rclone config file on the remote machine to find out where).

Configuring using SSH Tunnel

If you have an SSH client installed on your local machine, you can set up an SSH tunnel to redirect the port 53682 into the headless machine by using the following command:

ssh -L localhost:53682:localhost:53682 username@remote_server

Then on the headless machine run rclone config and answer Y to the question Use web browser to automatically authenticate rclone with remote?.

Use web browser to automatically authenticate rclone with remote?
 * Say Y if the machine running rclone has a web browser you can use
 * Say N if running rclone on a (remote) machine without web browser access
If not sure try Y. If Y failed, try N.

y) Yes (default)
n) No
y/n> y
NOTICE: Make sure your Redirect URL is set to "http://localhost:53682/" in your custom config.
NOTICE: If your browser doesn't open automatically go to the following link: http://127.0.0.1:53682/auth?state=xxxxxxxxxxxxxxxxxxxxxx
NOTICE: Log in and authorize rclone for access
NOTICE: Waiting for code...

Finally, copy and paste the presented URL http://127.0.0.1:53682/auth?state=xxxxxxxxxxxxxxxxxxxxxx to the browser on your local machine, complete the auth and you are done.

Filtering, includes and excludes

Filter flags determine which files rclone sync, move, ls, lsl, md5sum, sha1sum, size, delete, check and similar commands apply to.

They are specified in terms of path/file name patterns; path/file lists; file age and size, or presence of a file in a directory. Bucket based remotes without the concept of directory apply filters to object key, age and size in an analogous way.

Rclone purge does not obey filters.

To test filters without risk of damage to data, apply them to rclone ls, or with the --dry-run and -vv flags.

Rclone filter patterns can only be used in filter command line options, not in the specification of a remote.

E.g. rclone copy "remote:dir*.jpg" /path/to/dir does not have a filter effect. rclone copy remote:dir /path/to/dir --include "*.jpg" does.

Important Avoid mixing any two of --include..., --exclude... or --filter... flags in an rclone command. The results might not be what you expect. Instead use a --filter... flag.

Patterns for matching path/file names

Pattern syntax {#patterns}

Here is a formal definition of the pattern syntax, examples are below.

Rclone matching rules follow a glob style:

*         matches any sequence of non-separator (/) characters
**        matches any sequence of characters including / separators
?         matches any single non-separator (/) character
[ [ ! | ^ ] { character-range } ]
          character class (must be non-empty)
{ pattern-list }
          pattern alternatives
{{ regexp }}
          regular expression to match
c         matches character c (c != *, **, ?, \, [, {, })
\c        matches reserved character c (c = *, **, ?, \, [, {, }) or character class

character-range:

c         matches character c (c != \, -, ])
\c        matches reserved character c (c = \, -, ])
lo - hi   matches character c for lo <= c <= hi

pattern-list:

pattern { , pattern }
          comma-separated (without spaces) patterns

character classes (see Go regular expression reference) include:

Named character classes (e.g. [\d], [^\d], [\D], [^\D])
Perl character classes (e.g. \s, \S, \w, \W)
ASCII character classes (e.g. [[:alnum:]], [[:alpha:]], [[:punct:]], [[:xdigit:]])

regexp for advanced users to insert a regular expression - see below for more info:

Any re2 regular expression not containing `}}`

If the filter pattern starts with a / then it only matches at the top level of the directory tree, relative to the root of the remote (not necessarily the root of the drive). If it does not start with / then it is matched starting at the end of the path/file name but it only matches a complete path element - it must match from a / separator or the beginning of the path/file.

file.jpg   - matches "file.jpg"
           - matches "directory/file.jpg"
           - doesn't match "afile.jpg"
           - doesn't match "directory/afile.jpg"
/file.jpg  - matches "file.jpg" in the root directory of the remote
           - doesn't match "afile.jpg"
           - doesn't match "directory/file.jpg"

The top level of the remote might not be the top level of the drive.

E.g. for a Microsoft Windows local directory structure

F:
├── bkp
├── data
│   ├── excl
│   │   ├── 123.jpg
│   │   └── 456.jpg
│   ├── incl
│   │   └── document.pdf

To copy the contents of folder data into folder bkp excluding the contents of subfolder exclthe following command treats F:\data and F:\bkp as top level for filtering.

rclone copy F:\data\ F:\bkp\ --exclude=/excl/**

Important Use / in path/file name patterns and not \ even if running on Microsoft Windows.

Simple patterns are case sensitive unless the --ignore-case flag is used.

Without --ignore-case (default)

potato - matches "potato"
       - doesn't match "POTATO"

With --ignore-case

potato - matches "potato"
       - matches "POTATO"

Using regular expressions in filter patterns {#regexp}

The syntax of filter patterns is glob style matching (like bash uses) to make things easy for users. However this does not provide absolute control over the matching, so for advanced users rclone also provides a regular expression syntax.

Rclone generally accepts Perl-style regular expressions, the exact syntax is defined in the Go regular expression reference. Regular expressions should be enclosed in {{ }}. They will match only the last path segment if the glob doesn't start with / or the whole path name if it does. Note that rclone does not attempt to parse the supplied regular expression, meaning that using any regular expression filter will prevent rclone from using directory filter rules, as it will instead check every path against the supplied regular expression(s).

Here is how the {{regexp}} is transformed into an full regular expression to match the entire path:

{{regexp}}  becomes (^|/)(regexp)$
/{{regexp}} becomes ^(regexp)$

Regexp syntax can be mixed with glob syntax, for example

*.{{jpe?g}} to match file.jpg, file.jpeg but not file.png

You can also use regexp flags - to set case insensitive, for example

*.{{(?i)jpg}} to match file.jpg, file.JPG but not file.png

Be careful with wildcards in regular expressions - you don't want them to match path separators normally. To match any file name starting with start and ending with end write

{{start[^/]*end\.jpg}}

Not

{{start.*end\.jpg}}

Which will match a directory called start with a file called end.jpg in it as the .* will match / characters.

Note that you can use -vv --dump filters to show the filter patterns in regexp format - rclone implements the glob patterns by transforming them into regular expressions.

Filter pattern examples {#examples}

Description Pattern Matches Does not match
Wildcard *.jpg /file.jpg /file.png
/dir/file.jpg /dir/file.png
Rooted /*.jpg /file.jpg /file.png
/file2.jpg /dir/file.jpg
Alternates *.{jpg,png} /file.jpg /file.gif
/dir/file.png /dir/file.gif
Path Wildcard dir/** /dir/anyfile file.png
/subdir/dir/subsubdir/anyfile /subdir/file.png
Any Char *.t?t /file.txt /file.qxt
/dir/file.tzt /dir/file.png
Range *.[a-z] /file.a /file.0
/dir/file.b /dir/file.1
Escape *.\?\?\? /file.??? /file.abc
/dir/file.??? /dir/file.def
Class *.\d\d\d /file.012 /file.abc
/dir/file.345 /dir/file.def
Regexp *.{{jpe?g}} /file.jpeg /file.png
/dir/file.jpg /dir/file.jpeeg
Rooted Regexp /{{.*\.jpe?g}} /file.jpeg /file.png
/file.jpg /dir/file.jpg

How filter rules are applied to files {#how-filter-rules-work}

Rclone path/file name filters are made up of one or more of the following flags:

  • --include
  • --include-from
  • --exclude
  • --exclude-from
  • --filter
  • --filter-from

There can be more than one instance of individual flags.

Rclone internally uses a combined list of all the include and exclude rules. The order in which rules are processed can influence the result of the filter.

All flags of the same type are processed together in the order above, regardless of what order the different types of flags are included on the command line.

Multiple instances of the same flag are processed from left to right according to their position in the command line.

To mix up the order of processing includes and excludes use --filter... flags.

Within --include-from, --exclude-from and --filter-from flags rules are processed from top to bottom of the referenced file.

If there is an --include or --include-from flag specified, rclone implies a - ** rule which it adds to the bottom of the internal rule list. Specifying a + rule with a --filter... flag does not imply that rule.

Each path/file name passed through rclone is matched against the combined filter list. At first match to a rule the path/file name is included or excluded and no further filter rules are processed for that path/file.

If rclone does not find a match, after testing against all rules (including the implied rule if appropriate), the path/file name is included.

Any path/file included at that stage is processed by the rclone command.

--files-from, --files-from-raw and --files-from0 flags over-ride and cannot be combined with other filter options.

To see the internal combined rule list, in regular expression form, for a command add the --dump filters flag. Running an rclone command with --dump filters and -vv flags lists the internal filter elements and shows how they are applied to each source path/file. There is not currently a means provided to pass regular expression filter options into rclone directly though character class filter rules contain character classes. Go regular expression reference

How filter rules are applied to directories {#directory_filter}

Rclone commands are applied to path/file names not directories. The entire contents of a directory can be matched to a filter by the pattern directory/* or recursively by directory/**.

Directory filter rules are defined with a closing / separator.

E.g. /directory/subdirectory/ is an rclone directory filter rule.

Rclone commands can use directory filter rules to determine whether they recurse into subdirectories. This potentially optimises access to a remote by avoiding listing unnecessary directories. Whether optimisation is desirable depends on the specific filter rules and source remote content.

If any regular expression filters are in use, then no directory recursion optimisation is possible, as rclone must check every path against the supplied regular expression(s).

Directory recursion optimisation occurs if either:

  • A source remote does not support the rclone ListR primitive. local, sftp, Microsoft OneDrive and WebDAV do not support ListR. Google Drive and most bucket type storage do. Full list

  • On other remotes (those that support ListR), if the rclone command is not naturally recursive, and provided it is not run with the --fast-list flag. ls, lsf -R and size are naturally recursive but sync, copy and move are not.

  • Whenever the --disable ListR flag is applied to an rclone command.

Rclone commands imply directory filter rules from path/file filter rules. To view the directory filter rules rclone has implied for a command specify the --dump filters flag.

E.g. for an include rule

/a/*.jpg

Rclone implies the directory include rule

/a/

Directory filter rules specified in an rclone command can limit the scope of an rclone command but path/file filters still have to be specified.

E.g. rclone ls remote: --include /directory/ will not match any files. Because it is an --include option the --exclude ** rule is implied, and the /directory/ pattern serves only to optimise access to the remote by ignoring everything outside of that directory.

E.g. rclone ls remote: --filter-from filter-list.txt with a file filter-list.txt:

- /dir1/
- /dir2/
+ *.pdf
- **

All files in directories dir1 or dir2 or their subdirectories are completely excluded from the listing. Only files of suffix pdf in the root of remote: or its subdirectories are listed. The - ** rule prevents listing of any path/files not previously matched by the rules above.

Option exclude-if-present creates a directory exclude rule based on the presence of a file in a directory and takes precedence over other rclone directory filter rules.

When using pattern list syntax, if a pattern item contains either / or **, then rclone will not able to imply a directory filter rule from this pattern list.

E.g. for an include rule

{dir1/**,dir2/**}

Rclone will match files below directories dir1 or dir2 only, but will not be able to use this filter to exclude a directory dir3 from being traversed.

Directory recursion optimisation may affect performance, but normally not the result. One exception to this is sync operations with option --create-empty-src-dirs, where any traversed empty directories will be created. With the pattern list example {dir1/**,dir2/**} above, this would create an empty directory dir3 on destination (when it exists on source). Changing the filter to {dir1,dir2}/**, or splitting it into two include rules --include dir1/** --include dir2/**, will match the same files while also filtering directories, with the result that an empty directory dir3 will no longer be created.

--exclude - Exclude files matching pattern

Excludes path/file names from an rclone command based on a single exclude rule.

This flag can be repeated. See above for the order filter flags are processed in.

--exclude should not be used with --include, --include-from, --filter or --filter-from flags.

--exclude has no effect when combined with --files-from, --files-from-raw or --files-from0 flags.

E.g. rclone ls remote: --exclude *.bak excludes all .bak files from listing.

E.g. rclone size remote: --exclude "/dir/**" returns the total size of all files on remote: excluding those in root directory dir and sub directories.

E.g. on Microsoft Windows rclone ls remote: --exclude "*\[{JP,KR,HK}\]*" lists the files in remote: without [JP] or [KR] or [HK] in their name. Quotes prevent the shell from interpreting the \ characters.\ characters escape the [ and ] so an rclone filter treats them literally rather than as a character-range. The { and } define an rclone pattern list. For other operating systems single quotes are required ie rclone ls remote: --exclude '*\[{JP,KR,HK}\]*'

--exclude-from - Read exclude patterns from file

Excludes path/file names from an rclone command based on rules in a named file. The file contains a list of remarks and pattern rules.

For an example exclude-file.txt:

# a sample exclude rule file
*.bak
file2.jpg

rclone ls remote: --exclude-from exclude-file.txt lists the files on remote: except those named file2.jpg or with a suffix .bak. That is equivalent to rclone ls remote: --exclude file2.jpg --exclude "*.bak".

This flag can be repeated. See above for the order filter flags are processed in.

The --exclude-from flag is useful where multiple exclude filter rules are applied to an rclone command.

--exclude-from should not be used with --include, --include-from, --filter or --filter-from flags.

--exclude-from has no effect when combined with --files-from, --files-from-raw or --files-from0 flags.

--exclude-from followed by - reads filter rules from standard input.

--include - Include files matching pattern

Adds a single include rule based on path/file names to an rclone command.

This flag can be repeated. See above for the order filter flags are processed in.

--include has no effect when combined with --files-from, --files-from-raw or --files-from0 flags.

--include implies --exclude ** at the end of an rclone internal filter list. Therefore if you mix --include and --include-from flags with --exclude, --exclude-from, --filter or --filter-from, you must use include rules for all the files you want in the include statement. For more flexibility use the --filter-from flag.

E.g. rclone ls remote: --include "*.{png,jpg}" lists the files on remote: with suffix .png and .jpg. All other files are excluded.

E.g. multiple rclone copy commands can be combined with --include and a pattern-list.

rclone copy /vol1/A remote:A
rclone copy /vol1/B remote:B

is equivalent to:

rclone copy /vol1 remote: --include "{A,B}/**"

E.g. rclone ls remote:/wheat --include "??[^[:punct:]]*" lists the files remote: directory wheat (and subdirectories) whose third character is not punctuation. This example uses an ASCII character class.

--include-from - Read include patterns from file

Adds path/file names to an rclone command based on rules in a named file. The file contains a list of remarks and pattern rules.

For an example include-file.txt:

# a sample include rule file
*.jpg
file2.avi

rclone ls remote: --include-from include-file.txt lists the files on remote: with name file2.avi or suffix .jpg. That is equivalent to rclone ls remote: --include file2.avi --include "*.jpg".

This flag can be repeated. See above for the order filter flags are processed in.

The --include-from flag is useful where multiple include filter rules are applied to an rclone command.

--include-from implies --exclude ** at the end of an rclone internal filter list. Therefore if you mix --include and --include-from flags with --exclude, --exclude-from, --filter or --filter-from, you must use include rules for all the files you want in the include statement. For more flexibility use the --filter-from flag.

--include-from has no effect when combined with --files-from, --files-from-raw or --files-from0 flags.

--include-from followed by - reads filter rules from standard input.

--filter - Add a file-filtering rule

Specifies path/file names to an rclone command, based on a single include or exclude rule, in + or - format.

This flag can be repeated. See above for the order filter flags are processed in.

--filter + differs from --include. In the case of --include rclone implies an --exclude * rule which it adds to the bottom of the internal rule list. --filter...+ does not imply that rule.

--filter has no effect when combined with --files-from, --files-from-raw or --files-from0 flags.

--filter should not be used with --include, --include-from, --exclude or --exclude-from flags.

E.g. rclone ls remote: --filter "- *.bak" excludes all .bak files from a list of remote:.

--filter-from - Read filtering patterns from a file

Adds path/file names to an rclone command based on rules in a named file. The file contains a list of remarks and pattern rules. Include

rules start with + and exclude rules with - . ! clears existing rules. Rules are processed in the order they are defined.

This flag can be repeated. See above for the order filter flags are processed in.

Arrange the order of filter rules with the most restrictive first and work down.

Lines starting with # or ; are ignored, and can be used to write comments. Inline comments are not supported. Use -vv --dump filters to see how they appear in the final regexp.

E.g. for filter-file.txt:

# a sample filter rule file
- secret*.jpg
+ *.jpg
+ *.png
+ file2.avi
- /dir/tmp/** # WARNING! This text will be treated as part of the path.
- /dir/Trash/**
+ /dir/**
# exclude everything else
- *

rclone ls remote: --filter-from filter-file.txt lists the path/files on remote: including all jpg and png files, excluding any matching secret*.jpg and including file2.avi. It also includes everything in the directory dir at the root of remote, except remote:dir/Trash which it excludes. Everything else is excluded.

E.g. for an alternative filter-file.txt:

- secret*.jpg
+ *.jpg
+ *.png
+ file2.avi
- *

Files file1.jpg, file3.png and file2.avi are listed whilst secret17.jpg and files without the suffix .jpg or .png are excluded.

E.g. for an alternative filter-file.txt:

+ *.jpg
+ *.gif
!
+ 42.doc
- *

Only file 42.doc is listed. Prior rules are cleared by the !.

--files-from - Read list of source-file names

Adds path/files to an rclone command from a list in a named file. Rclone processes the path/file names in the order of the list, and no others.

Other filter flags (--include, --include-from, --exclude, --exclude-from, --filter and --filter-from) are ignored when --files-from, --files-from-raw or --files-from0 is used.

--files-from expects a list of files as its input. Leading or trailing whitespace is stripped from the input lines. Lines starting with # or ; are ignored.

--files-from followed by - reads the list of files from standard input.

Rclone commands with a --files-from flag traverse the remote, treating the names in --files-from as a set of filters.

If the --no-traverse and --files-from flags are used together an rclone command does not traverse the remote. Instead it addresses each path/file named in the file individually. For each path/file name, that requires typically 1 API call. This can be efficient for a short --files-from list and a remote containing many files.

Rclone commands do not error if any names in the --files-from file are missing from the source remote.

The --files-from flag can be repeated in a single rclone command to read path/file names from more than one file. The files are read from left to right along the command line.

Paths within the --files-from file are interpreted as starting with the root specified in the rclone command. Leading / separators are ignored. See --files-from-raw if you need the input to be processed in a raw manner.

E.g. for a file files-from.txt:

# comment
file1.jpg
subdir/file2.jpg

rclone copy --files-from files-from.txt /home/me/pics remote:pics copies the following, if they exist, and only those files.

/home/me/pics/file1.jpg        → remote:pics/file1.jpg
/home/me/pics/subdir/file2.jpg → remote:pics/subdir/file2.jpg

E.g. to copy the following files referenced by their absolute paths:

/home/user1/42
/home/user1/dir/ford
/home/user2/prefect

First find a common subdirectory - in this case /home and put the remaining files in files-from.txt with or without leading /, e.g.

user1/42
user1/dir/ford
user2/prefect

Then copy these to a remote:

rclone copy --files-from files-from.txt /home remote:backup

The three files are transferred as follows:

/home/user1/42       → remote:backup/user1/important
/home/user1/dir/ford → remote:backup/user1/dir/file
/home/user2/prefect  → remote:backup/user2/stuff

Alternatively if / is chosen as root files-from.txt will be:

/home/user1/42
/home/user1/dir/ford
/home/user2/prefect

The copy command will be:

rclone copy --files-from files-from.txt / remote:backup

Then there will be an extra home directory on the remote:

/home/user1/42       → remote:backup/home/user1/42
/home/user1/dir/ford → remote:backup/home/user1/dir/ford
/home/user2/prefect  → remote:backup/home/user2/prefect

--files-from-raw - Read list of source-file names without any processing

This flag is the same as --files-from except that input is read in a raw manner. Lines with leading / trailing whitespace, and lines starting with ; or # are read without any processing. rclone lsf has a compatible format that can be used to export file lists from remotes for input to --files-from-raw.

--files-from0 - Read NUL separated list of source-file names

This flag is the same as --files-from-raw except that input is split on NUL (\0) characters instead of newlines. This allows handling filenames that contain embedded newline characters.

It is similar to the -print0 option of find and the -0 option of xargs.

E.g. to copy files listed by find:

find /path -print0 | rclone copy --files-from0 - / remote:path

--ignore-case - make searches case insensitive

By default, rclone filter patterns are case sensitive. The --ignore-case flag makes all of the filters patterns on the command line case insensitive.

E.g. --include "zaphod.txt" does not match a file Zaphod.txt. With --ignore-case a match is made.

Quoting shell metacharacters

Rclone commands with filter patterns containing shell metacharacters may not as work as expected in your shell and may require quoting.

E.g. linux, OSX (* metacharacter)

  • --include \*.jpg
  • --include '*.jpg'
  • --include='*.jpg'

Microsoft Windows expansion is done by the command, not shell, so --include *.jpg does not require quoting.

If the rclone error Command .... needs .... arguments maximum: you provided .... non flag arguments: is encountered, the cause is commonly spaces within the name of a remote or flag value. The fix then is to quote values containing spaces.

Other filters

--min-size - Don't transfer any file smaller than this

Controls the minimum size file within the scope of an rclone command. Default units are KiB but abbreviations B, K, M, G, T or P are valid.

E.g. rclone ls remote: --min-size 50k lists files on remote: of 50 KiB size or larger.

See the size option docs for more info.

--max-size - Don't transfer any file larger than this

Controls the maximum size file within the scope of an rclone command. Default units are KiB but abbreviations B, K, M, G, T or P are valid.

E.g. rclone ls remote: --max-size 1G lists files on remote: of 1 GiB size or smaller.

See the size option docs for more info.

--max-age - Don't transfer any file older than this

Controls the maximum age of files within the scope of an rclone command.

--max-age applies only to files and not to directories.

E.g. rclone ls remote: --max-age 2d lists files on remote: of 2 days old or less.

See the time option docs for valid formats.

--min-age - Don't transfer any file younger than this

Controls the minimum age of files within the scope of an rclone command. (see --max-age for valid formats)

--min-age applies only to files and not to directories.

E.g. rclone ls remote: --min-age 2d lists files on remote: of 2 days old or more.

See the time option docs for valid formats.

--hash-filter - Deterministically select a subset of files {#hash-filter}

The --hash-filter flag enables selecting a deterministic subset of files, useful for:

  1. Running large sync operations across multiple machines.
  2. Checking a subset of files for bitrot.
  3. Any other operations where a sample of files is required.

Syntax

The flag takes two parameters expressed as a fraction:

--hash-filter K/N
  • N: The total number of partitions (must be a positive integer).
  • K: The specific partition to select (an integer from 0 to N).

For example:

  • --hash-filter 1/3: Selects the first third of the files.
  • --hash-filter 2/3 and --hash-filter 3/3: Select the second and third partitions, respectively.

Each partition is non-overlapping, ensuring all files are covered without duplication.

Random Partition Selection

Use @ as K to randomly select a partition:

--hash-filter @/M

For example, --hash-filter @/3 will randomly select a number between 0 and 2. This will stay constant across retries.

How It Works

  • Rclone takes each file's full path, normalizes it to lowercase, and applies Unicode normalization.
  • It then hashes the normalized path into a 64 bit number.
  • The hash result is reduced modulo N to assign the file to a partition.
  • If the calculated partition does not match K the file is excluded.
  • Other filters may apply if the file is not excluded.

Important: Rclone will traverse all directories to apply the filter.

Usage Notes

  • Safe to use with rclone sync; source and destination selections will match.
  • Do not use with --delete-excluded, as this could delete unselected files.
  • Ignored if --files-from, --files-from-raw or --files-from0 is used.

Examples

Dividing files into 4 partitions

Assuming the current directory contains file1.jpg through file9.jpg:

$ rclone lsf --hash-filter 0/4 .
file1.jpg
file5.jpg

$ rclone lsf --hash-filter 1/4 .
file3.jpg
file6.jpg
file9.jpg

$ rclone lsf --hash-filter 2/4 .
file2.jpg
file4.jpg

$ rclone lsf --hash-filter 3/4 .
file7.jpg
file8.jpg

$ rclone lsf --hash-filter 4/4 . # the same as --hash-filter 0/4
file1.jpg
file5.jpg
Syncing the first quarter of files
rclone sync --hash-filter 1/4 source:path destination:path
Checking a random 1% of files for integrity
rclone check --download --hash-filter @/100 source:path destination:path

Other flags

--delete-excluded - Delete files on dest excluded from sync

Important this flag is dangerous to your data - use with --dry-run and -v first.

In conjunction with rclone sync, --delete-excluded deletes any files on the destination which are excluded from the command.

E.g. the scope of rclone sync --interactive A: B: can be restricted:

rclone --min-size 50k --delete-excluded sync A: B:

All files on B: which are less than 50 KiB are deleted because they are excluded from the rclone sync command.

--dump filters - dump the filters to the output

Dumps the defined filters to standard output in regular expression format.

Useful for debugging.

Exclude directory based on a file

The --exclude-if-present flag controls whether a directory is within the scope of an rclone command based on the presence of a named file within it. The flag can be repeated to check for multiple file names, presence of any of them will exclude the directory.

This flag has a priority over other filter flags.

E.g. for the following directory structure:

dir1/file1
dir1/dir2/file2
dir1/dir2/dir3/file3
dir1/dir2/dir3/.ignore

The command rclone ls --exclude-if-present .ignore dir1 does not list dir3, file3 or .ignore.

Metadata filters {#metadata}

The metadata filters work in a very similar way to the normal file name filters, except they match metadata on the object.

The metadata should be specified as key=value patterns. This may be wildcarded using the normal filter patterns or regular expressions.

For example if you wished to list only local files with a mode of 100664 you could do that with:

rclone lsf -M --files-only --metadata-include "mode=100664" .

Or if you wished to show files with an atime, mtime or btime at a given date:

rclone lsf -M --files-only --metadata-include "[abm]time=2022-12-16*" .

Like file filtering, metadata filtering only applies to files not to directories.

The filters can be applied using these flags.

  • --metadata-include - Include metadatas matching pattern
  • --metadata-include-from - Read metadata include patterns from file (use - to read from stdin)
  • --metadata-exclude - Exclude metadatas matching pattern
  • --metadata-exclude-from - Read metadata exclude patterns from file (use - to read from stdin)
  • --metadata-filter - Add a metadata filtering rule
  • --metadata-filter-from - Read metadata filtering patterns from a file (use - to read from stdin)

Each flag can be repeated. See the section on how filter rules are applied for more details - these flags work in an identical way to the file name filtering flags, but instead of file name patterns have metadata patterns.

Common pitfalls

The most frequent filter support issues on the rclone forum are:

  • Not using paths relative to the root of the remote
  • Not using / to match from the root of a remote
  • Not using ** to match the contents of a directory

rclone gui

The rclone gui command starts the official Web GUI that comes bundled with rclone.

With this command, rclone can serve a web-based GUI (graphical user interface) that is accessible from a normal web browser.

Run it in a terminal and rclone will initialize and then start the GUI.

rclone gui

This will produce logs like this. The terminal window needs to stay open to continue to run the GUI:

2026/04/14 11:36:04 NOTICE: Serving remote control on http://127.0.0.1:50803/
2026/04/14 11:36:04 NOTICE: Serving GUI on http://127.0.0.1:50802/
2026/04/14 11:36:04 NOTICE: GUI available at http://127.0.0.1:50802/login?pass=XXX&url=http%3A%2F%2F127.0.0.1%3A50803%2F&user=gui

You can also add debugging flags when running the GUI, such as -v, which will show more logging output from the rc server.

Using the GUI

Once the browser opens, you will be presented with the Dashboard, the main screen where you can see the status of your remotes and system.

At the top, starting from the left, you will see a series of tabs you can click on. On the right side you will have the logout button and potentially an "Update available" message if a new rclone version has been released.

Dashboard

See live metrics, learn if your remotes are near capacity, and read the changelog for the current version.

Explorer

Explore and manage both local disks and remotes, download files and directories and start transfers.

Remotes

Scroll the list of remotes and tap on it to navigate to the explorer. There you can navigate the contents of your remotes, transfer, download, and even upload files.

Mounts

Mount remotes as local drives on your computer and check on your existing mounts.

Serves

Get quick info about your active serve instances, and start new ones.

Settings

Edit your rclone.conf file directly, set logging flags, and performance parameters.

How it works

When you run rclone gui this is what happens

  • Rclone starts the remote control API ("rc").
  • Rclone starts a second server to serve the Web GUI.
  • If a port, username or password is not specified, then missing values will be auto-generated.
  • Unless --no-open-browser is passed, a browser window will open.
  • The URL already contains the username & password, in which case the GUI will use those values and log you in automatically.

The rc address is passed to the GUI via the url query parameter (the URL printed in the terminal); append ?url=http://... to point it at a different rc server.

Security

It's important to think first about what rclone has access to and what you might be sharing.

A few good measures:

  • Don't use --no-auth (this is for testing only).
  • Do not expose to the local network (eg with --api-addr :5572 --addr :8080) unless you trust all devices on your local network. Prefer 127.0.0.1 or localhost (the default).
  • Use a strong password and non-obvious usernames like "admin" or "rclone" if you are using --user and --pass.
  • If you want to host it on a server and access it remotely, make sure you're only exposing the GUI and not the RC API. They listen on different ports.

If you want to access it remotely but want to avoid running a proxy and exposing ports, you can use Cloudflare Tunnels or localhost.run or Tailscale (all free).

Options

      --addr stringArray       IPaddress:Port for the GUI server (default auto-chosen localhost port)
      --api-addr stringArray   IPaddress:Port for the RC API server (default auto-chosen localhost port)
      --enable-metrics         Enable OpenMetrics/Prometheus compatible endpoint at /metrics
  -h, --help                   help for gui
      --no-auth                Don't require auth for the RC API
      --no-open-browser        Skip opening the browser automatically
      --pass string            Password for RC authentication
      --user string            User name for RC authentication

History

In v1.74 the GUI was redone and embedded within rclone for ease of use. The GUI bundle ships as a compressed zip embedded in the rclone binary and is served from the zip at runtime.

Remote controlling rclone with its API

If rclone is run with the --rc flag then it starts an HTTP server which can be used to remote control rclone using its API.

You can either use the rc command to access the API or use HTTP directly.

If you just want to run a remote control then see the rcd command.

Security {#security}

Access to the rc API is equivalent to shell access as the user running rclone. Treat the rc port as you would an interactive login on the host.

Any caller who can reach the API (and pass authentication, if it is enabled) can, among other things:

  • Run OS commands as the rclone user. core/command re-executes the rclone binary with arbitrary arguments, and several backend options shell out to programs, so even creating a remote with config/create can lead to command execution.
  • Read and write any file reachable by the rclone process, by pointing operations/* or sync/* at a local remote (or via --rc-files / --rc-serve). Writing arbitrary files as the rclone user is itself a route to code execution.
  • Read back stored credentials. rclone configs routinely hold cloud-provider secrets. config/dump and friends expose them, so a compromise of the rc reaches every configured backend.
  • Change rclone's runtime behaviour with options/set, manage remotes with config/*, and stop the process with core/quit.

There is currently no per-endpoint capability or scope system: authentication is all-or-nothing. Granting any access grants all of the above.

Consequently:

  • Do not bind the rc to a network address you do not control. The default bind is loopback (localhost:5572); keep it there unless you have a specific reason to change it.
  • Do not use --rc-no-auth on a non-loopback bind. It disables authentication on the endpoints that access remotes — see --rc-no-auth.
  • Use authentication and TLS (--rc-user/--rc-pass or --rc-htpasswd, plus --rc-cert/--rc-key) whenever the port is reachable by anyone you do not fully trust, and raise --rc-min-tls-version.

Supported parameters

--rc

Flag to start the http server listen on remote requests.

--rc-addr=IP

IPaddress:Port or :Port to bind server to. (default "localhost:5572").

--rc-cert=KEY

SSL PEM key (concatenation of certificate and CA certificate).

--rc-client-ca=PATH

Client certificate authority to verify clients with.

--rc-htpasswd=PATH

htpasswd file - if not provided no authentication is done.

--rc-key=PATH

TLS PEM private key file.

--rc-max-header-bytes=VALUE

Maximum size of request header (default 4096).

--rc-min-tls-version=VALUE

The minimum TLS version that is acceptable. Valid values are "tls1.0", "tls1.1", "tls1.2" and "tls1.3" (default "tls1.0").

--rc-user=VALUE

User name for authentication.

--rc-pass=VALUE

Password for authentication.

--rc-realm=VALUE

Realm for authentication (default "rclone").

--rc-server-read-timeout=DURATION

Timeout for server reading data (default 1h0m0s).

--rc-server-write-timeout=DURATION

Timeout for server writing data (default 1h0m0s).

--rc-serve

Enable the serving of remote objects via the HTTP interface. This means objects will be accessible at http://127.0.0.1:5572/ by default, so you can browse to http://127.0.0.1:5572/ or http://127.0.0.1:5572/* to see a listing of the remotes. Objects may be requested from remotes using this syntax http://127.0.0.1:5572/[remote:path]/path/to/object

Unless the rc server has authentication configured (--rc-user/--rc-pass or --rc-htpasswd) or the --rc-no-auth flag is set, only remotes already present in the config file may be served this way. Inline remotes (e.g. [:webdav,url=...:]), connection string parameters and bare local paths are rejected, since instantiating them from an unauthenticated request could run commands or read arbitrary local files.

Default Off.

global.* connection string options and the rc

Remotes instantiated by the rc do not let connection string global.* options change rclone's process-wide configuration. Remotes created directly on the command line or defined in the config file are unaffected.

A global.* option still takes effect for the individual backend it is set on (exactly like an override.* option), it just does not leak into the global config for the rest of the process.

--rc-serve-no-modtime

Set this flag to skip reading the modification time (can speed things up).

Default Off.

--rc-files /path/to/directory

Path to local files to serve on the HTTP server.

If this is set then rclone will serve the files in that directory. It will also open the root in the web browser if specified. This is for implementing browser based GUIs for rclone functions.

If --rc-user or --rc-pass is set then the URL that is opened will have the authorization in the URL in the http://user:pass@localhost/ style.

Default Off.

--rc-enable-metrics

Enable OpenMetrics/Prometheus compatible endpoint at /metrics. If more control over the metrics is desired (for example running it on a different port or with different auth) then endpoint can be enabled with the --metrics-* flags instead.

Default Off.

--rc-web-gui

Set this flag to serve the default web gui on the same port as rclone.

Default Off.

--rc-allow-origin

Set the allowed Access-Control-Allow-Origin for rc requests.

This enables cross-origin (CORS) requests from browsers on the origin given, for example --rc-allow-origin https://app.example.com. Browser CORS preflight OPTIONS requests are answered by rclone without authentication as browsers never send credentials with them. All other requests, including other OPTIONS requests, must be authenticated when authentication is enabled.

The rclone gui command sets this automatically to the origin the GUI is served from if it is not set explicitly.

Setting this to * allows any origin which is a security risk on an authenticated server.

Default is off (no CORS).

--rc-web-fetch-url

Set the URL to fetch the rclone-web-gui files from.

Default https://api.github.com/repos/rclone/rclone-webui-react/releases/latest.

--rc-web-gui-update

Set this flag to check and update rclone-webui-react from the rc-web-fetch-url.

Default Off.

--rc-web-gui-force-update

Set this flag to force update rclone-webui-react from the rc-web-fetch-url.

Default Off.

--rc-web-gui-no-open-browser

Set this flag to disable opening browser automatically when using web-gui.

Default Off.

--rc-job-expire-duration=DURATION

Expire finished async jobs older than DURATION (default 60s).

--rc-job-expire-interval=DURATION

Interval duration to check for expired async jobs (default 10s).

--rc-no-auth

By default rclone will require authorisation to have been set up on the rc interface in order to use any methods which access any rclone remotes. Eg operations/list is denied as it involved creating a remote as is sync/copy.

If this is set then no authorisation will be required on the server to use these methods. The alternative is to use --rc-user and --rc-pass and use these credentials in the request.

Without authorisation or this flag, POST request bodies are limited to 64 KiB.

Default Off.

--rc-baseurl

Prefix for URLs.

Default is root

--rc-template

User-specified template.

Accessing the remote control via the rclone rc command {#api-rc}

Rclone itself implements the remote control protocol in its rclone rc command.

You can use it like this:

$ rclone rc rc/noop param1=one param2=two
{
    "param1": "one",
    "param2": "two"
}

If the remote is running on a different URL than the default http://localhost:5572/, use the --url option to specify it:

rclone rc --url http://some.remote:1234/ rc/noop

Or, if the remote is listening on a Unix socket, use the --unix-socket option instead:

rclone rc --unix-socket /tmp/rclone.sock rc/noop

Run rclone rc on its own, without any commands, to see the help for the installed remote control commands. Note that this also needs to connect to the remote server.

JSON input

rclone rc also supports a --json flag which can be used to send more complicated input parameters.

$ rclone rc --json '{ "p1": [1,"2",null,4], "p2": { "a":1, "b":2 } }' rc/noop
{
    "p1": [
        1,
        "2",
        null,
        4
    ],
    "p2": {
        "a": 1,
        "b": 2
    }
}

If the parameter being passed is an object then it can be passed as a JSON string rather than using the --json flag which simplifies the command line.

rclone rc operations/list fs=/tmp remote=test opt='{"showHash": true}'

Rather than

rclone rc operations/list --json '{"fs": "/tmp", "remote": "test", "opt": {"showHash": true}}'

Paths on Windows

Prefer forward slashes (/) in Windows paths passed to the rc API, for example C:/path1/path2. If you use backslashes (\) in a JSON string, escape each one as \\. These JSON objects contain equivalent Windows paths:

{"remote": "C:/path1/path2"}
{"remote": "C:\\path1\\path2"}

This applies to JSON request bodies and to JSON-encoded object parameters, such as parameters in config/create, even when that JSON is sent in a query string or form data. An unescaped backslash can cause a JSON parsing error or be interpreted as an escape sequence, such as \t for a tab.

String parameters sent directly in a query string or form data, without JSON encoding, do not need JSON escaping. Use your HTTP client's URL encoding for query strings and form data.

Special parameters

The rc interface supports some special parameters which apply to all commands. These start with _ to show they are different.

Running asynchronous jobs with _async = true

Each rc call is classified as a job and it is assigned its own id. By default jobs are executed immediately as they are created or synchronously.

If _async has a true value when supplied to an rc call then it will return immediately with a job id and execute id, and the task will be run in the background. The job/status call can be used to get information of the background job. The job can be queried for up to 1 minute after it has finished.

It is recommended that potentially long running jobs, e.g. sync/sync, sync/copy, sync/move, operations/purge are run with the _async flag to avoid any potential problems with the HTTP request and response timing out.

As an alternative to _async, you can send the HTTP header Prefer: respond-async (RFC 7240). This has the same effect as _async = true but additionally returns HTTP status 202 (Accepted) instead of 200, and includes a Preference-Applied: respond-async response header. The 202 status code makes it easy for clients to distinguish an async response from a completed one without inspecting the body.

Starting a job with the _async flag:

$ rclone rc --json '{ "p1": [1,"2",null,4], "p2": { "a":1, "b":2 }, "_async": true }' rc/noop
{
    "jobid": 2,
    "executeId": "d794c33c-463e-4acf-b911-f4b23e4f40b7"
}

The jobid is a unique identifier for the job within this rclone instance. The executeId identifies the rclone process instance and changes after rclone restart. Together, the pair (executeId, jobid) uniquely identifies a job across rclone restarts.

Query the status to see if the job has finished. For more information on the meaning of these return parameters see the job/status call.

$ rclone rc --json '{ "jobid":2 }' job/status
{
    "duration": 0.000124163,
    "endTime": "2018-10-27T11:38:07.911245881+01:00",
    "error": "",
    "executeId": "d794c33c-463e-4acf-b911-f4b23e4f40b7",
    "finished": true,
    "id": 2,
    "output": {
        "_async": true,
        "p1": [
            1,
            "2",
            null,
            4
        ],
        "p2": {
            "a": 1,
            "b": 2
        }
    },
    "startTime": "2018-10-27T11:38:07.911121728+01:00",
    "success": true
}

job/list can be used to show running or recently completed jobs along with their status

$ rclone rc job/list
{
    "executeId": "d794c33c-463e-4acf-b911-f4b23e4f40b7",
    "finished_ids": [
        1
    ],
    "jobids": [
        1,
        2
    ],
    "running_ids": [
        2
    ]
}

This shows:

  • executeId - the current rclone instance ID (same for all jobs, changes after restart)
  • jobids - array of all job IDs (both running and finished)
  • running_ids - array of currently running job IDs
  • finished_ids - array of finished job IDs

Setting config flags with _config

If you wish to set config (the equivalent of the global flags) for the duration of an rc call only then pass in the _config parameter.

Alternatively, you can pass config options flat at the top level of the parameter map. The option names are the same as their CLI flags without -- and with - replaced by _ (e.g. transfers instead of Transfers inside _config).

This should be in the same format as the main key returned by options/get.

rclone rc --loopback options/get blocks=main

You can see more help on these options with this command (see the options blocks section for more info).

rclone rc --loopback options/info blocks=main

For example, if you wished to run a sync with the --checksum parameter, you would pass this parameter in your JSON blob.

"_config":{"CheckSum": true}

Or pass it flat at the top level:

"checksum": true

If using rclone rc this could be passed as

rclone rc sync/sync ... _config='{"CheckSum": true}'

Or simply flat:

rclone rc sync/sync ... checksum=true

If both flat parameters and _config are supplied, the parameters in the legacy _config block will take precedence.

Any config parameters you don't set will inherit the global defaults which were set with command line flags or environment variables.

Note that it is possible to set some values as strings or integers - see data types for more info. Here is an example setting the equivalent of --buffer-size in string or integer format.

"_config":{"BufferSize": "42M"}
"_config":{"BufferSize": 44040192}

Or flat:

"buffer_size": "42M"
"buffer_size": 44040192

If you wish to check the _config assignment has worked properly then calling options/local will show what the value got set to.

Setting LogLevel sets the log level for the logs made by the rc call only, for example to debug a single call.

"_config":{"LogLevel": "DEBUG"}

Note that not all of rclone's logs support this yet. In particular the backends log at the global log level.

Setting filter flags with _filter

If you wish to set filters for the duration of an rc call only then pass in the _filter parameter.

Alternatively, you can pass filter options flat at the top level of the parameter map. The option names are the same as their CLI flags without -- and with - replaced by _ (e.g. exclude instead of Exclude inside _filter).

This should be in the same format as the filter key returned by options/get.

rclone rc --loopback options/get blocks=filter

You can see more help on these options with this command (see the options blocks section for more info).

rclone rc --loopback options/info blocks=filter

For example, if you wished to run a sync with these flags

--max-size 1M --max-age 42s --include "a" --include "b"

you would pass this parameter in your JSON blob.

"_filter":{"MaxSize":"1M", "IncludeRule":["a","b"], "MaxAge":"42s"}

Or pass them flat at the top level:

"max_size":"1M", "include":["a","b"], "max_age":"42s"

If using rclone rc this could be passed as

rclone rc ... _filter='{"MaxSize":"1M", "IncludeRule":["a","b"], "MaxAge":"42s"}'

Or simply flat:

rclone rc ... max_size=1M include="a,b" max_age=42s

If both flat parameters and _filter are supplied, the parameters in the legacy _filter block will take precedence.

Any filter parameters you don't set will inherit the global defaults which were set with command line flags or environment variables.

Note that it is possible to set some values as strings or integers - see data types for more info. Here is an example setting the equivalent of --buffer-size in string or integer format.

"_filter":{"MinSize": "42M"}
"_filter":{"MinSize": 44040192}

Or flat:

"min_size": "42M"

If you wish to check the _filter assignment has worked properly then calling options/local will show what the value got set to.

Assigning operations to groups with _group = value

Each rc call has its own stats group for tracking its metrics. By default grouping is done by the composite group name from prefix job/ and id of the job like so job/1.

If _group has a value then stats for that request will be grouped under that value. This allows caller to group stats under their own name.

Stats for specific group can be accessed by passing group to core/stats:

$ rclone rc --json '{ "group": "job/1" }' core/stats
{
    "speed": 12345
    ...
}

Data types {#data-types}

When the API returns types, these will mostly be straight forward integer, string or boolean types.

However some of the types returned by the options/get call and taken by the options/set calls as well as the vfsOpt, mountOpt and the _config parameters.

  • Duration - these are returned as an integer duration in nanoseconds. They may be set as an integer, or they may be set with time string, eg "5s". See the options section for more info.
  • Size - these are returned as an integer number of bytes. They may be set as an integer or they may be set with a size suffix string, eg "10M". See the options section for more info.
  • Enumerated type (such as CutoffMode, DumpFlags, LogLevel, VfsCacheMode - these will be returned as an integer and may be set as an integer but more conveniently they can be set as a string, eg "HARD" for CutoffMode or DEBUG for LogLevel.
  • BandwidthSpec - this will be set and returned as a string, eg "1M".

Option blocks {#option-blocks}

The calls options/info (for the main config) and config/providers (for the backend config) may be used to get information on the rclone configuration options. This can be used to build user interfaces for displaying and setting any rclone option.

These consist of arrays of Option blocks. These have the following format. Each block describes a single option.

Field Type Optional Description
Name string N name of the option in snake_case
FieldName string N name of the field used in the rc - if blank use Name. May contain "." for nested fields.
Help string N help, started with a single sentence on a single line
Groups string Y groups this option belongs to - comma separated string for options classification
Provider string Y set to filter on provider
Default any N default value, if set (and not to nil or "") then Required does nothing
Value any N value to be set by flags
Examples Examples Y predefined values that can be selected from list (multiple-choice option)
ShortOpt string Y the short command line option for this
Hide Visibility N if non zero, this option is hidden from the configurator or the command line
Required bool N this option is required, meaning value cannot be empty unless there is a default
IsPassword bool N set if the option is a password
NoPrefix bool N set if the option for this should not use the backend prefix
Advanced bool N set if this is an advanced config option
Exclusive bool N set if the answer can only be one of the examples (empty string allowed unless Required or Default is set)
Sensitive bool N set if this option should be redacted when using rclone config redacted

An example of this might be the --log-level flag. Note that the Name of the option becomes the command line flag with _ replaced with -.

{
    "Advanced": false,
    "Default": 5,
    "DefaultStr": "NOTICE",
    "Examples": [
        {
            "Help": "",
            "Value": "EMERGENCY"
        },
        {
            "Help": "",
            "Value": "ALERT"
        },
        ...
    ],
    "Exclusive": true,
    "FieldName": "LogLevel",
    "Groups": "Logging",
    "Help": "Log level DEBUG|INFO|NOTICE|ERROR",
    "Hide": 0,
    "IsPassword": false,
    "Name": "log_level",
    "NoPrefix": true,
    "Required": true,
    "Sensitive": false,
    "Type": "LogLevel",
    "Value": null,
    "ValueStr": "NOTICE"
},

Note that the Help may be multiple lines separated by \n. The first line will always be a short sentence and this is the sentence shown when running rclone help flags.

Specifying remotes to work on

Remotes are specified with the fs=, srcFs=, dstFs= parameters depending on the command being used.

The parameters can be a string as per the rest of rclone, eg s3:bucket/path or :sftp:/my/dir. They can also be specified as JSON blobs.

If specifying a JSON blob it should be a object mapping strings to strings. These values will be used to configure the remote. There are 3 special values which may be set:

  • type - set to type to specify a remote called :type:
  • _name - set to name to specify a remote called name:
  • _root - sets the root of the remote - may be empty

One of _name or type should normally be set. If the local backend is desired then type should be set to local. If _root isn't specified then it defaults to the root of the remote.

For example this JSON is equivalent to remote:/tmp

{
    "_name": "remote",
    "_root": "/tmp"
}

And this is equivalent to :sftp,host='example.com':/tmp

{
    "type": "sftp",
    "host": "example.com",
    "_root": "/tmp"
}

And this is equivalent to /tmp/dir

{
    "type": "local",
    "_root": "/tmp/dir"
}

Supported commands

backend/command: Runs a backend command. {#backend-command}

This takes the following parameters:

  • command - a string with the command name
  • fs - a remote name string e.g. "drive:"
  • arg - a list of arguments for the backend command
  • opt - a map of string to string of options

Returns:

  • result - result from the backend command

Example:

rclone rc backend/command command=noop fs=. -o echo=yes -o blue -a path1 -a path2

Returns

{
	"result": {
		"arg": [
			"path1",
			"path2"
		],
		"name": "noop",
		"opt": {
			"blue": "",
			"echo": "yes"
		}
	}
}

Note that this is the direct equivalent of using this "backend" command:

rclone backend noop . -o echo=yes -o blue path1 path2

Note that arguments must be preceded by the "-a" flag

See the backend command for more information.

cache/expire: Purge a remote from cache {#cache-expire}

Purge a remote from the cache backend. Supports either a directory or a file. Params:

  • remote = path to remote (required)
  • withData = true/false to delete cached data (chunks) as well (optional)

Eg

rclone rc cache/expire remote=path/to/sub/folder/
rclone rc cache/expire remote=/ withData=true

cache/fetch: Fetch file chunks {#cache-fetch}

Ensure the specified file chunks are cached on disk.

The chunks= parameter specifies the file chunks to check. It takes a comma separated list of array slice indices. The slice indices are similar to Python slices: start[:end]

start is the 0 based chunk number from the beginning of the file to fetch inclusive. end is 0 based chunk number from the beginning of the file to fetch exclusive. Both values can be negative, in which case they count from the back of the file. The value "-5:" represents the last 5 chunks of a file.

Some valid examples are: ":5,-5:" -> the first and last five chunks "0,-2" -> the first and the second last chunk "0:10" -> the first ten chunks

Any parameter with a key that starts with "file" can be used to specify files to fetch, e.g.

rclone rc cache/fetch chunks=0 file=hello file2=home/goodbye

File names will automatically be encrypted when the a crypt remote is used on top of the cache.

cache/stats: Get cache stats {#cache-stats}

Show statistics for the cache remote.

Authentication is not required for this call.

config/create: create the config for a remote. {#config-create}

This takes the following parameters:

  • name - name of remote
  • parameters - a map of { "key": "value" } pairs
  • type - type of the new remote
  • opt - a dictionary of options to control the configuration
    • obscure - declare passwords are plain and need obscuring
    • noObscure - declare passwords are already obscured and don't need obscuring
    • noOutput - don't print anything to stdout
    • nonInteractive - don't interact with a user, return questions
    • continue - continue the config process with an answer
    • all - ask all the config questions not just the post config ones
    • state - state to restart with - used with continue
    • result - result to restart with - used with continue

See the config create command for more information on the above.

config/delete: Delete a remote in the config file. {#config-delete}

Parameters:

  • name - name of remote to delete

See the config delete command for more information on the above.

config/dump: Dumps the config file. {#config-dump}

Returns a JSON object:

  • key: value

Where keys are remote names and values are the config parameters.

See the config dump command for more information on the above.

config/get: Get a remote in the config file. {#config-get}

Parameters:

  • name - name of remote to get

See the config dump command for more information on the above.

config/listremotes: Lists the remotes in the config file and defined in environment variables. {#config-listremotes}

Returns

  • remotes - array of remote names

See the listremotes command for more information on the above.

config/oauthstatus: Get the status of the OAuth authentication server. {#config-oauthstatus}

Returns the current status of the OAuth authentication server.

Returns a JSON object:

  • status - "running" or "stopped"
  • authUrl - URL for the authorization (only if status is "running")

Eg

{
    "status": "running",
    "authUrl": "http://127.0.0.1:53682/auth?state=..."
}

config/oauthstop: Stop any running OAuth authentication server. {#config-oauthstop}

Stops the OAuth authentication server if one is running.

This can be used to recover from an interrupted OAuth flow without restarting rclone. If no OAuth authentication is in progress, an error is returned.

config/password: password the config for a remote. {#config-password}

This takes the following parameters:

  • name - name of remote
  • parameters - a map of { "key": "value" } pairs

See the config password command for more information on the above.

config/paths: Reads the config file path and other important paths. {#config-paths}

Returns a JSON object with the following keys:

  • config: path to config file
  • cache: path to root of cache directory
  • temp: path to root of temporary directory

Eg

{
    "cache": "/home/USER/.cache/rclone",
    "config": "/home/USER/.rclone.conf",
    "temp": "/tmp"
}

See the config paths command for more information on the above.

config/providers: Shows how providers are configured in the config file. {#config-providers}

Returns a JSON object:

  • providers - array of objects

See the config providers command for more information on the above.

Note that the Options blocks are in the same format as returned by "options/info". They are described in the option blocks section.

config/setpath: Set the path of the config file {#config-setpath}

Parameters:

  • path - path to the config file to use

config/unlock: Unlock the config file. {#config-unlock}

Unlocks the config file if it is locked.

Parameters:

  • 'configPassword' - password to unlock the config file

A good idea is to disable AskPassword before making this call

config/unset: Unset keys in a remote in the config file. {#config-unset}

Parameters:

  • name - name of remote
  • keys - a list of key names to remove

Returns:

  • removed - a list of the keys that were actually removed

See the config unset command for more information on the above.

config/update: update the config for a remote. {#config-update}

This takes the following parameters:

  • name - name of remote
  • parameters - a map of { "key": "value" } pairs
  • opt - a dictionary of options to control the configuration
    • obscure - declare passwords are plain and need obscuring
    • noObscure - declare passwords are already obscured and don't need obscuring
    • noOutput - don't print anything to stdout
    • nonInteractive - don't interact with a user, return questions
    • continue - continue the config process with an answer
    • all - ask all the config questions not just the post config ones
    • state - state to restart with - used with continue
    • result - result to restart with - used with continue

See the config update command for more information on the above.

core/bwlimit: Set the bandwidth limit. {#core-bwlimit}

This sets the bandwidth limit to the string passed in. This should be a single bandwidth limit entry or a pair of upload:download bandwidth.

Eg

rclone rc core/bwlimit rate=off
{
    "bytesPerSecond": -1,
    "bytesPerSecondTx": -1,
    "bytesPerSecondRx": -1,
    "rate": "off"
}
rclone rc core/bwlimit rate=1M
{
    "bytesPerSecond": 1048576,
    "bytesPerSecondTx": 1048576,
    "bytesPerSecondRx": 1048576,
    "rate": "1M"
}
rclone rc core/bwlimit rate=1M:100k
{
    "bytesPerSecond": 1048576,
    "bytesPerSecondTx": 1048576,
    "bytesPerSecondRx": 131072,
    "rate": "1M"
}

If the rate parameter is not supplied then the bandwidth is queried

rclone rc core/bwlimit
{
    "bytesPerSecond": 1048576,
    "bytesPerSecondTx": 1048576,
    "bytesPerSecondRx": 1048576,
    "rate": "1M"
}

The format of the parameter is exactly the same as passed to --bwlimit except only one bandwidth may be specified.

In either case "rate" is returned as a human-readable string, and "bytesPerSecond" is returned as a number.

core/command: Run a rclone terminal command over rc. {#core-command}

This takes the following parameters:

  • command - a string with the command name.
  • arg - a list of arguments for the backend command.
  • opt - a map of string to string of options.
  • returnType - one of ("COMBINED_OUTPUT", "STREAM", "STREAM_ONLY_STDOUT", "STREAM_ONLY_STDERR").
    • Defaults to "COMBINED_OUTPUT" if not set.
    • The STREAM returnTypes will write the output to the body of the HTTP message.
    • The COMBINED_OUTPUT will write the output to the "result" parameter.

Returns:

  • result - result from the backend command.
    • Only set when using returnType "COMBINED_OUTPUT".
  • error - set if rclone exits with an error code.
  • returnType - one of ("COMBINED_OUTPUT", "STREAM", "STREAM_ONLY_STDOUT", "STREAM_ONLY_STDERR").

Example:

rclone rc core/command command=ls -a mydrive:/ -o max-depth=1
rclone rc core/command -a ls -a mydrive:/ -o max-depth=1

Returns:

{
	"error": false,
	"result": "<Raw command line output>"
}

OR
{
	"error": true,
	"result": "<Raw command line output>"
}

core/disks: List the local disks {#core-disks}

This does not take any parameters

This call is for rclone GUI programs to enumerate local disks and important directories for doing transfers to and from. The list returned will include the root directory and the user's home directory and any mounted disks. The returned items should be usable directly as remotes.

Returns:

  • disks
    • This is an array of strings of local disk names

core/du: Returns disk usage of a locally attached disk. {#core-du}

This returns the disk usage for the local directory passed in as dir.

If the directory is not passed in, it defaults to the directory pointed to by --cache-dir.

  • dir - string (optional)

Returns:

{
	"dir": "/",
	"info": {
		"Available": 361769115648,
		"Free": 361785892864,
		"Total": 982141468672
	}
}

Authentication is not required for this call.

core/gc: Runs a garbage collection. {#core-gc}

This tells the go runtime to do a garbage collection run. It isn't necessary to call this normally, but it can be useful for debugging memory problems.

core/group-list: Returns list of stats. {#core-group-list}

This returns list of stats groups currently in memory.

Returns the following values:

{
	"groups":  an array of group names:
		[
			"group1",
			"group2",
			...
		]
}

Authentication is not required for this call.

core/memstats: Returns the memory statistics {#core-memstats}

This returns the memory statistics of the running program. What the values mean are explained in the go docs: https://golang.org/pkg/runtime/#MemStats

The most interesting values for most people are:

  • HeapAlloc - this is the amount of memory rclone is actually using
  • HeapSys - this is the amount of memory rclone has obtained from the OS
  • Sys - this is the total amount of memory requested from the OS
    • It is virtual memory so may include unused memory

core/obscure: Obscures a string passed in. {#core-obscure}

Pass a clear string and rclone will obscure it for the config file:

  • clear - string

Returns:

  • obscured - string

core/pid: Return PID of current process {#core-pid}

This returns PID of current process. Useful for stopping rclone process.

core/quit: Terminates the app. {#core-quit}

(Optional) Pass an exit code to be used for terminating the app:

  • exitCode - int

core/stats: Returns stats about current transfers. {#core-stats}

This returns all available stats:

rclone rc core/stats

If group is not provided then summed up stats for all groups will be returned.

Parameters

  • group - name of the stats group (string, optional)
  • short - if true will not return the transferring and checking arrays (boolean, optional)

Returns the following values:

{
	"bytes": total transferred bytes since the start of the group,
	"checks": number of files checked,
	"deletedDirs": number of directories deleted,
	"deletes" : number of files deleted,
	"elapsedTime": time in floating point seconds since rclone was started,
	"errors": number of errors,
	"eta": estimated time in seconds until the group completes,
	"fatalError": boolean whether there has been at least one fatal error,
	"lastError": last error string,
	"renames" : number of files renamed,
	"listed" : number of directory entries listed,
	"retryError": boolean showing whether there has been at least one non-NoRetryError,
        "serverSideCopies": number of server side copies done,
        "serverSideCopyBytes": number bytes server side copied,
        "serverSideMoves": number of server side moves done,
        "serverSideMoveBytes": number bytes server side moved,
	"speed": average speed in bytes per second since start of the group,
	"totalBytes": total number of bytes in the group,
	"totalChecks": total number of checks in the group,
	"totalTransfers": total number of transfers in the group,
	"transferTime" : total time spent on running jobs,
	"transfers": number of transferred files,
	"updatedDirs": number of directories updated (modtime or metadata set),
	"transferring": an array of currently active file transfers:
		[
			{
				"bytes": total transferred bytes for this file,
				"eta": estimated time in seconds until file transfer completion
				"name": name of the file,
				"percentage": progress of the file transfer in percent,
				"speed": average speed over the whole transfer in bytes per second,
				"speedAvg": current speed in bytes per second as an exponentially weighted moving average,
				"size": size of the file in bytes
			}
		],
	"checking": an array of names of currently active file checks
		[]
}

Values for "transferring", "checking" and "lastError" are only assigned if data is available. The value for "eta" is null if an eta cannot be determined.

Authentication is not required for this call.

core/stats-delete: Delete stats group. {#core-stats-delete}

This deletes entire stats group.

Parameters

  • group - name of the stats group (string)

core/stats-reset: Reset stats. {#core-stats-reset}

This clears counters, errors and finished transfers for all stats or specific stats group if group is provided.

Parameters

  • group - name of the stats group (string)

core/transferred: Returns stats about completed transfers. {#core-transferred}

This returns stats about completed transfers:

rclone rc core/transferred

If group is not provided then completed transfers for all groups will be returned.

Note only the last 100 completed transfers are returned.

Parameters

  • group - name of the stats group (string)

Returns the following values:

{
	"transferred":  an array of completed transfers (including failed ones):
		[
			{
				"name": name of the file,
				"size": size of the file in bytes,
				"bytes": total transferred bytes for this file,
				"checked": if the transfer is only checked (skipped, deleted),
				"what": the purpose of the transfer (transferring, deleting, checking, importing, hashing, merging, listing, moving, renaming),
				"timestamp": integer representing millisecond unix epoch,
				"error": string description of the error (empty if successful),
				"jobid": id of the job that this transfer belongs to
			}
		]
}

Authentication is not required for this call.

core/version: Shows the current version of rclone, Go and the OS. {#core-version}

This shows the current versions of rclone, Go and the OS:

  • version - rclone version, e.g. "v1.71.2"
  • decomposed - version number as [major, minor, patch]
  • isGit - boolean - true if this was compiled from the git version
  • isBeta - boolean - true if this is a beta version
  • os - OS in use as according to Go GOOS (e.g. "linux")
  • osKernel - OS Kernel version (e.g. "6.8.0-86-generic (x86_64)")
  • osVersion - OS Version (e.g. "ubuntu 24.04 (64 bit)")
  • osArch - cpu architecture in use (e.g. "arm64 (ARMv8 compatible)")
  • arch - cpu architecture in use according to Go GOARCH (e.g. "arm64")
  • goVersion - version of Go runtime in use (e.g. "go1.25.0")
  • linking - type of rclone executable (static or dynamic)
  • goTags - space separated build tags or "none"

Authentication is not required for this call.

debug/set-block-profile-rate: Set runtime.SetBlockProfileRate for blocking profiling. {#debug-set-block-profile-rate}

SetBlockProfileRate controls the fraction of goroutine blocking events that are reported in the blocking profile. The profiler aims to sample an average of one blocking event per rate nanoseconds spent blocked.

To include every blocking event in the profile, pass rate = 1. To turn off profiling entirely, pass rate <= 0.

After calling this you can use this to see the blocking profile:

go tool pprof http://localhost:5572/debug/pprof/block

Parameters:

  • rate - int

debug/set-gc-percent: Call runtime/debug.SetGCPercent for setting the garbage collection target percentage. {#debug-set-gc-percent}

SetGCPercent sets the garbage collection target percentage: a collection is triggered when the ratio of freshly allocated data to live data remaining after the previous collection reaches this percentage. SetGCPercent returns the previous setting. The initial setting is the value of the GOGC environment variable at startup, or 100 if the variable is not set.

This setting may be effectively reduced in order to maintain a memory limit. A negative percentage effectively disables garbage collection, unless the memory limit is reached.

See https://pkg.go.dev/runtime/debug#SetMemoryLimit for more details.

Parameters:

  • gc-percent - int

debug/set-mutex-profile-fraction: Set runtime.SetMutexProfileFraction for mutex profiling. {#debug-set-mutex-profile-fraction}

SetMutexProfileFraction controls the fraction of mutex contention events that are reported in the mutex profile. On average 1/rate events are reported. The previous rate is returned.

To turn off profiling entirely, pass rate 0. To just read the current rate, pass rate < 0. (For n>1 the details of sampling may change.)

Once this is set you can look use this to profile the mutex contention:

go tool pprof http://localhost:5572/debug/pprof/mutex

Parameters:

  • rate - int

Results:

  • previousRate - int

debug/set-soft-memory-limit: Call runtime/debug.SetMemoryLimit for setting a soft memory limit for the runtime. {#debug-set-soft-memory-limit}

SetMemoryLimit provides the runtime with a soft memory limit.

The runtime undertakes several processes to try to respect this memory limit, including adjustments to the frequency of garbage collections and returning memory to the underlying system more aggressively. This limit will be respected even if GOGC=off (or, if SetGCPercent(-1) is executed).

The input limit is provided as bytes, and includes all memory mapped, managed, and not released by the Go runtime. Notably, it does not account for space used by the Go binary and memory external to Go, such as memory managed by the underlying system on behalf of the process, or memory managed by non-Go code inside the same process. Examples of excluded memory sources include: OS kernel memory held on behalf of the process, memory allocated by C code, and memory mapped by syscall.Mmap (because it is not managed by the Go runtime).

A zero limit or a limit that's lower than the amount of memory used by the Go runtime may cause the garbage collector to run nearly continuously. However, the application may still make progress.

The memory limit is always respected by the Go runtime, so to effectively disable this behavior, set the limit very high. math.MaxInt64 is the canonical value for disabling the limit, but values much greater than the available memory on the underlying system work just as well.

See https://go.dev/doc/gc-guide for a detailed guide explaining the soft memory limit in more detail, as well as a variety of common use-cases and scenarios.

SetMemoryLimit returns the previously set memory limit. A negative input does not adjust the limit, and allows for retrieval of the currently set memory limit.

Parameters:

  • mem-limit - int

fscache/clear: Clear the Fs cache. {#fscache-clear}

This clears the fs cache. This is where remotes created from backends are cached for a short while to make repeated rc calls more efficient.

If you change the parameters of a backend then you may want to call this to clear an existing remote out of the cache before re-creating it.

fscache/entries: Returns the number of entries in the fs cache. {#fscache-entries}

This returns the number of entries in the fs cache.

Returns

  • entries - number of items in the cache

job/batch: Run a batch of rclone rc commands concurrently. {#job-batch}

This takes the following parameters:

  • concurrency - int - do this many commands concurrently. Defaults to --transfers if not set.
  • inputs - an list of inputs to the commands with an extra _path parameter
{
    "_path": "rc/path",
    "param1": "parameter for the path as documented",
    "param2": "parameter for the path as documented, etc",
}

The inputs may use _async, _group, _config and _filter as normal when using the rc.

Returns:

  • results - a list of results from the commands with one entry for each in inputs.

For example:

rclone rc job/batch --json '{
  "inputs": [
    {
      "_path": "rc/noop",
      "par