Skip to content

armgit / Pulse

Public Imported from GitHub
0

Real-time monitoring dashboard for Proxmox VE, PBS, Docker, Kubernetes, TrueNAS and vSphere. Self-hosted, with alerts for failed backups, full disks, unhealthy containers and offline nodes.

Prepare v6.8.0-beta.1

pulse-triage[bot]8d9276caf617
diff --git a/docs/releases/RELEASE_NOTES_v6.8.0-beta.1.md b/docs/releases/RELEASE_NOTES_v6.8.0-beta.1.md
index 38b6dc9fb6..35b83362b3 100644
--- a/docs/releases/RELEASE_NOTES_v6.8.0-beta.1.md
+++ b/docs/releases/RELEASE_NOTES_v6.8.0-beta.1.md
@@ -2,7 +2,7 @@
This beta opens the `v6.8.0` candidate line after stable `v6.7.0`, which remains the rollback target.
-Pulse v6.8.0 adds clearer backup summaries, two weeks of backup job outcomes and individual host sensor thresholds. It also improves notification recovery, agent installation checks and how settings, logs and downloads behave when access changes.
+Pulse v6.8.0 makes backup problems easier to identify, adds individual temperature sensor limits, and improves notification recovery. It also makes agent installation failures clearer and prevents outdated settings, logs and downloads from following you between accounts or workspaces.
## Highlights
@@ -12,101 +12,111 @@ Pulse v6.8.0 adds clearer backup summaries, two weeks of backup job outcomes and
## What's improved
-- **Latest readings remain visible.** Resource drawers keep CPU, memory and other available readings when history fails or the selected range exceeds the plan. Older readings stay labelled.
-- **Settings failures have a recovery path.** Restricted panels hide after an unsuccessful access check and offer a retry instead of retaining stale controls.
-- **Saved systems are distinguished from failed reads.** Infrastructure shows connection-list and saved-configuration errors with retry controls instead of presenting empty onboarding or incomplete editors.
-- **Role editing preserves existing choices.** Saved permissions remain visible even outside the shortcut list, and unchanged access refreshes preserve drafts.
-- **Help handles loading failures clearly.** Missing guides, refused access and connection failures have distinct messages. Retry keeps the same guide, and repeated heading links restore keyboard focus.
+- **Latest readings stay visible.** Resource drawers keep available CPU, memory and other readings when history fails or the selected range exceeds your plan. Retained readings remain labelled.
+- **Settings offers recovery after access checks fail.** Restricted panels hide outdated controls and offer retry instead of continuing with stale permissions.
+- **Infrastructure errors no longer look like first setup.** Failed connection-list and saved-configuration reads show retry controls instead of empty onboarding or incomplete editors.
+- **Role editing preserves saved choices.** Existing permissions remain visible outside the shortcut list, and unchanged access refreshes preserve drafts.
+- **Help explains loading failures.** Docs distinguishes missing guides, refused access and connection failures. Retry keeps the same guide, and repeated heading links restore keyboard focus.
- **French is available.** Supported interface surfaces and the getting-started guide now include French.
## Alerts and notifications
-- **Notifications survive graceful restarts.** Queued deliveries and buffered alert groups retain their work and grouping deadline. Startup restores saved destinations and delivery policy before sending.
-- **Acknowledgements avoid duplicate requests.** Individual, group and bulk actions share pending state across navigation. Partial group failures report both successful and failed counts.
-- **Active-alert markers return.** Threshold rows show the red dot for matching open alerts, including machine filesystem and disk-temperature alerts. Help explains what the marker means.
+- **Notifications survive graceful restarts.** Queued deliveries and buffered alert groups keep their work and grouping deadline. Saved destinations and delivery policy load before sending resumes.
+- **Acknowledgements avoid duplicate requests.** Individual, group and bulk actions share pending state across navigation. Partial group failures report successful and failed counts separately.
+- **Active-alert markers return.** Threshold rows show the red dot for matching open alerts, including machine filesystem and disk-temperature alerts. Help distinguishes these markers from readings and delivery confirmation.
- **Threshold notes survive editing.** Saved notes reappear when reopening an override, and edits that change only the note now refresh correctly.
-- **Alert settings recover safely.** Thresholds, destinations, delivery history and alert-intent controls discard results from a previous account or workspace. Unread saved destinations stay hidden until retry succeeds.
-- **Delivery errors stay readable.** Recent delivery activity preserves ordinary error messages while continuing to redact credentials in webhook URLs.
+- **Alert settings stay with your current access.** Thresholds, destinations, delivery history and alert-intent controls discard previous-account or workspace replies. Unread saved destinations stay hidden until retry succeeds.
+- **Delivery errors stay readable.** Recent delivery activity preserves ordinary error messages while continuing to hide credentials in webhook URLs.
+- **Failed PBS maintenance jobs raise datastore alerts.** Sync, verify, prune and garbage-collection failures also appear in Backup health, separately from guest protection. Outages, partial reads and running retries no longer clear known failures.
## Disks and storage
-- **Full disks remain visible in summaries.** Compact Proxmox node rows use the fullest measured disk, including when another disk lacks a reading. Wide host rows retain individual disk bars.
-- **Disk controls are easier to read.** Disk, alert and view toggles make the selected option clearer. Narrow disk cells preserve the percentage instead of clipping it beside the disk label.
-- **Storage guidance protects retained history.** Help now recommends persistent metric storage and explains that changing the database path neither migrates history nor fixes excessive writes.
+- **Full disks remain visible in summaries.** Compact Proxmox node rows use the fullest measured disk even when another disk lacks a reading. Wide host rows retain individual disk bars.
+- **Disk controls are easier to read.** Disk, alert and view toggles show the selected option more clearly. Narrow disk cells preserve the percentage beside the disk label.
+- **Storage guidance protects retained history.** Help recommends persistent metric storage and explains that changing the database path neither migrates history nor fixes excessive writes.
## Proxmox, PBS and backups
- **Backup jobs show two weeks of outcomes.** Backups adds daily Proxmox VE job results and a last-24-hours summary, distinguishing failures, warnings, running jobs and gaps in observed history.
-- **PBS maintenance results appear per datastore.** Backup servers shows garbage collection, prune, verify and sync outcomes. Status reads now use PBS run-status endpoints, and unclear unrelated task entries no longer mark every datastore unknown.
-- **Confirmed backups retain their status.** A freshly read backup can remain protected when another storage location is unreadable. Coverage names the affected node, storage, datastore or namespace without treating retained entries as fresh confirmation.
-- **Backups follows a clearer order.** Health comes first, then backup servers and restore points. The restore-points-per-day chart is removed, and metric bars use the dashboard’s usual colours.
-- **Backup rows identify the right location.** Server rows lead with datastore names, with each PBS server named once. Restore-point and Coverage rows stay compact and distinguish connections when names overlap. Host figures remain in the server drawer.
-- **Ceph manager counts include standby daemons.** Healthy standby managers count as running for API and agent readings. Critical status filters also recognise ERR.
-- **Ceph recovery clears old incident details.** A current healthy report no longer inherits warnings from an older or lower-priority source.
-- **Node alerts match the affected node.** Proxmox rows include their own physical-disk and local-storage alerts without taking on a guest’s alert just because its node name matches.
-- **Node details include power readings.** Proxmox hardware details show power measurements from the linked agent and label retained readings as last known.
-- **HTTPS redirect recovery is documented.** Help explains how to correct an existing Proxmox or PBS connection to direct HTTPS while retaining credentials, certificate verification and saved fingerprints.
+- **PBS job results match the datastore.** Backup servers shows GC, prune, verify and sync outcomes from PBS run-status endpoints. Plain error text counts as failure, and unclear unrelated tasks no longer make every datastore unknown.
+- **Confirmed backups retain their status.** A freshly read backup can remain protected when another storage is unreadable. Coverage names affected nodes, storage, datastores or namespaces without treating retained entries as fresh confirmation.
+- **Backups opens on Coverage.** The page orders health, backup servers and restore points, removes the restore-points-per-day chart, and uses the dashboard's usual metric-bar colours.
+- **Backup servers are counted and grouped.** Each PBS server has one header with reachability and datastore count. Datastore names lead the rows, and host figures remain in the server drawer.
+- **Backup lists stay compact and distinct.** Restore-point and Coverage rows remain on one line and distinguish connections when names overlap.
+- **Ceph counts and health are more accurate.** Standby managers count as running for API and agent readings. Critical filters recognise ERR, and healthy reports stop inheriting old incident details.
+- **Node alert colours match the affected node.** Proxmox rows include their own physical-disk and local-storage alerts without borrowing a guest's alert because its node name matches.
+- **Node details include power readings.** Proxmox hardware details show measurements from the linked agent and label retained values as last known.
+- **HTTPS recovery keeps saved connections intact.** Help explains how to correct a Proxmox or PBS endpoint to direct HTTPS while preserving credentials, certificate verification and saved fingerprints.
## TrueNAS, vSphere and Docker
- **Namesake TrueNAS systems stay separate.** Connections with the same hostname no longer merge into one displayed machine.
-- **TrueNAS history stays with the selected system.** Drawer requests read only that appliance, discard results from replaced connections and retain native disk I/O beside stored CPU history without claiming one source supplied everything.
-- **TrueNAS settings ignore outdated replies.** Tests, saves and access errors remain tied to the account and editor that requested them, including when a draft changes.
-- **Docker health indicators update live.** Container row dots follow refreshed health results instead of retaining the first healthy state.
+- **TrueNAS history stays with the selected system.** Drawer requests query only that appliance and discard replaced-connection results. Native disk I/O remains beside stored CPU history, with accurate source labels.
+- **TrueNAS and vSphere settings ignore outdated replies.** Connection tests, saves and access errors stay with the account, workspace and editor that requested them. Failed saved-configuration reads offer recovery.
+- **Docker health indicators update live.** Container row dots follow refreshed health results instead of retaining the initial healthy state.
- **Compose respects the image healthcheck.** Bundled and installer-generated Compose files use the HTTPS-aware image probe and preserve its failure result.
## Install, updates and agents
-- **Failed startup means failed installation.** The agent installer returns an error when startup verification fails, even if files were installed, and points to service logs.
-- **Invalid server URLs leave existing installs intact.** The downloaded agent checks URL policy before the installer replaces services or credentials. Certificate bypass does not silently permit plain HTTP.
-- **NAS installation handles older utilities.** URL, hostname and fingerprint processing works with older NAS versions of the text tools used by the installer.
-- **Incomplete installer downloads are not executed.** The container-agent wrapper downloads the complete script into a private file and checks its syntax before running it.
-- **Command-access setup preserves installed credentials.** Changing the command-execution choice prepares a new token without revoking earlier tokens that agents may still use. Concurrent requests no longer mint the same credential twice.
-- **Agent updates retain clearer outcomes.** Binary replacement and persistent NAS copies are made more durable. Failed recovery no longer reports an old update as newly completed.
-- **Agent Doctor stays current while open.** Expanded rows update their version, diagnosis and repair commands when fresh fleet information arrives.
-- **Helper updates have explicit instructions.** Help and agent logs explain that automatic updates replace the collector only. Updating the privilege helper requires the agent installer’s update flow.
-- **Uninstall failures describe stopped services.** When server removal cannot be confirmed, the installer distinguishes retained files from running monitoring and explains how to reconcile the existing registration.
-- **Compose setup separates image choices from authentication.** The environment example now explains which values Compose forwards, where runtime settings belong and why legacy token variables do not create v6 tokens.
+- **Failed startup means failed installation.** The agent installer returns an error when startup verification fails, even if files were installed, and directs you to service logs.
+- **Invalid server URLs leave existing installs intact.** The downloaded agent checks URL policy before services or credentials are replaced. Certificate bypass does not silently permit plain HTTP.
+- **NAS installation handles older utilities.** URL, hostname and fingerprint processing works with older NAS text tools.
+- **Incomplete installer downloads are not executed.** The container-agent wrapper downloads the whole script into a private file and checks its syntax before running it.
+- **Command-access setup preserves installed credentials.** Changing command execution prepares a new token without revoking tokens agents may still use. Concurrent requests no longer create the same install credential twice.
+- **Agent updates report more reliable outcomes.** Binary replacement and persistent NAS copies are more durable. Failed recovery no longer presents an earlier update as newly completed.
+- **Agent Doctor refreshes expanded rows.** Version, diagnosis and repair commands update as fresh fleet information arrives.
+- **Privilege-helper updates have explicit instructions.** Help and agent logs explain that automatic updates replace only the collector. Updating the helper requires the agent installer's update flow.
+- **Uninstall failures describe stopped services.** If server removal is unconfirmed, the installer distinguishes retained files from running monitoring and explains how to reconcile the existing registration.
+- **Compose setup separates image choice from authentication.** The environment example explains which values Compose forwards, where runtime settings belong and why legacy token variables do not create v6 tokens.
+- **Community-scripts LXC installs are recognised again.** Pulse detects the newer update helper so these installations receive the appropriate update controls.
## Pulse Pro, AI and hosted
-- **Reporting preserves selected resources during outages.** A failed resource read offers retry instead of appearing as an empty scope. Previous-account results and downloads cannot populate the current reporting panel.
-- **Organization and billing actions require current reads.** Workspace switches and failed reads clear outdated details and controls. Sharing drafts cannot reuse a removed destination.
+- **Reporting distinguishes outages from empty scope.** Failed resource reads offer retry. Earlier-account selections, results and downloads cannot populate the current reporting panel, including after live updates reconnect.
+- **Organization and billing actions use current details.** Workspace switches and failed reads clear outdated controls. Sharing drafts cannot reuse removed destinations.
+- **Container update checks stay in the selected organization.** Infrastructure update lists, summaries and check commands use that organization's agents.
- **Audit administration stays with the current view.** Audit results, verification feedback and webhook actions discard replies from an earlier account, workspace or page.
-- **Patrol requires sensor-specific recovery evidence.** It no longer clears an individual sensor-temperature alert using a snapshot that contains no reading for that sensor.
+- **Patrol cannot clear sensor alerts without readings.** Individual sensor-temperature alerts remain open when the review snapshot contains no reading for that sensor.
+- **Retired organization monitors stay stopped.** Late requests cannot reopen their stores or restart monitoring during shutdown or runtime replacement.
+- **MSP evaluation instructions use the published bundle.** Setup guidance now points to the v6.7.0 provider bundle.
+- **MSP networking guidance explains overlapping subnets.** Separate workspaces still share the server's routes, so clients with identical private addresses need distinct reachable addresses or tunnel NAT.
+- **Pro terms name the current AI modes.** The terms describe Ask first, Safe auto-fix, Autopilot and connected external agents, including which modes can run fixes.
## Monitoring and service health
-- **Buffered reports no longer inflate rates.** Host and Docker disk and network rates use collection times instead of rapid delivery after reconnect. Libvirt CPU and I/O rates use their collection interval too. Invalid intervals remain unknown.
-- **System Logs shows connection loss.** Connecting, reconnecting and disconnected states replace a misleading Live label. Uncertain access clears buffered lines and provides an explicit reconnect path.
-- **Diagnostics results stay with their request.** Changing account, workspace or panel discards outdated results and feedback instead of displaying them in the new context.
-- **Linux sensor collection retains more detail.** Agents report individual temperature and fan inputs with stable identities, including stopped fans and invalid readings. Sensor thresholds remain off until configured.
+- **Buffered reports no longer inflate rates.** Host and Docker disk and network rates use collection times after reconnect. Libvirt CPU and I/O rates use their collection interval too. Invalid intervals remain unknown.
+- **System Logs shows connection loss.** Connecting, reconnecting and disconnected states replace a misleading Live label. Uncertain access clears buffered lines and offers an explicit reconnect.
+- **Diagnostics discards outdated results.** Switching account, workspace or panel clears earlier results and feedback.
+- **Linux sensor collection retains more detail.** Agents report individual temperature and fan inputs with stable identities, including stopped fans and invalid readings. New sensor alerts stay off until configured.
## Security
-- **Sign-out confirms the Pulse session ended.** Proxy logout navigation waits for local confirmation and browser cleanup. Failed confirmation or incomplete cleanup produces a visible warning, and the previous live connection is closed.
-- **Private reads stop when access ends.** Metric and availability history, audit data, configuration exports, reports, support bundles and protected progress streams recheck the requesting session or token before sending further data.
-- **Settings actions stay with the active account.** API tokens, SSO, User Access and workload metadata discard outdated replies and confirmations after navigation or access changes. Returning requires a fresh read.
-- **Support-bundle summaries expose less configuration.** System information uses a limited diagnostic summary and selected environment fields, with the actual Pulse, Go and operating-system versions. Included logs still need private handling.
-- **API clients refuse credential-bearing redirects.** CLI, MCP and external-agent requests stay within the configured Pulse origin instead of forwarding credentials to another scheme, host or port.
-- **Saved credentials use safer file writes.** Configuration, bootstrap and billing saves avoid reusing stale temporary or backup paths that could expose secrets or overwrite other files.
-- **Assistant responses remain non-executable.** Protected JSON responses reject unsafe response formats, and duplicate-import checks better protect retained passphrase fingerprints.
-- **Token reveals keep current copy feedback.** Replacing or dismissing a reveal clears outdated dialog feedback. Help explains that closing the dialog does not revoke the token or erase clipboard copies.
-- **Network dependencies address security advisories.** The updated Go networking dependency fixes HTML parser and renderer vulnerabilities.
+- **Sign-out confirms the Pulse session ended.** Proxy logout navigation waits for local confirmation and browser cleanup. Failed confirmation or cleanup shows a warning, and the previous live connection closes.
+- **Private responses stop when access ends.** History, audit data, configuration exports, reports, support bundles, Assistant results and protected progress streams recheck the requesting session or token before sending more data.
+- **Browser results stay with the active account.** Resource lists, agent snapshots, API tokens, SSO, User Access and workload metadata discard outdated replies and confirmations after navigation or access changes.
+- **Support-bundle summaries expose less configuration.** System information includes a limited diagnostic summary, selected environment fields and actual software versions. Included logs still need private handling.
+- **API clients refuse credential-bearing redirects.** CLI, MCP and external-agent requests stay within the configured Pulse scheme, host and port instead of forwarding credentials elsewhere.
+- **Saved credentials use safer file writes.** Configuration, bootstrap and billing saves avoid stale temporary or backup paths that could expose secrets or overwrite other files.
+- **Protected responses reject executable formats.** Assistant results and private downloads enforce their expected response formats. Duplicate-import checks also better protect retained passphrase fingerprints.
+- **Token reveals keep copy feedback current.** Replacing or dismissing a reveal clears outdated feedback. Help explains that closing the dialog neither revokes the token nor clears clipboard copies.
+- **Security dependencies are updated.** Go networking fixes address HTML parser and renderer vulnerabilities, alongside updates to authentication and other runtime libraries.
## Other improvements
-- **Registration guidance matches authentication requirements.** Help distinguishes public discovery from protected registration and directs users to scoped tokens created in Settings.
-- **API examples keep body secrets private.** Guidance uses private request files for passwords and tokens instead of placing them in shell arguments or history.
-- **Recovery guidance preserves authentication.** Help prefers authenticated configuration transfers and explains that direct-loopback recovery creates a client-bound session rather than disabling authentication for everyone.
+- **Registration guidance matches authentication requirements.** Help distinguishes public discovery from protected registration and directs users to scoped tokens in Settings.
+- **API examples keep body secrets private.** Password and token examples use private request files instead of shell arguments or history.
+- **Recovery guidance preserves authentication.** Help prefers authenticated configuration transfers and explains that direct-loopback recovery creates a client-bound session.
+- **Wrong agent-to-node links have recovery instructions.** Help documents the existing unlink API so separate machines can keep reporting without removal or reinstallation.
+- **Organization recovery avoids duplicate changes.** Help explains how to reload saved details, access and sharing after failed reads or uncertain saves without recreating missing-looking configuration.
+- **Missing Roles controls have clearer guidance.** Help distinguishes licence, Pro runtime and administrator requirements without recommending broader access or a data reset.
## Known issues
-- **Guest freezes and Windows restarts remain unresolved.** Related fixes do not establish recovery for every guest. If monitoring was stopped after a guest froze, do not restart it solely to test this beta or this release.
-- **Affected backups still need precautions.** Stop Pulse server monitoring before a freeze-enabled backup. Keep it stopped until independent checks confirm thaw, fresh writes on every backed-up filesystem and workload liveness. An OK task does not prove thaw.
-- **Kubernetes pauses need controller and storage checks.** Confirm retained server storage before stopping pods and prevent controllers from replacing them. Deleting a pod or stopping only the collector does not confirm server polling has stopped.
-- **Windows disk readings can remain unavailable.** Keep using the existing working Machines disk view. Do not remove or reinstall agents, restart guests or force guest-agent checks solely to recover Proxmox disk readings.
-- **Elevated server memory remains unresolved.** Keep the working setup and report elevated steady usage or continued growth from existing readings at comparable uptimes and workloads. These changes do not establish a lower baseline.
+- **Guest freezes and Windows restarts remain unresolved.** Related fixes do not establish recovery for every guest. If monitoring was stopped after a guest froze, do not restart it solely to test this beta or release.
+- **Affected backups still need precautions.** Stop Pulse server monitoring before freeze-enabled backups. Keep it stopped until independent checks confirm thaw, fresh writes on every backed-up filesystem and workload liveness. An OK task does not prove thaw.
+- **Kubernetes pauses need storage and controller checks.** Confirm persistent server storage before stopping pods and prevent controllers replacing them. Deleting a pod or stopping only the collector does not confirm server polling stopped.
+- **Windows disk readings may remain unavailable.** Keep using the working Machines disk view. Do not remove or reinstall agents, restart guests or force guest-agent checks solely to recover Proxmox disk readings.
+- **Elevated server memory remains unresolved.** Keep the working setup. Use existing readings at comparable uptimes and workloads when reporting high steady usage or growth. These changes do not establish a lower memory baseline.
- **Previously affected hosts still need caution.** If agents or Patrol were disabled after a host lockup, keep that setup. Preserve existing evidence privately without re-enabling them, changing GPU load or altering hardware to recreate the incident.
## Before you upgrade
diff --git a/docs/releases/V6_CHANGELOG_v6.8.0-beta.1.md b/docs/releases/V6_CHANGELOG_v6.8.0-beta.1.md
index bc17e4b9cf..5fb66992ef 100644
--- a/docs/releases/V6_CHANGELOG_v6.8.0-beta.1.md
+++ b/docs/releases/V6_CHANGELOG_v6.8.0-beta.1.md
@@ -4,6 +4,31 @@ This changelog describes the changes since `v6.7.0`. Stable `v6.7.0` remains the
## Changes since v6.7.0
+- Alert on failed PBS datastore jobs, apart from guest health (#3826)
+- docs(msp): stop promising overlapping subnets route themselves (#3832)
+- Document safe organisation-switch recovery (#3831)
+- Resolve the request organization for infra updates (#3828)
+- Explain missing RBAC controls without widening access (#3827)
+- Stop tests from adding new source-text reads (#3825)
+- Reserve both loopback addresses for notification fallback tests (#3824)
+- Group Backup servers by server, with a count and header-only status (#3790)
+- Run the shared template audit instead of restating it in tests (#3823)
+- Recognise community-scripts LXC installs again (#3822)
+- Open Backups on Coverage and name health reasons plainly (#3821)
+- Publish the monitor's discovery service under its lock (#3820)
+- Explain non-destructive recovery of a wrong agent node link (#3819)
+- Keep VMware settings requests with their admitted reader (#3818)
+- Refresh installer slice costs from complete current CI (#3817)
+- Reconcile the public Go patch graph and its safety floors (#3816)
+- Compose the canonical Markdown and router maintenance graph (#3815)
+- Preserve whole-thread affected versions during issue intake (#3814)
+- Exercise withdrawn readers across unified resource queries (#3813)
+- Refresh governed Node and Go container build inputs together (#3812)
+- Protect TrueNAS reader isolation through the real API transport (#3807)
+- Keep explicit no-topic replies out of decomposition triage (#3808)
+- Exercise authority withdrawal across agent snapshot reads (#3806)
+- Release v6.8.0-beta.1 (#3805)
+- Compress public frontend CI transfer without changing its identity (#3804)
- Cover availability history authority through delayed body reads (#3803)
- Cover linked-agent power in the Proxmox node drawer (#3802)
- Preserve image healthcheck failures in Docker Compose (#3801)