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.

Explain Proxmox VM and linked-agent I/O sources

pulse-triage[bot]e744bce8d4b5
diff --git a/docs/METRICS_HISTORY.md b/docs/METRICS_HISTORY.md
index 37a0dfe404..0f011dd1c8 100644
--- a/docs/METRICS_HISTORY.md
+++ b/docs/METRICS_HISTORY.md
@@ -32,6 +32,11 @@ establish the observations in the selected stored range. For a PBS host, use
the [Backup Server chart](PBS.md#pbs-is-connected-but-history-stays-empty),
not its datastore or the VM that happens to host it.
+For a Proxmox VM, linking a Pulse Agent does not select agent I/O for the VM's
+live or stored Disk I/O chart. Passed-through storage can be busy while the
+Proxmox VM chart is quiet; compare the [VM and agent measurement sources](VM_DISK_MONITORING.md#vm-disk-io-is-quiet-but-linked-agent-disks-are-busy)
+before treating that trace as an idle guest.
+
Start with the existing chart, not a restart or a storage change. A current
reading does not prove that it was stored, and a selected time range does not
prove continuous coverage. **Window start** and **Window end** bound the
diff --git a/docs/VM_DISK_MONITORING.md b/docs/VM_DISK_MONITORING.md
index 233c6cec85..26c0291ffe 100644
--- a/docs/VM_DISK_MONITORING.md
+++ b/docs/VM_DISK_MONITORING.md
@@ -4,6 +4,9 @@ Pulse can read filesystem usage inside a Proxmox VM through the QEMU Guest
Agent. Allocated virtual-disk size is not the same as used space inside the
guest. LXC filesystem collection uses a different path.
+For a quiet VM **Disk I/O** chart alongside busy linked-agent disks, see
+[which readings the VM chart uses](#vm-disk-io-is-quiet-but-linked-agent-disks-are-busy).
+
## Backup safety
**Do not run manual guest-agent probes during a backup, freeze or thaw.** A
@@ -319,6 +322,38 @@ custom monitoring role if the required read privileges are absent.
## 🔧 Troubleshooting
+### VM Disk I/O is quiet but linked-agent disks are busy
+
+**Filesystem usage and Disk I/O measure different things.** Filesystem usage
+is occupied space, not read/write throughput or disk busy time. A linked
+**Pulse Agent** can supply filesystem usage and disk inventory, but that does
+not select the source of the VM's I/O charts.
+
+The Proxmox VM's **Disk I/O** chart uses the VM counters reported by Proxmox.
+Linking a Pulse Agent does not switch the VM’s live Disk I/O or stored Disk I/O
+history to that agent. There is no VM I/O source selector in this build.
+
+With USB storage or a PCIe HBA/NVMe device passed directly into the VM, storage
+activity can bypass QEMU's virtual-block-device accounting. A quiet Proxmox
+VM chart therefore does not establish that the guest’s disks are idle. This
+differs from a virtual SCSI/VirtIO disk whose traffic Proxmox can account for;
+the two views are not necessarily measuring the same devices.
+
+For that activity, use readings already available in the **Pulse Agent-backed
+machine or per-disk view**, or the guest's established storage monitoring.
+Compare the same guest, device, observation time and
+[History range](METRICS_HISTORY.md#read-a-history-panel). A temperature,
+SMART result or filesystem reading alone does not establish current I/O.
+Do not add Proxmox and agent totals, or sum partitions, RAID members and
+logical devices as if they were independent disks: the same traffic can be
+counted more than once.
+
+Use existing observations, not a new I/O stress test, guest-agent probe,
+restart or hardware reconfiguration. Do not widen token permissions or install
+another agent to change this chart's source. If a backup or unresponsive guest
+is involved, [Backup safety](#backup-safety) takes precedence; a busy or quiet
+chart does not establish thaw or successful workload writes.
+
### Current, retained and unavailable disk readings
In **Workloads**, select the disk reading indicator for its explanation, or
diff --git a/frontend-modern/public/docs/METRICS_HISTORY.md b/frontend-modern/public/docs/METRICS_HISTORY.md
index 37a0dfe404..0f011dd1c8 100644
--- a/frontend-modern/public/docs/METRICS_HISTORY.md
+++ b/frontend-modern/public/docs/METRICS_HISTORY.md
@@ -32,6 +32,11 @@ establish the observations in the selected stored range. For a PBS host, use
the [Backup Server chart](PBS.md#pbs-is-connected-but-history-stays-empty),
not its datastore or the VM that happens to host it.
+For a Proxmox VM, linking a Pulse Agent does not select agent I/O for the VM's
+live or stored Disk I/O chart. Passed-through storage can be busy while the
+Proxmox VM chart is quiet; compare the [VM and agent measurement sources](VM_DISK_MONITORING.md#vm-disk-io-is-quiet-but-linked-agent-disks-are-busy)
+before treating that trace as an idle guest.
+
Start with the existing chart, not a restart or a storage change. A current
reading does not prove that it was stored, and a selected time range does not
prove continuous coverage. **Window start** and **Window end** bound the
diff --git a/frontend-modern/public/docs/VM_DISK_MONITORING.md b/frontend-modern/public/docs/VM_DISK_MONITORING.md
index 233c6cec85..26c0291ffe 100644
--- a/frontend-modern/public/docs/VM_DISK_MONITORING.md
+++ b/frontend-modern/public/docs/VM_DISK_MONITORING.md
@@ -4,6 +4,9 @@ Pulse can read filesystem usage inside a Proxmox VM through the QEMU Guest
Agent. Allocated virtual-disk size is not the same as used space inside the
guest. LXC filesystem collection uses a different path.
+For a quiet VM **Disk I/O** chart alongside busy linked-agent disks, see
+[which readings the VM chart uses](#vm-disk-io-is-quiet-but-linked-agent-disks-are-busy).
+
## Backup safety
**Do not run manual guest-agent probes during a backup, freeze or thaw.** A
@@ -319,6 +322,38 @@ custom monitoring role if the required read privileges are absent.
## 🔧 Troubleshooting
+### VM Disk I/O is quiet but linked-agent disks are busy
+
+**Filesystem usage and Disk I/O measure different things.** Filesystem usage
+is occupied space, not read/write throughput or disk busy time. A linked
+**Pulse Agent** can supply filesystem usage and disk inventory, but that does
+not select the source of the VM's I/O charts.
+
+The Proxmox VM's **Disk I/O** chart uses the VM counters reported by Proxmox.
+Linking a Pulse Agent does not switch the VM’s live Disk I/O or stored Disk I/O
+history to that agent. There is no VM I/O source selector in this build.
+
+With USB storage or a PCIe HBA/NVMe device passed directly into the VM, storage
+activity can bypass QEMU's virtual-block-device accounting. A quiet Proxmox
+VM chart therefore does not establish that the guest’s disks are idle. This
+differs from a virtual SCSI/VirtIO disk whose traffic Proxmox can account for;
+the two views are not necessarily measuring the same devices.
+
+For that activity, use readings already available in the **Pulse Agent-backed
+machine or per-disk view**, or the guest's established storage monitoring.
+Compare the same guest, device, observation time and
+[History range](METRICS_HISTORY.md#read-a-history-panel). A temperature,
+SMART result or filesystem reading alone does not establish current I/O.
+Do not add Proxmox and agent totals, or sum partitions, RAID members and
+logical devices as if they were independent disks: the same traffic can be
+counted more than once.
+
+Use existing observations, not a new I/O stress test, guest-agent probe,
+restart or hardware reconfiguration. Do not widen token permissions or install
+another agent to change this chart's source. If a backup or unresponsive guest
+is involved, [Backup safety](#backup-safety) takes precedence; a busy or quiet
+chart does not establish thaw or successful workload writes.
+
### Current, retained and unavailable disk readings
In **Workloads**, select the disk reading indicator for its explanation, or