Fault Description

In the PVE management interface overview, an alarm was found indicating a Ceph health warning.

Clicking the yellow icon provided further information: two OSD disks were full.

By clicking the exclamation mark to the right of the first line “2 nearfull osd (?)”, it was confirmed that disks numbered 5 and 7 were full.

Processing Steps

Identify which nodes the data-filled OSDs are located on, then delete non-running or unnecessary virtual machines on those nodes to free up disk space.

• Identify which nodes the disks are located on.

The two data-filled disks were located on two different physical nodes. Log into both nodes and use the command line `df -h` to check disk usage, further confirming the issue.

• In the Proxmox VE management interface, delete non-running virtual machines.

Manually input the IDs of the virtual machines to be deleted.

Result Confirmation

Log into any node, run the command `ceph health`, and if the output is OK, the issue is resolved. You can also check the cluster overview in the web management interface. If the health status icon is green, the issue is temporarily resolved.