Backups keep failing verification

Solus

Member
Sep 18, 2024
1
0
6
I'm trying to get PBS working smoothly but I keep running into a snag. I'm able to backup the VM/CT's fine, but it occurs often that the backup can't be verified. I've tried removing the backups and taking new ones with no change, they keep failing. One just keeps failing the initial backup verification.

I read that it could be corrupt VM disks, so I cloned the disk to a second on I added, all on the same VM, removed the old disk and tried then. It appeared to have worked, at least on the initial backup. Then, subsequent backups after that initial one just keeps failing to be verified.

I'm at a loss on what I should do, is my Proxmox VE nodes (there are two) corrupted in some way? (It wouldn't surprise me I guess, I messed with it a lot when setting it up)
1795028818094.png

1795028844421.png

Let me know what information you need from me to help understand what's going on! I'd like to learn what I'm doing wrong so I can prevent it in the future...

This is what I get for that task...

Code:
2026-11-18T13:37:35-05:00: verify External:vm/102/2026-11-18T18:35:38Z
2026-11-18T13:37:35-05:00:   check qemu-server.conf.blob
2026-11-18T13:37:35-05:00:   check drive-scsi0.img.fidx
2026-11-18T13:48:27-05:00:   verified 57346.65/114060.00 MiB in 651.48 seconds, speed 88.03/175.08 MiB/s (3 errors)
2026-11-18T13:48:27-05:00: verify External:vm/102/2026-11-18T18:35:38Z/drive-scsi0.img.fidx failed: chunks could not be verified
2026-11-18T13:48:27-05:00: Failed to verify the following snapshots/groups:
2026-11-18T13:48:27-05:00:     vm/102/2026-11-18T18:35:38Z
2026-11-18T13:48:27-05:00: TASK ERROR: verification failed - please check the log for details
 
Last edited:
I'd start with the verify task on the PBS host that owns External. The excerpt narrows it down to chunks referenced by drive-scsi0.img.fidx. In the full task log, look between that archive's check line and the summary with the error count. The detail there can distinguish a failed chunk read from a check that failed after the chunk was read. Could you share the first detailed error and a few lines around it, if present? PBS verify code

New snapshots in the same datastore can refer to chunks already stored there, so seeing the same problem after a fresh backup can still point to the datastore. If the log identifies the chunks, compare the IDs across failed snapshots. If it reports an I/O error, check the PBS host's storage logs at that timestamp. Those details would help narrow it down before changing the VM disk again. PBS technical overview