Critical: All Proxmox Backups Missing After Node Failure (NFS/QNAP)

L

lakshmi prasad

New Member
Apr 5, 2026
6
0
1
Hi Team,

I need help understanding a critical issue in my Proxmox VE cluster.


Setup:

3-node cluster
15+ infrastructure VMs hosted across the nodes
Shared storage: QNAP via NFS
Nightly incremental VM backups to the same NFS storage

Issue:

One node had a disk failure and went offline overnight

Cluster became unstable

Next day, all backup files are missing (backup folder is completely empty)

Important Points:

No manual deletion was done

Configuration looks correct

QNAP storage is working fine for other data

No clear logs found related to deletion

Questions:

Can Proxmox delete all backups automatically in any scenario?

Can a node failure or cluster issue trigger backup removal/pruning?

Could an NFS mount issue cause Proxmox to see/delete backups?

What logs should I check to identify what happened?

Has anyone seen similar behavior?



Any help to identify the root cause would be greatly help us.
 
Hmm, thats not recommend. use local ssd storage for Proxmox BS Datasets not a nfs share.
Can you look on your Qnap Dataspace direct without the NFS share or setup a second Samba CIFS share to see the over 10^6 files in the Proxmox BS folder?
 
can your Proxmox BS USER, delete data on the Proxmox BS Dataset?
Do you run ZFS over all nodes and on the Proxmox BS?
 
Last edited:
We are not using Proxmox Backup Server (PBS); backups are stored via Proxmox VE on a QNAP NFS share.
I checked directly on the QNAP storage, and the backup files are not present there either.

Since we had an NFS disconnection earlier, we suspect that the storage appeared empty during a backup run and the retention policy may have removed the existing backups.
Can you confirm if this behavior is expected in such scenarios?
 
No you don't! look to the Promox VE WIKI for that fact!

I asume
* You have only ONE mountpoint to only ONE NFS share on your QNAP
* on all of your nodes (is locking actice?), then one or more NODE lost the connection.
* The mountpoint, example /mn/nfs-share, will be reset (unmount) and point to your local harddrive AND the local filesystem is now EMPTY.
* The replication over your cluster begin and 0 files -- file deletion -- will replicated to the other nodes.
* So you will lost all of our data in this shared not locked folder on your QNAP.

I'm done.
we have sunday in germany and "Osterfest".
good luck.
 
news said:
No you don't! look to the Promox VE WIKI for that fact!

I asume
* You have only ONE mountpoint to only ONE NFS share on your QNAP
* on all of your nodes (is locking actice?), then one or more NODE lost the connection.
* The mountpoint, example /mn/nfs-share, will be reset (unmount) and point to your local harddrive AND the local filesystem is now EMPTY.
* The replication over your cluster begin and 0 files -- file deletion -- will replicated to the other nodes.
* So you will lost all of our data in this shared not locked folder on your QNAP.
Click to expand...
Thanks for the detailed explanation, this aligns closely with what we observed.


We do have a single NFS share from QNAP mounted across all nodes, and we experienced an NFS disconnection.
After checking directly on QNAP, the backup files are no longer present.


It seems likely that when the NFS mount dropped, the mount point fell back to an empty local directory, and subsequent backup/retention operations treated the storage as empty, leading to deletion of existing backups.


Could you please confirm if Proxmox VE would trigger retention cleanup in such a scenario when the storage appears empty?
 
You must log in or register to reply here.
Top Bottom
Back