backup Server restore loops forever restoring VM

R

ronalddevos

New Member
Feb 9, 2025
4
1
3
I have the following problem:
I have a Windows 2019 Server VM with 4 disk (175g plus 3x 150G) that I have backed up to 1) ProxMox Backup Server and 2) a NAS device (QNAP) over SMB/CIFS
When I restore the VM from the NAS Device it works OK, without any problems.
When I restote the VM from the ProxMox Backup Server it restores and when the restore is finished it restart the restore from scratch and never finsihes (i.e. it keeps on restarting the restore).
Running ProxMox 9.2.1 and Backup Server 4.2.1 It had the same problem on the previous version (3.4.x I believe)
Restoring other VMs works fine.
Any suggestions? Any information I can provide?
 
You reported the same thing in February: https://forum.proxmox.com/threads/restore-loop.181059/ Sviat suggested there that with 4 disks it may be one restore per disk. That is still the most likely explanation, and the task log will settle it.

Could you post the full restore task log? Task History on the PVE node, or /var/log/pve/tasks/.

Each disk is restored by its own pbs-restore command, and each one counts its own progress from 0 to 100, so a four-disk VM shows the counter run four times. In the log every pass begins with a line like:

restore proxmox backup image: /usr/bin/pbs-restore --repository ... drive-scsi0.img.fidx ...

If the drive name changes on each pass, the restore is working normally and simply takes four times as long as your single-disk VMs. If the same drive name repeats, it is a real loop and the log will show that.
 
Thanks for the answer.
Yes, you are absolutely right, it takes 1 restore per disk.
The reason for my confusion is that, when I do the restore from the QNAP NAS (SMB/CIFS) it all restores in 1 pass, not 1 restore per disk.....
Logs available, if you want.
 
The difference is the backup format. A vzdump backup to the QNAP is a single archive (vma or tar) that holds all four disks, so the restore reads it once from beginning to end and you get one progress run. PBS stores each disk as its own separate index, so PVE calls pbs-restore once per disk and each call reports its own progress.

One thing worth confirming: your first post said it never finishes. Does the task now complete after the fourth pass? If it keeps going beyond that, the log would be useful after all.
 
You must log in or register to reply here.
Top Bottom
Back