[SOLVED] Unable to restore PBS backup

[proxycat]

Dear all,

I'm trying to restore a VM backup (done with PBS). Unfortunatly it's not working.

I'm using the following command :

# qmrestore catalog-pbs:backup/vm/102/2026-06-13T17:00:18Z 7103 --storage local-zfs

where I'm restoring the backup to a new VM.

Indeed the process starts without error but hangs at the end :

starting to restore snapshot 'vm/102/2026-06-13T17:00:18Z'
download and verify backup index
fetching up to 16 chunks in parallel
...
progress 99% (read 29766975488 bytes, zeroes = 0% (247463936 bytes), duration 27 sec)
progress 100% (read 30064771072 bytes, zeroes = 0% (247463936 bytes), duration 27 sec)
restore image complete (bytes=30064771072, duration=27.58s, speed=1039.57MB/s)

From there I don't the prompt back. And on the GUI I can see a lock icon on the new VM.

I have no idea how to investigate further to be able to solve my issue.

Does anyone has any clue to solve or help investigate this ?

Note: I ended up interrupting the process after ~1 hour.

 

[j.theisen]

Hi @proxycat

thanks for posting in the forum!

may i ask is there a particular reason why you are using the CLI for this restore?
It shouldn't be relevant to the problem but just out of curiosity.

Apart from that, a few questions:

• Is this a universal problem for all the backups of your PBS and / or this PVE server or just this particular VM?
• What version of PVE and PBS are you running?
  pveversion -v
proxmox-backup-manager version --verbose
• Were there any other Tasks running during the restore?
• What user are you using to login? i.e. PAM user or PVE authentication server?

Yours sincerely
Jonas

 

[leesteken]

It's possible that you have consumer drives that slow down extremely during sustained writes and that the command is still waiting for the data to actually be written to the drive. I know that an hour seems too long but it's 28GiB and the drive might also be busy doing other things.
There have been other threads with question about why the restore process seems to freeze at 100% because of slow drives.

 

[proxycat]

Many thanks to both of you for your replies!

Hi @proxycat

thanks for posting in the forum!

may i ask is there a particular reason why you are using the CLI for this restore?
It shouldn't be relevant to the problem but just out of curiosity.

Apart from that, a few questions:

• Is this a universal problem for all the backups of your PBS and / or this PVE server or just this particular VM?
• What version of PVE and PBS are you running?
  pveversion -v
proxmox-backup-manager version --verbose
• Were there any other Tasks running during the restore?
• What user are you using to login? i.e. PAM user or PVE authentication server?

Yours sincerely
Jonas

We switched to the command line because we were hoping to have a more verbose output. However we realized it's the same we have with the web interface.

Now to answer your question in order.

1. Yes it's universal.

2. The PVE versions :

proxmox-ve: 9.2.0 (running kernel: 6.14.8-2-pve)
pve-manager: 9.2.2 (running version: 9.2.2/b9984c6d90a4bd80)

and the PBS version :

proxmox-backup                      4.2.0         running kernel: 6.17.2-2-pve
proxmox-backup-server               4.2.1-1       running version: 4.1.0

3. I'm not aware of executing any supplementary task.

4. We're using PAM authentication.

It's possible that you have consumer drives that slow down extremely during sustained writes and that the command is still waiting for the data to actually be written to the drive. I know that an hour seems too long but it's 28GiB and the drive might also be busy doing other things.
There have been other threads with question about why the restore process seems to freeze at 100% because of slow drives.

That's it ! We decided to wait. And it wasn't totally frozen but slow.

We were also tricked by this message :

restore image complete (bytes=30064771072, duration=29.10s, speed=985.29MB/s)

which made us think the process was completed but something hanged. If we wait enough time, the following is added to the output :

rescan volumes...

I'm going to mark this thread as solved if I can do it myself.

 

[Onslow]

I'm going to mark this thread as solved if I can do it myself.

Yes, as the author of the thread you can mark it as SOLVED by editing the title and from the drop-down menu selecting the [SOLVED] prefix.