[SOLVED] VM stops working if backup is started

[André Koppany]

Dear PBS Users,

we are using the PBS Server (running on a separate Hardware) to backup VMs running on our 3 node cluster. (Backup Mode is Snapshot, Compression ZSTD)
On each backup run after the backup of a vm did start, the VM stopped working. Means the CPU rate rised to 25% percent, and the VM does not respond anymore (no login possible).

I did read that maybe disabling the qemu-agent will be a solution. We did this, but the last VM with the described problems did not have Qemu-agent enabled.

Here is the output of the backup task:

ERROR: VM 111 qmp command 'backup' failed - got timeout
INFO: aborting backup job
ERROR: VM 111 qmp command 'backup-cancel' failed - got timeout
ERROR: Backup of VM 111 failed - VM 111 qmp command 'backup' failed - got timeout
INFO: Failed at 2021-02-04 00:13:36

On this VM we are using Ubuntu 20.04.1 LTS.

We also get a subscription of the PBS to get the latest version 1.0-6 which was also updated to yesterday. But still the error did occur.

Any Ideas how we can solve this problem?
This way we cannot use the PBS because the danger is high that each night where the backup is running another VM stops working.

I am happy to provide you more system details if you need to.

Best,
André

 

[t.lamprecht]

Hi,

is this still an issue? There was a pve-qemu-kvm package update moved to the enterprise repository just the day you posted this here.

If this is still a problem please also post details about:
* storage used form VMs and/or Backup
* server HW (especially CPU model)

Here is the output of the backup task:

Is that the full output?

On this VM we are using Ubuntu 20.04.1 LTS.

Can you please also post its config?

 

[André Koppany]

hi,

we updated the pve-qemu-kvm package and restarted the VMs and the problem did not appear again.

Think it could be marked as solved.

Best,
André