vzdump with snapshot hangs if a VM reboots during the backup process?

[RodinM]

Hi, the situation is this. During the backup process, when the snapshot method is used, the process may hang (while the VM itself functions properly more or less). I noticed that it happens when a VM (Windows guest only. It NEVER happened with Linux KVM guests) reboots. This occurs when I VM installs new updates and automatically reboots. I suspected this once when I logged on to the Windows KVM guest and noticed the fact the machine rebooted. After I logged off I switched to the proxmox terminal of the server and noticed that the backup process immediately continued (vzdump........................dat folder in the backup directory began to grow in accordance with the size of the KVM guest) and finished succesfully.
Is that a bug or am I in the wrong way? I'm concerned about it because I can't monitor the backup process daily.

 

[dietmar]

Maybe you storage is a bit too slow (snapshots slow down storage access): What is the output of

# pveperf

(please run when there is no load on the servers).

 

[RodinM]

It is complicated to do a check with no load on the servers. I can try it later. Now it shows the following (20 KVM guests on the storage):
CPU BOGOMIPS: 14932.37
REGEX/SECOND: 585656
HD SIZE: 2.21 GB (/dev/mapper/pve-root)
BUFFERED READS: 65.20 MB/sec
AVERAGE SEEK TIME: 4.57 ms
FSYNCS/SECOND: 59.84
DNS EXT: 94.85 ms
DNS INT: 0.85 ms

 

[RodinM]

Tha additional question to that topic is: how do I stop the vzdump process correctly if I suppose it hang?
Should I simply kill the processes with corresponding PIDs and issue the command qm unlock VMID?