TASK ERROR: unable to commit configuration file

[jpbartle]

Recently built 3 node, all based on 3.4-6/102d4547 (kernel 2.6.32-39). This as part of overall network upgrade, moving from a 2 node setup that had been in place for some years.

All 3 current nodes are using local storage for disk images. Have been able to restore and run a couple of the vms (Windows DC, Freepbx instance, Windows 8 instance). But having difficult in 2 areas. I've researched this extensively on forums - and would appreciate some tips for resolving.

1 - trying to restore some of the backup images from earlier PM setup results in "TASK ERROR: unable to commit configuration file '/etc/pve/nodes/pm3/qemu-server/120.conf'" This after system working thru 100 restore / space reduction msg (attached text file.)

2 - following this, I am seeing the disk image content in the server view, disk "local" selection -- but am not able to free up the files from the GUI. the REMOVE button being disabled. (screenshot attached.)

Any guidance on this appreciated

 

[t.lamprecht]

Hi,

1) We write the config to a temp file and then rename (move) it to the correct one, this guarantees atomicity, and in this case the rename fails, which may be the case if a VM/CT with the same VMID already exists, this would be my first guess.

2) This is by design. We want to avoid an accidental deletion of a disk file, which would be quite bad. Disk images must be deleted from the respective VMs Hardware Tab, where it first goes to unused and then can be finally deleted completely. In your case you probably cannot do that as there is no VM if the restore failed, then fallback to a manual deletion through the command line, but be sure what disk to delete.

 

[jpbartle]

Thomas --

I'm looking at the /etc/pve/nodes/ vicinity and seeing for each node files like
- 163.conf.130833.tmp
- 163.conf.148349.tmp
I believe these are indicating examples of failed restores. I've deleted all of the .tmp occurrences.

Looking at /var/lib/vz/images/ I'm seeing empty folders for these failed restores. I've deleted these.
On doing the above, I'm able to successfully restore the same backups that failed previously.
Understood about VMs hardware tab for VM removal. That makes perfect sense.

Thank you for your help sorting this out, Jonathan