Restoring a container can potentially remove it from Proxmox interface

[Sigfried]

Hi there

I'm not sure this is know or is suppose to be, I'd say this is a bug. But earlier today I had a storage failure. I found out that the pool was still online but something went wrong and reading and writing is no longer possible (Input/output errors).
This meant that one container which was using the storage was also no longer available.
I wanted to get that container back up, so I went through the Proxmox selected a backup, changed the storage to a working pool and clicked restore.

However, instead of actually restoring the container, it deleted it because the task failed. Here is the log:

recovering backed-up configuration from 'BACKUP-1:backup/vzdump-lxc-103-2021_06_20-01_19_28.tar.zst'
  Failed to parse thin params: Error.
  Failed to parse thin params: Error.
  Failed to parse thin params: Error.
  Failed to parse thin params: Error.
  device-mapper: message ioctl on  (253:4) failed: Input/output error
  Failed to process message "delete 2".
  Failed to suspend SSD/SSD with queued messages.
TASK ERROR: unable to restore CT 103 - lvremove 'SSD/vm-103-disk-0' error:   Failed to update pool SSD/SSD.

I assume it wanted to write some changes to the SSD storage pool (which was no longer available), crashed and then caused it to be removed from the Proxmox interface.
Obviously I manually restored it via CLI but it seems a bit odd that when it crashes it removes it from the interface. I would expect it to stop, or ignore the drive error on a restore and just restore it anyway on the other specified disk.

Is this really how it should be or is this a bug?

Regards
Sigfried

 

[Stefan_R]

A restore needs to remove the old instance first, this is true for both CTs and VMs. So yes, while not optimal, this is how it currently is intended to work.

Note that you shouldn't need the CLI for trying the restore again - you can select the storage that contains the backup in the tree on the left, select 'VM/CT backups' and start the restore from there, selecting the same VMID as before.

 

[Sigfried]

Hi there, sorry for my late reply.

IT makes sense that it needs to remove the image, but I found it a bit weird that when it fails to remove it, it also deletes the configuration. When it fails I would expect it to keep the configuration. Deleting the image before the configuration?

I just found this interesting.