Verify jobs
- Thread starter TwiX
- Start date
Hi,
A verification guarantees that the backup data is still all there and valid and could be, at time of verification, restored to the same state it got backed up.
That may not be the same then a restore resulting in a working VM/CT setup, e.g., if one missed backing up one of the virtual guest disk, or if the storage of the virtual guest was already corrupted when the data got backed up.
Those risks are unrelated to Proxmox Backup Server, and it cannot really check that.
That's why periodic end-to-end restore tests will always have value (untested backups are more a gamble than anything else).
A verification guarantees that the backup data is still all there and valid and could be, at time of verification, restored to the same state it got backed up.
That may not be the same then a restore resulting in a working VM/CT setup, e.g., if one missed backing up one of the virtual guest disk, or if the storage of the virtual guest was already corrupted when the data got backed up.
Those risks are unrelated to Proxmox Backup Server, and it cannot really check that.
That's why periodic end-to-end restore tests will always have value (untested backups are more a gamble than anything else).
Some of that may be automated, that also makes it easier to do it constantly periodically.
But yeah, even then doing it for all snapshots is overkill and unpractical, so spot checking (random samples) is normally a good trade off which will catch most issues (especially if VMs and Proxmox VE setups are relatively homogenous).
When doing it manually one could, for example, restore a handful of recent snapshots once a week or even month (better than nothing), boot them (ensure network is disabled or isolated, to avoid interruption with production) and check if all comes up and seems OK.
Automatic would be script using the API for the restore and some automatic check if the VM lives, qemu guest-agent could be of help here.
Last edited: