VM Backups on NFS Storage

[Chris Thompson]

I have a ProxMox VE environment in which my primary storage is an NFS NAS. From this NAS, I backup the images folder containing my ProxMox KVM VM's. I'm not using the inbuilt Proxmox Backup feature.

I'm curious what the scenario looks like should I ever need to do a one-off restore of a VM and/or a full bare metal restoration using my current setup. Could it be as simple as copying the contents of the backed up images folder back to the NFS share?

Thanks in advance for any thoughts in this regard.

Best Regards,
Chris

 

[wolfgang]

Hi,

general VM images are a container which is used as a block device.
This means it is nearly the same as every Harddrive, but it is not physical it is virtual.

If you have to restore the VM disk, you can copy the image back, and it should run.
But the problem is if you make the backup (copy the file) at VM runtime the image will not be consistent, and you can't boot the VM.

So if you make Backups on image file level be sure the VM is turned off.

 

[Vengance]

Hi,

general VM images are a container which is used as a block device.
This means it is nearly the same as every Harddrive, but it is not physical it is virtual.

If you have to restore the VM disk, you can copy the image back, and it should run.
But the problem is if you make the backup (copy the file) at VM runtime the image will not be consistent, and you can't boot the VM.

So if you make Backups on image file level be sure the VM is turned off.

The qemu guest agent should provide a higher consistency, right?
Would it still be necessary to shut down the VM?

 

[wolfgang]

The qemu guest agent should provide a higher consistency, right?

Yes, you can freeze the fs of the VM.
But you must freeze so long the backup process will run and in this time the VM is running but can't write what will make it normally unavailable.

Would it still be necessary to shut down the VM?

It depends on the VM, if it would be a read-only VM then it would be possible.
The most VM are not read-only because the logging is running.
So a common VM has to be turned off.