[SOLVED] Restoring LXC from PBS fails

[fabian]

Again not sure how to do that. When downloading through the web ui (either from PBS or PVE) I get a zip file "root.pxar.didx.zip" downloaded to e.g. my laptop. How would I download directly to the PVE? And get the root.pxar instead of the oot.pxar.didx.zip?

just for completeness' sake - you hopefully won't need this in this instance

in the content tree of PBS, next to the 'root.pxar.didx' entry there are two buttons in the action column - the first downloads the full archive, the second opens the file browser where you can download files/directories. I meant the first one, which should give you a file 'root.pxar' which can be manipulated with the 'pxar' CLI tool.

 

[fabian]

http://download.proxmox.com/debian/...binary-amd64/libpve-common-perl_7.2-2_all.deb

available on pvetest now

 

[Hr. Schmid]

okay, that is good! I was afraid you'd run into a second related issue, but this looks like it's not the case.

can you try upgrading to libpve-common-perl 7.2-2 once that hits the repositories? it should fix the restore issue:

https://git.proxmox.com/?p=pve-common.git;a=commit;h=6647801cb374e75b48d488484893559ffcfb1203

if you feel comfortable, you can also apply the diff to /usr/share/perl5/PVE/Tools.pm (apt install --reinstall libpve-common-perl will revert to the old version if something goes wrong).

Applied the patch - and "Heureka" it works"

@fabian : Thanks a lot for your time, patience and effort!

One question still bothers me:
Is my backup strategy - backing up also the mounted volumes - the right one? Or is there a better solution, lets say backing up guest system and mounted volumes separately, If so, how could that be achieved?

 

[fabian]

backups (and restore) is currently "all or nothing" - you can exclude mountpoints from the backup, but then upon restore they will be empty (and if you restore over an existing container, the existing volume will be destroyed as part of the restore!). if you want to go down that route, you want to restore into a new (not-yet-existing) VMID, then remove the empty MP and re-assign the one containing data from the old CT to the newly restored CT. backing up the full container including all mountpoints is obviously the "safer" choice!

 

[Hr. Schmid]

backups (and restore) is currently "all or nothing" - you can exclude mountpoints from the backup, but then upon restore they will be empty (and if you restore over an existing container, the existing volume will be destroyed as part of the restore!). if you want to go down that route, you want to restore into a new (not-yet-existing) VMID, then remove the empty MP and re-assign the one containing data from the old CT to the newly restored CT. backing up the full container including all mountpoints is obviously the "safer" choice!

Ok, then I'll stick to my backup strategy.

Thanks again!