[SOLVED] Proxmox backup and restore

[Mrt12]

Dear colleagues

OK I've got a PVE node in my cluster whose system disk crashed. So the Proxmox installation with all its settings on this particular node is lost.
However I believe that the ZFS pool on the other disk should still be alive, and I do have full backups of all my VMs. Two questions:

a) one of the VMs is 2.7TB to restore. While I believe that this would work (Proxmox showed to be extremely reliable), it would take a bit too long. On the other hand, I think the VM's disk image on the ZFS pool should still be there. So if I do a fresh installation of Proxmox on this node, can I somehow restore only the configuration for this particular VM, and use the disk image which is on the ZFS pool? that would save me a lot of time, plus amount of work lost would be minimised (backup is 2 days old).

b) to avoid this problem in the future: what would be the best strategy to backup the Proxmox system itself? how do you do it?

Thanks!

 

[oguz]

hi,

a) one of the VMs is 2.7TB to restore. While I believe that this would work (Proxmox showed to be extremely reliable), it would take a bit too long. On the other hand, I think the VM's disk image on the ZFS pool should still be there. So if I do a fresh installation of Proxmox on this node, can I somehow restore only the configuration for this particular VM, and use the disk image which is on the ZFS pool? that would save me a lot of time, plus amount of work lost would be minimised (backup is 2 days old).

yes - if you have a backup of the config this is easy, if you don't then you can try recreating the configuration to the best of your memory and tweak it until it works.

b) to avoid this problem in the future: what would be the best strategy to backup the Proxmox system itself? how do you do it?

to backup the proxmox related configurations you can make a full backup of /var/lib/pve-cluster/config.db (this is the database behind the cluster filesystem on /etc/pve)

otherwise regular linux backup suggestions apply, (backup /etc/ at minimum)

 

[Mrt12]

dear oguz,
MANY thanks for your hint!
Today, I tried it and it works!
I did it as follows: in the "storage" view, one can show the configuration of a VM which was used when it was backed up.
So I made a fresh installation of Proxmox, then I created a new VM and renamed the ZFS subvolumes accordingly and attached them to the newly created VM. Then, I copied the configuration info from the backup into the files under `/etc/pve/qemu-server/<vmid>.conf` and it just WORKS! great! many thanks.


One last question:
I had two nodes together in one cluster, and node PVE3 is now inexistent, as it blew up and I had to reinstall it.
I named it now PVE0, and tried to remove PVE3 from the cluster with `pvecm delnode`. That worked, but interestingly, I still see the node PVE3 in the web interface, but its just greyed out. How can I completely remove it?

hi,



yes - if you have a backup of the config this is easy, if you don't then you can try recreating the configuration to the best of your memory and tweak it until it works.



to backup the proxmox related configurations you can make a full backup of /var/lib/pve-cluster/config.db (this is the database behind the cluster filesystem on /etc/pve)

otherwise regular linux backup suggestions apply, (backup /etc/ at minimum)

 

[fiona]

Hi,
is there still a directory for PVE3 in /etc/pve/nodes/? If you find some useful configuration files in there (e.g. the host firewall config), you might want to transfer/adapt it to the re-installed node. When you made sure there's nothing you need in there, remove that directory. Also please check your replication jobs (if you have any) and your /etc/pve/storage.cfg for mentions of the removed node.

 

[Mrt12]

Hi,
is there still a directory for PVE3 in /etc/pve/nodes/? If you find some useful configuration files in there (e.g. the host firewall config), you might want to transfer/adapt it to the re-installed node. When you made sure there's nothing you need in there, remove that directory. Also please check your replication jobs (if you have any) and your /etc/pve/storage.cfg for mentions of the removed node.

that did the trick. Cool, thanks!

 

[oguz]

great, you can mark the thread as [SOLVED] so others know what to expect as well