[SOLVED] Error when restoring LXC backups

[nicoska]

Hi, this is my current situation:
1 Cluster with 2 nodes

Node n. 1 running:​

PBS (with Node n. 2 LXCs backups on it)​

PVE​

​

Node n .2​

PVE - now dead​

​

I was trying to restore backups into Node n.1 from the PBS, but this is the message I got when trying to restore from GUI:



How can I restore all my LXCs backup into Node n. 1?

Thank you
Nico

 

[sterzy]

Hi,

two node clusters aren't supported by us without a QDevice [1]. Once one node fails the entire cluster will lose quorum which will lead to exactelly the situation you are in now.

You can temporarily work around this issue you can use the following command:

pvecm expected 1

However, be aware that if you change things on a node that has no quorum, it may not be possible to recover the cluster anymore. At this point the two nodes in your cluster would be diverged and there is no automatic way of getting them to converge again. This includes restoring VM that previously ran on another node. So be careful here.

[1]: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_corosync_external_vote_support

 

[nicoska]

So, what will be the best solution to restore the LXCs that are backup'd on the PBS on the "active" node?
Un-cluster it?

Thanks
Nico

 

[sterzy]

So, what will be the best solution to restore the LXCs that are backup'd on the PBS on the "active" node?

That depends, do you plan on bringing the “dead” node back up? Can you essentially re-install the dead node, or would you lose some important data?

It really depends on your situation what is easiest/most efficient. If you want to remove the dead node from the cluster you can use pvecm delnode to remove the dead node. You may have to run the pvecm expected command before that in order for the comand to work. After that you could re-install the node or re-use it for other purposes, but adding it back to the cluster without re-installing is probably a bad idea.

If you can boot the “dead” node and want to remove it from the cluster, you can also follow these instructions: [1]. You may also need to follow them on the one remaining node to revert it to a non-clustered state.

The third alternative would be to use the pvecm expected 1 command. Then restore the VMs to different VM IDs and manually merge the /etc/pve folder if the “dead” node fails to re-join the cluster properly after you bring it back up. After the node is back up and joined the cluster, you can remove the old VMs.

[1]: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#pvecm_separate_node_without_reinstall

 

[nicoska]

That depends, do you plan on bringing the “dead” node back up? Can you essentially re-install the dead node, or would you lose some important data?

It really depends on your situation what is easiest/most efficient. If you want to remove the dead node from the cluster you can use pvecm delnode to remove the dead node. You may have to run the pvecm expected command before that in order for the comand to work. After that you could re-install the node or re-use it for other purposes, but adding it back to the cluster without re-installing is probably a bad idea.

If you can boot the “dead” node and want to remove it from the cluster, you can also follow these instructions: [1]. You may also need to follow them on the one remaining node to revert it to a non-clustered state.

The third alternative would be to use the pvecm expected 1 command. Then restore the VMs to different VM IDs and manually merge the /etc/pve folder if the “dead” node fails to re-join the cluster properly after you bring it back up. After the node is back up and joined the cluster, you can remove the old VMs.

[1]: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#pvecm_separate_node_without_reinstall

pvecm expected 1 solved my problem.
I was able to restore LXCs backup into the "good" node.

Now I'm wondering how to remove the "dead" node. Is there any safe way?

 

[sterzy]

As stated before: Follow the steps I linked to in my previous reply [1] on the “good” node. This will make it a stand-alone node again. Afterward, you should decommission the “dead” node.

[1]: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#pvecm_separate_node_without_reinstall