[SOLVED] Another "cluster not ready - no quorum? (500)" case

[PeterZ]

Hi all, new Proxmox user since a few months for a home server.
I have some basic experience with vmware & virtualbox but recently got a refurbished mini pc which I geared up with Proxmox and a few containers and vm's; initially this was a test but because I liked it, I continued this path.

Everything was fine, I installed a second node and thought of moving containers and vm's to this machine so that I could reinstall the main system with a recently repurposed SSD - now all is on 1 HD.
So I created a cluster, tried through migration in the web interface but got stuck on the systems not being reachable.
Moved them back to first node and everything was OK.

Yesterday I got the 'no valid subscription message and after making some changes to the sources files, I got some updates, which I installed.
Now my containers and vm's don't boot anymore with this "cluster not ready - no quorum? (500)" message.

The second node is still showing but not manageable anymore from the first and if I click on it, I get "hostname lookup 'pve2' failed - failed to get address info for: pve2: No address associated with hostname (500)" although it's running and I can manage it from it's own web interface.

I don't need this second node right anymore, so no matter what happens with it but I have some data in a vm on the first node that I would like to get my hands on.
If possible, recovering the first node with the systems, would be a bonus.

Is this possible and if so, how?
TIA!
Peter

 

[mira]

You can temporarily set 'pvecm expected 1' which sets the expected votes to 1. Make sure the other node has no access to any of the VMs or storages the first node uses.

Please post your corosync config (/etc/pve/corosync.conf), your /etc/hosts file and your /etc/network/interfaces file. If there are any public IP addresses in there you might want to mask them.

 

[PeterZ]

Hi mira, thanks for picking this up!
It's not clear to me if changing that setting was the only thing to solve this since you're asking for these files, but the node is back up and running.

Thanks a LOT!
Peter

 

[Syrrys]

Hey,

Just wanted to add a note for anyone else who comes across this thread looking to setup a small homelab with 2 nodes.

according to man corosync.conf and man votequorum explain that adding two_node: 1
to the quorum{} section of /etc/corosync/corosync.conf will create a special case where the usual (nodes/2)+1 quorum rule doesn't apply.

I'm not sure yet what other changes might be needed to get or keep this setup stable but for me, this is letting me get on with spinning up new VMs etc.

 

[ieronymous]

Hey,

Just wanted to add a note for anyone else who comes across this thread looking to setup a small homelab with 2 nodes.

according to man corosync.conf and man votequorum explain that adding two_node: 1
to the quorum{} section of /etc/corosync/corosync.conf will create a special case where the usual (nodes/2)+1 quorum rule doesn't apply.

I'm not sure yet what other changes might be needed to get or keep this setup stable but for me, this is letting me get on with spinning up new VMs etc.

By that you mean the line two_node: 1 to be inserted on both nodes?
Like

quorum {
provider: corosync_votequorum
two_node: 1
}
...... because didnt seem to work for me? Did you have to update any service afterwards even though I rebooted the server

 

[UdoB]

(Please do not re-animate very old topics - which are already labeled "Solved" in this case.)

Perhaps you might read https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_quorum regarding "QDevice" to get quorum by simulating a third node.

Best regards

 

[t.lamprecht]

Please don't necro bump old threads, and we recommend avoiding generic articles for cluster problems, they can make the actual underlying issue often much worse.

Also note that this is a designated international English-speaking forum, we currently do not have a Portuguese-speaking one, I'm afraid.