Altering Proxmox Quorum Votes
Recently, I converted my Rocky Linux 9 servers, which I had set up QEMU/Libvirt to run my VMs, to Proxmox. Why? That’s another post. This one concerns me adjusting the quorum votes a node gets in a cluster.
Why do I need to adjust the votes? I have an oddball setup in my homelab that would not typically be seen in a production environment. In the homelab is a single Dell R420. I also have a VM on my desktop inside of Virtual Box. Both of these are set up in a cluster. I would not have the VM running on the desktop in a typical operation. Its only job is to let me move the VMs to “somewhere” if there is a problem or about to be one—also clustering.
The problem this causes is that for Proxmox to start any VM in a cluster or allow configuration changes, it has to have a quorum. Unfortunately, that is 51% or better. By default, every node in the cluster gets a single vote. If the server rebooted, it would come up to a 50% quorum, not enough votes. Thankfully, getting around this is easy, but it is not recommended for anyone unless you have a random setup like this.
The howto:
As root
1
2
3
systemctl stop pve-cluster.service
pmxcvs -l
nano /etc/pve/corosync.conf
In the nodelist section, edit the node you want to have more votes to include
quorum_votes: 3
Then, in the totem section, update the config_version by one number.
Save and close.
1
2
killall pmxcfs
systemctl start pve-cluster.service
Reboot the node and check pvecm status to ensure votequorum show proper expected votes and quorum numbers.