2-Node Cluster Best Practices

[baron164]

I currently have two proxmox hosts running in a cluster for my homelab. I am concerned about what will happen if one of the hosts goes offline. Whether due to temporary reboot or possible hardware failure. What is the best way to ensure that the remaining host works properly if one of the nodes goes offline?

I have read a couple of options, one is to setup a quorum disk which sound similar to Microsoft's Clustering. I've also seen mention of changing the number of votes a node gets to ensure it can stay running in the event another node goes offline.

I got around this issue with Microsoft's Hyper-V by having a SMB share on my Synology act as the disk quorum. Would something like that be possible with Proxmox?

 

[Dunuin]

I currently have two proxmox hosts running in a cluster for my homelab. I am concerned about what will happen if one of the hosts goes offline.

The remaining node will switch read-only and you won't be able to change configs or start VMs until there is quorum again.

What is the best way to ensure that the remaining host works properly if one of the nodes goes offline?

You get a third PVE node or at least a Thin-Client/SBC/PBS/VM on a NAS and set that up as a qdevice: https://pve.proxmox.com/wiki/Cluster_Manager#_corosync_external_vote_support

've also seen mention of changing the number of votes a node gets to ensure it can stay running in the event another node goes offline.

This will only help if you always shut down the same node. If the node with more votes is unavailable for some reason, the node with less votes will stop working too.

There is no proper way other than having 3+ voters if any single node should be able to be restarted/fail.

 

[baron164]

The remaining node will switch read-only and you won't be able to change configs or start VMs until there is quorum again.


You get a third PVE node or at least a Thin-Client/SBC/PBS/VM on a NAS and set that up as a qdevice: https://pve.proxmox.com/wiki/Cluster_Manager#_corosync_external_vote_support

This will only help if you always shut down the same node. If the node with more votes is unavailable for some reason, the node with less votes will stop working too.

There is no proper way other than having 3+ voters if any single node should be able to be restarted/fail.

Ok, so it looks like I should be able to run corosync in a container on the Synology using docker. Plan B would be running an entire VM for it, but hopefully a container would be enough. However, before I dig into that more, is there a disk quorum option?