Setting up a cluster with only 2 hosts

Thechaser

New Member
Mar 28, 2024
14
8
3
Hi all,

If I set up a cluster with only two hosts, everyting crashed when one host fails. Can't use cli or even remove the dead node from the cluster.
I wonder what modifications I need to do on the system to get it running on only two hosts and not crash when one fails.

This is a production at home use. that's why I only have two nodes and not three.

Thank you
 
Hi,

first of, I'd suggest you read the chapter on Cluster Management in our admin guide. This explains the basics and what to be aware of when running clusters very well.

If I set up a cluster with only two hosts, everyting crashed when one host fails.
Two-node clusters are not support, see the link above. You can work around that using an external QDevice.
 
  • Like
Reactions: Johannes S
Thank you both for the information.
I will read it (doing it at the moment).

Maybe its important to note that I don't want to use this cluser for HA. but only for easy migration of a vm from one node to another.
and to mange both in one gui.
so maybe a cluster is not the right method?

If no other solution I will buy a extra node.
 
if you only have 2 nodes,
you need to use CLI to move vms. (it's not possible with gui).


when a node crash:

1) verify that the node is really crashed and vms are not running anymore

2) on the other node : #pvecm expected 1 (to be able to write in /etc/pve )

3) on the other node : #mv /etc/pve/nodes/<crashednode>/qemu-server/* /etc/pve/nodes/<othernodes>qemu-server/

then start all vms on the other node with gui



if you have a third node. (could be a small nuc , without any vms running on it), you could enable HA and the failover done automatically)
 
if you only have 2 nodes,
you need to use CLI to move vms. (it's not possible with gui).


when a node crash:

1) verify that the node is really crashed and vms are not running anymore

2) on the other node : #pvecm expected 1 (to be able to write in /etc/pve )

3) on the other node : #mv /etc/pve/nodes/<crashednode>/qemu-server/* /etc/pve/nodes/<othernodes>qemu-server/

then start all vms on the other node with gui



if you have a third node. (could be a small nuc , without any vms running on it), you could enable HA and the failover done automatically)

Hello,

Thank you for this information.
Your reply was giving me the solution I needed I think.
If I set up my cluser with 2 server + small nuc to have three nodes and only start the nuc when one fails I assume the gui will kick in when the nuc is up and running.

I have read the other links that where provided, and they have given me indeed more information.

thank you all
 
Hi all,

If I set up a cluster with only two hosts, everyting crashed when one host fails. Can't use cli or even remove the dead node from the cluster.
I wonder what modifications I need to do on the system to get it running on only two hosts and not crash when one fails.

This is a production at home use. that's why I only have two nodes and not three.

Thank you
Here is your solution: https://forum.proxmox.com/threads/proxmox-datacenter-manager-first-alpha-release.159323/

https://enix.io/en/blog/proxmox-datacenter-manager/
 
Hi all,
I have the same situation. Only 2 hosts.

After reading this thread I see I cannot move vms between the 2 nodes via gui?
(or is this just a fact when a node is disconnected/down)

As for a 3rd quorum/arbiter I could setup a Corosync External Vote Support but I also could setup a nuc .
What could be the best?

And if I setup a corosync could I setup multiple or would that not be a good idea? (thinking of setup 2 because of possible site failure an losing both pve node and corosync vote)
 
Last edited:
Hi all,
I have the same situation. Only 2 hosts.

After reading this thread I see I cannot move vms between the 2 nodes via gui?
(or is this just a fact when a node is disconnected/down)

As for a 3rd quorum/arbiter I could setup a Corosync External Vote Support but I also could setup a nuc .
What could be the best?

And if I setup a corosync could I setup multiple or would that not be a good idea? (thinking of setup 2 because of possible site failure an losing both pve node and corosync vote)
I actually found the answer myself..

So in short you need an External vote whenever you have an even number of nodes.
so i will not be setting up multiple corosyns (i. don't even think it is possible now that I have seen the configure command)
 
  • Like
Reactions: UdoB