2 Proxmox VE Hosts Cluster issue

S

svefro

New Member
Nov 23, 2025
3
0
1
I have just set up my first proxmox env.

The plan was to run 2 nodes, one main and one ready as backup. I have a Smb share mounted as a Directory for backup. this is available on both nodes.
so if one fails i can restore backup on the other node. Assuming my smb share is still operational, of not i can take the disk and plug it in to the backup node and restore the files that way. anyway the files are not lost.

Node 1 Main runs all vms and LXC containers- Done
Node 2 Ready as backup node. can restore vms here if i need to. Also i might power down this node. and only plug it in if i need it.

Everything is set up this way now.
Note: During testing i added one node to HA just to test. i can migrate this node fine back and forth.
But while testing i had one vm migrated to the backup host (Node 2) and then shut down the main node (to install another disk)

i then noticed that the vm running on node 2 was unable to start. no ouorum.
i rebooted node 1 and shut down node 2, the vms continued to run but if i stop them i can not start them again.

I need node 1 to be able to operate regardles of node 2.

i could always run them separatly and connect both to the same network storage for backup.
But it was nice to be able to migrate vms to the backup node if i needed them running while doing mainenance on node 1

What can i do to fix this? Do i need to de-cluster them? if so how can i do this? i have seen threads on this but it looked like there was some issues.

Right now all my services are running outside of proxmox. so if i need to do a complete reinstall to be able to de-cluster the two node. this is the time.

1763915860569.png
1763915903881.png
 
Last edited:
  • Like
Reactions: Johannes S
I just saw a video about this (here). i wish i knew this before i made the cluster :p
i think i will skip qdevice. even if i have a spare pi. this can cause issues since it will have to be on another fuse than one of the nodes to be effective.

i think i will go the separate PVE's with shared network storage as backup.

So how to de-cluster 2 nodes? i see info on removing one node. But not specificaly on splitting 2 nodes in a cluster, going clusterless.
 
I did it! i think :p

here is what i did. from my own notes
<!--#############################################################################################-->
# Proxmox remove node from cluster (PVE 9.1.1)
<!--#############################################################################################-->

thanks to: https://www.youtube.com/watch?v=xTpaCAq1ruM
and misc other forum posts

Remove/Migrate all nodes to one PVE
on Main node (the one with all the VMS that we want to keep):
Back up your configuration files. Key directories to back up are /etc/pve/ and /etc/corosync/ + /etc/cluster/
cp -R /etc/pve/ /etc/pve_cluster_backup/
cp -R /etc/corosync/ /etc/corosync_cluster_backup/
cp -R /etc/cluster/ /etc/cluster_cluster_backup/
````
pvecm nodes
pvecm delnode home-pm2 # The other node without any vm's
pvecm nodes
pvecm expected 1
systemctl stop pve-cluster corosync
killall pmxcfs # you might get "no process found" in response
pmxcfs -l
````
Delete all the cluster configuration files
````
rm -f /etc/pve/cluster.conf
rm -f /etc/pve/corosync.conf
rm -f /etc/cluster/cluster.conf
rm -f /etc/corosync/corosync.conf
rm /var/lib/pve-cluster/.pmxcfs.lockfile
rm -R /etc/corosync/*
````
Stop cluster from running
````
systemctl stop pve-cluster corosync
killall pmxcfs # you might get "no process found" in response
````
disconnect the other node from the network (pulled cable)
reboot
rm /etc/corosync/authkey # no need, already deleted

but i'm still left with the other node in HA. is there a way to remove it?
and is there anything else i need to remove since i had HA enabled on one node before?
1763928429216.png
 
You must log in or register to reply here.
Top Bottom