Delete unconnected nodes from cluster

m1n10n

New Member
Aug 23, 2023
2
0
1
I have 6 boxes in a cluster that are all running proxmox. Since initial configuration, I have reconfigured the physical network connection and subnetted each node in the cluster, so that theyre all together in the same subnet, separate from my main network. Here is my problem; since i changed the physical wiring configuration, and each node now has a new local IP address, there is no connectivity between them in the previously configured cluster (since they are looking to resolve original IPs that no longer exist). I have no physical access to each node (they are offsite), but i am able to connect to to each of them independently via SSH and webgui.

Is there any way to delete the cluster configuration, or change the existing configuration to point to the new subnet IP addresses, without first connecting the cluster all together, and deleting it (since reconnecting the cluster would require me to physically change the configuration of the cat cables, and i dont have them next to me for a few months)

Any help is appreciated!
 
You need to update the corosync configuration on each node to include the new IP. You should be able to do it in the following way:

Disable corosync and pve-cluster on ALL nodes:
Code:
systemctl stop pve-cluster corosync

Then on one node mount pmxcfs locally and edit the /etc/pve/corosync.conf file:
Code:
pmxcfs -l
[vim|nano] /etc/pve/corosync.conf

Change the ring0 addresses to your new IPs and don't forget to increase the version number. It might make sense to read our respective wiki entry for editing the configuration [1].

Since you should be having network connectivity on the new IPs (I assume you have configured the interfaces correctly). You can then mount pmxcfs locally on the remaining nodes and manually copy over the new corosync.conf file (e.g. via scp). You need to do this on all other nodes:

Code:
pmxcfs -l
scp root@<node1>:/etc/pve/corosync.conf /etc/pve/corosync.conf

After you have done this, reboot ALL nodes (systemctl reboot), and they should form a cluster again.

[1] https://pve.proxmox.com/wiki/Cluster_Manager#_corosync_configuration
 
Last edited: