[SOLVED] Cluster Node Name Problem

[troycarpenter]

I have a node in my cluster that somehow ended up with the wrong spelling in /etc/corosync/corosync.conf. I have tried changing the file itself, but clearly it gets regenerated at some point. The node entry looks like this:

nodelist {
  node {
    name: bagh
    nodeid: 2
    quorum_votes: 1
    ring0_addr: baugh-corosync
  }

As you can see, the ring0_addr entry is wrong. Now on that node, whenever cororsync restarts, the node complains that there isn't a quorum, and the other nodes think it has dropped out of the cluster. If I change the file manually and restart corosync, then the node comes up.

How do I permanently fix that entry so I don't have to manually work around this problem whenever corosync restarts on that node (like when the corosync-pve package updates)? And I'd rather not have to reinstall the node with a different name.

 

[troycarpenter]

I found the database in /var/lib/pve-cluster where that config file is generated. Using the sqlite3 cli command, I can see it with:

select data from tree where name='corosync.conf';

However, I'm afraid to just modify the database, and I'm not sure how that database gets propagated around the cluster.

So for now the best way I've found to work around the issue is to create an entry in /etc/hosts with the wrong fqdn for the typo in the configuration file on each of my nodes.

 

[wolfgang]

Hi,

You must edit it in the /etc/pve/corosync.conf not in the /etc/corosync/corosync.conf.
The /etc/corosync/corosync.conf will be overwritten on every restart.

 

[troycarpenter]

Thank you! I can't believe I overlooked that file. That took care of the problem.

 

[emoxam]

Hi,

You must edit it in the /etc/pve/corosync.conf not in the /etc/corosync/corosync.conf.
The /etc/corosync/corosync.conf will be overwritten on every restart.

And what to do after editing? Restart something ?
systemctl restart corosync.service Will be sufficient ?

 

[UdoB]

Please look here: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#pvecm_edit_corosync_conf

In any case a reboot should work...


Best regards