Proxmox 7 -8 upgrade Use of uninitialized value $nodename

[dazza76]

We are in the process of upgrading our cluster from 7 to 8, and had an issue where all the confgis had dissapeared
luckly we had a back up of all the config files -- extracted from backups.

but now we have an issue with the ha-manager
ha-manager status | head
unable to read file '/etc/pve/nodes//lrm_status'
quorum OK
master Node4 (old timestamp - dead?, Tue Oct 14 03:00:31 2025)
lrm (unable to read lrm status)
lrm Node1 (wait_for_agent_lock, Tue Oct 14 22:41:09 2025)
lrm Node2 (wait_for_agent_lock, Tue Oct 14 22:41:10 2025)
lrm Node3 (wait_for_agent_lock, Tue Oct 14 22:41:09 2025)
lrm Node4 (wait_for_agent_lock, Tue Oct 14 22:41:09 2025)


ournalctl -u pve-ha-crm -n 50
journalctl -u pve-ha-lrm -n 50
Oct 14 03:00:41 Node4 pve-ha-crm[165430]: Use of uninitialized value $nodename in hash element at /usr/share/perl5/PVE/HA/Usage/Basic.pm line 38.
Oct 14 03:00:41 Node4 pve-ha-crm[165430]: Use of uninitialized value $current_node in hash element at /usr/share/perl5/PVE/HA/Manager.pm line 181.
Oct 14 03:00:41 Node4 pve-ha-crm[165430]: Use of uninitialized value $fenced_node in concatenation (.) or string at /usr/share/perl5/PVE/HA/Manager.pm line 993.



Thoughts or ideas to fix ?

Cheers D

 

[dazza76]

Fix for this was to
systemctl restart pve-ha-crm
then wait for some time for it to resolve
we then created a tool to cycle through the vms on each host and deactivate ha and re-enable

Hope this helps somebody