Master HA node failover

Hi All

Ive setup a HA broker cluster - 2 nodes using pacemaker + corosync (TL 4.20).

The HA cluster appears to work fine.

A failover test - node 1 was taken off line with an active session on VDA/agent server.

The HA failover is ok - able to login via the VIP using the client and the VP web admin page is redirecting to node 2. However, node 2 shows 0 active sessions so the session DB failed to sync from node 1 to node 2.

/opt/thinlinc/etc/conf.d/cluster.hconf
HA is showing as enabled (1) and that both nodes are listed in the cluster
nodes= node1.example.com node2.example.com

Bringing node 1 back on line - node 2 stays primary and active node.

Unable to reconnect to the disconnected session on the VDA

Any ideas ?

Thanks

HArj

Please check the server logs for any information related to HA. You can find the logs at /var/log/vsmserver.log on the master server. Let us know if there are any clues there.

Hi Aaron

Thanks for the headsup on the log files. I’ll re test fail over and see if its behaving or not and check the logs accordingly.

Update:

from the logs on one of the brokers:

2026-06-29 08:58:10 ERROR vsmserver.HA: Failed to do hostname lookup on node ‘rocktldc-02.lab.internal’ from /HA/nodes
2026-06-29 08:58:10 ERROR vsmserver.HA: 0 matches when comparing hostnames from /HA/nodes with result of gethostname() - should be exactly one match
2026-06-29 08:58:10 ERROR vsmserver.HA: Can’t determine address of other node safely. 2026-06-29 08:58:10 ERROR vsmserver.HA: STARTING WITH HA DISABLED! 2026-06-29 08:58:10 WARNING vsmserver.loadinfo: Draining agents: [‘127.0.0.1’, ‘rocktldc-02.lab.internal’]
2026-06-29 08:58:10 INFO vsmserver.loadinfo: Agent 127.0.0.1 is back up

HA/Nodes was originally configured with the fqdn of each host.
Replaced that with just the hostname.

On both brokers the /opt/thinlinc/etc/conf.d/cluster.hconf file is :

nodes=rocktldc-01 rocktldc-02

Update 2 - got it working.

set the hostnames to fqdn’s on both nodes, set the HA/nodes back to fqdn and that’s worked. HA is working and session failover is seamless - and session disconnects and reconnects.

KR

Harj

Hi @Harj,

Great that you got it working and thank you for sharing your solution!