We are facing since few days a strange behavior with a 3-nodes Galera cluster and Haproxy frontend, deployed by ClusterControl 1.5.1
Sometimes, nodes do not respond suddenly to queries, no event in mysqld logs nor haproxy.log
We have to failover the vip (keepalived restart) to get the cluster working, immediatly.
I have activated long queries log and waiting the next similar event.
Mariadb 10.2.22 and Haproxy 1.6
You can see the event starting at 20:06 , connections are peeking but queries are collapsing, and come back few minutes after, when a haproxy failover is manually done.
Have you any idea ?
Thank you for your help.
Please sign in to leave a comment.