Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Not a Bug
-
2.4.12
-
None
-
Ubuntu 18.04.5 LTS, MariaDB Galera cluster
Description
I have configured connection_keepalive=1 parameter under ReadWriteSplit router, out MariaDB servers are configured with wait_timeout=28800 and interactive_timeout=28800 but we are still experiencing 'MySQL Server has gone away' errors during operation of our cluster. According to documentation, MaxScale should keep connections open by sending periodic keepalive packets. I've checked our MariaDB cluster and can confirm that no nodes were failing to cause this issue, but indeed the disconnect was caused by a long open connection that wasn't sending any queries. Actually, application was sending queries like 'SELECT 1' by it looks like they were routed to slaves allowing master connection to time out.