← Back to team overview

maria-discuss team mailing list archive

Problems with running out of connections in cluster



We are running a 7 node cluster (6 data nodes and 1 arbiter node) which span 2 data centers using version 10.1.13. We have had several occasions where the application have gone off and created so many connections that the max connections are filled up which leads to the database becoming unresponsive. The connections cannot be killed and and in the ‘killed’ state forever. Today when this occurred all six cluster node maxed out on connections. The only way to get the servers back is by restarting the databases but when the first node we restarted it tried to connect to the cluster is to IST it could not connected. After shutdown a second node the cluster seemed to sort its self out and everything came back. Is anyone aware of any bugs that might cause this issue.