Memory Leak? 8.0.33-25.1 Percona XtraDB Cluster

MySQL & MariaDB Percona XtraDB Cluster 8.x
1

Since updating from .29 I have seen a large increase in memory usage. Even to the point where I have had to restart the DB process on each server in the cluster. On Average Every 10 days or so… Mind you this is on all the servers not just some… For instance this server more or less sits as a third or fourth server fail over backup and has exceeded the memory use by quite a bit already… Once this happens things also tend to slow down and cause application issues… Thoughts?

innodb-buffer-pool-size = 30G

 ps aux --sort=-%mem | awk 'NR<=11{print $0}' | awk '{print $11, $2, $4, $6/1024 "GB"}'
COMMAND PID %MEM 0GB
/usr/sbin/mysqld 43492 76.7 39256.1GB

image
image1060×152 8.44 KB

settings.txt (19.1 KB)

2

Hello @meyerder,
Use the sys schema and some of the memory views within to determine where the memory is being used. That might help isolate if this is indeed a memory leak or a connection that needs to be closed.

3

@matthewb this seems VERY high… Though I will admit Im not sure exactly what I am looking for yet…

image
image1214×445 34.9 KB

image
image973×65 5.02 KB

image
image1166×108 8.34 KB

Ok the root issue APPEARS to be that the applier threads are using much more memory than in the prior versions… I reduced the amount of applier threads and it “seems” after a good 12 hours that things are not using as much as before… I will continue to monitor. Thank you for the Tip… I had flat out forgotten about those views