Percona novice… I noticed an odd pattern and I’m still researching it. We have a multi-master setup of several nodes (6 main + 1 used for backups). We are running on Debian (version now unsupported), but we still patch the underlying VMware hosts. We’ve determined that doing vMotions of the VMs can cause our applications to lose their connections, so we decided to pin VMs to the host and stop/restart the DBs as required as we patch.
If you look at the attachment, the growth appears linear, but if you look more closely, there’s several bumps in the disk space. This is very closely associated with the stop/restart of the VMs.
Might I be missing something else? When stopping/restarting Percona, is there a risk of leaving around unused bin files?