@ "thekendog"#p19875
It's generally not ideal to give MySQL optimization recommendations without knowing your exact setup, but since you didn’t provide details, here are some general tips:
Apply these settings:
max_user_connections=25
tmp_table_size = 64M
max_heap_table_size = 64M
max_allowed_packet = 128M
wait_timeout = 300
interactive_timeout = 300
Restart MariaDB, let it run for 24 hours, then run this script and post the output:
https://gist.github.com/cPFence/98c359cfade030fd62adb6681312a97a
If you want to mimic how v11 handled this, you can create a systemd override file with auto-restart blindly enabled, similar to what Docker was doing. When a spike happens, it will just keep restarting MariaDB over and over (obviously, MySQL will be down while this is happening). Whether it lasts a few minutes or a few hours, only time (or luck) will tell.
Then the admin checks in, sees the MySQL service is running, and everything looks fine ,but in reality, it isn’t. In my humble opinion, this is the wrong approach.
If you insist on using a systemd override to mimic Docker behavior, make sure you set up proper monitoring for mission-critical sites so you actually know what’s happening, when, and how it’s performing.
