Auto restart MySQL service if it went down

[bari86]

Sometimes I see that MySQL service went down and then stays down. Would be good if there is something to monitor the service and restart it should it went down for 1 minute? I believe cPanel use something like this where if ever MySQL service is down, it can auto-restart it automatically.

[jessuppi]

Thanks for posting @bari86

Interestingly right after you mentioned this, I had a client contact me who is using SlickStack on an older Ubuntu 20.04 server saying his website was crashed... the error log was full of Redis object cache related errors.

I have to assume whatever he experienced is likely the same as you.

I'm going to open a new Issue to investigate that further, but the "quick fix" is probably migrating your site to a brand new build on Ubuntu 22.04 which is incredibly stable for us.

About auto-restarting MySQL it's a great idea and something we used to have (kinda) with Monit, but after we stopped bundling monitoring apps there hasn't been any further attempt at auto-restarting services.

[hj-collab]

@jessuppi You can add the below line to make redis / mysql restart always when it runs out of memory. Ofcourse this won't restart for configuration errors. Override the systemd service file by creating an override file else update will override your change.

[Service]
Restart=always

[jessuppi]

Thanks again for the suggestion. We added a lightweight MySQL health check to 02-cron-often.

Ref: https://github.com/littlebizzy/slickstack/blob/master/crons/02-cron-often.txt

The cron now checks local MySQL using a cheap mysqladmin socket ping before running WP-Cron. If MySQL fails 3 consecutive checks, SlickStack will restart MySQL using the existing ss-restart-mysql script.

This is intentionally conservative:

• skips remote database setups
• runs only every few minutes via 02-cron-often
• uses a 5-second timeout so the check cannot hang the cron
• stores only a small failure counter in /var/www/meta/mysql-watchdog.count
• clears the counter after a successful ping
• restarts MySQL only after repeated failures, not after one brief spike

The goal is to help recover from cases where MySQL becomes unresponsive after traffic spikes, bot floods, or amateur DoS attacks, without adding heavy monitoring or restarting MySQL too aggressively.

Ref: MAY2026L

This has not been tested much, so I'm not going to close this issue quite yet. Testing and feedback greatly desired.

(With help from ChatGPT.)