Seen yesterday on two 5.x installs updated to the latest (.6) - that’s why I got the answer.
If the system suffers an unclean shutdown (e.g. power cut) the “mysql.sock” file lingers behind, and the startup script fails to start mysql on reboot, saying there’s another instance running.
I guess they’re assuming that if there’s that SOCK file mysql is up, and not bothering to really check if there’s an instance running with a “mysqld.pid” file or grepping a “ps”.
Package is not ours, but you could patch and repackage on your repo.
Do you fancy to file a bug with the maintainers? I’m not so keen.
Great, but look around in the forum because somebody that had the stray mysql.sock file problem just upgraded to .7 and the script spewed out errors on him about the DB, again.
Do we have a race condition where mysql doesn’t remove the .sock file when we reboot from a script?
Clutching at straws here (and not upgrading to .7 ATM).