I’m running 5.5.19-55-log. This is the second month that this has happened, and it happens on two of my servers that have nothing else in common that I can find.
At 1:03 am on the first of the month, percona shuts down. I can’t figure out why. Here is the error log file:
120102 10:33:34 [Note] /usr/sbin/mysqld: ready for connections.
Version: ‘5.5.18-55-log’ socket: ‘/tmp/mysql.sock’ port: 3306 Percona Server (GPL), Release rel23.0, Revision 203
120201 1:03:37 [Note] /usr/sbin/mysqld: Normal shutdown
120201 1:03:37 [Note] Event Scheduler: Purging the queue. 0 events
120201 1:03:37 InnoDB: Starting shutdown…
120201 1:03:42 InnoDB: Shutdown completed; log sequence number 9983757491
120201 1:03:42 [Note] /usr/sbin/mysqld: Shutdown complete
120201 01:03:42 mysqld_safe mysqld from pid file /ssd/mysqldata/foo.com.pid ended
And what does your syslog file look like at that same time??
Since mysqld reports normal shutdown it will get an external signal to shut down at that moment. So my first guess is to take a look at the crontabs on the two machines.
Or if somebody has prepared some backup job that remotely is scheduled to stop the servers and then run a file backup and then start the servers so that you get a consistent snapshot, but that would be in the crontab on the backup server or in a backup program itself where you can set a precommand and a postcommand that are going to be executed before and after the backup is taken.
Those are the two I would guess, and since you say it’s twice at 01:03 at the first of the month then it’s way to close to not be a scheduled cron job.
I figured it out. I run a yum update then, and it automatically picked up the percona server.