Without warning, MySQL went on strike. After starting with the startup script, the mysqld process ends without any prompt information, and naturally the mysqld process does not start.
Redirect the output of $bindir/mysqld_safe --datadir=$datadir --pid-file=$server_pid_file >/dev/null 2>&1, and then run it. As a result, it was found that an error was reported: File size limit exceeded nohup ... ...
This seems that there are too large files. But after checking the relevant database files and log files in the mysql data directory, the largest one is only a few hundred megabytes. ulimit -a View also does not limit the file size.
After N minutes of depression, I decided to use find / -size +2048000k to find suspicious files on the entire hard drive. The result is: /var/log/pacct This 2G large file emerged. Obviously, this is the log file generated after SA activates accton and is not rolled back. Without saying a word, clear /var/log/pacct and start MySQL again, and everything returns to normal!
What I don't understand is why the mysql startup script is affected by such an inconsistency process statistics log, which causes the mysqld process to be completely unable to start. It is probably still a problem with /mysqld_safe script logic. .