>>>>> "Paul" == Paul Schilling <pschill@stripped> writes:
Paul> Between Feb 8th and last night we have experienced 4 lockups by the mysql
Paul> server. Once locked up the safe_mysql and 1 mysqld process are still
Paul> running but you can not connect to the server.
Paul> A ps shows this:
Paul> USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
Paul> <user> 17001 0.0 0.2 868 240 ? S Mar 09 0:00 /bin/sh
Paul> /.../mysql/bin/safe_mysqld --log=/.../MYSQL/query.log
Paul> <user> 17012 0.0 9.31925617712 ? S Mar 09 45:12
Paul> /.../mysql-3.22.19b/libexec/mysqld --basedir=/.../mysql-3.22.19b
Paul> --datadir=/.../mysql-3.22.19b/var --log=/.../MYSQL/query.log
Paul> Each of the lockups occurred at about 1:20am after at least a week of
Paul> mysql uptime. Nothing is croned for that time. It also happened on a
Paul> different weekday each time. Other server information did not show anything
Paul> out of the ordinary at this time.
Paul> 3 of the 4 lockups were while running the prebuilt 3.21.33b Solaris
Paul> 2.6 binary. The 4th was a mysql 3.22.19b I build w/ --with-debug and
Paul> -ggdb. A gcore followed by gdb didn't show anything useful. The machine is a
Paul> SPARC 20.
Paul> If any more information is needed please contact me.
Did you get a back trace from gdb?
Can you verify from the log if it's some specific query that
kills mysqld ?
Please check your tables with isamchk -s */*.ISM !
Did you read the MySQL manual entry 'What to do if 'MySQL' keeps
crashing' ? We have tried to put a lot of useful information in this
chapter to help locate this kind of problems.
Any change you can run mysqld under gdb to get a proper back trace
when this happens?