All of a sudden, today, Radiator kept locking up on us every 20 minutes.
Looking in the logs for MySQL and for Radiator, MySQL was crashing and
restarting regularly, so I turned on -trace 9 on radiator, hoping to see if
there was any relationship between radiator crashing and MySQL core dumping.
It turns out that there was no relationship between coredumps and crashes --
MySQL would continue to stay running without a recent coredump when Radiator
locked up. The problem was that MySQL was locking up instead of
core-dumping.
An upgrade from 3.22.25 to 3.22.30 made the problem worse! MySQL was not
coredumping and locking up more regularly than before. I started to get a
little worried, but then remembered isamchk.
Turns out radius/RADONLINE was corrupt! No wonder MySQL was crashing often.
Our RADONLINE table has a LOT of write and read activity.
TIPS: 1. Run ISAMCHK to check your databases regularly.
2. Repair broken databases.
Information on using ISAMCHK:
http://mysql.cyberus.ca/Manual/manual.html#Table_maintenance
--
Roy Hooper ([EMAIL PROTECTED])
Director, Network Architecture
eisa.com
===
Archive at http://www.thesite.com.au/~radiator/
To unsubscribe, email '[EMAIL PROTECTED]' with
'unsubscribe radiator' in the body of the message.
