mysqld CPU spike issue still around...

[JaseP]

I am running and apt-get upgrade on my core and MDs (MDs got done relatively quickly)...

I was trying to shut down unnecessary things sucking up bandwidth or traffic on the whole network when I just happened to run a top on the Core, to see if there was anything hogging up system resources that might possibly slow down the upgrade,... That's when I noticed the CPU pegged near 100% on the mysqld... (turns out it had nothing to do with the slowness,... apparently that's just my connection to the LinuxMCE servers)...

I searched the forums and noticed that this appeared back in the 810 version, ... and that several people had noticed that if you kill mysqld it doesn't hog CPU cycles when it respawns... So I did that, and whataya know... same thing on 1004... a well behaved, respawned mysqld after killing it... (of course it had nothing to do with download efficiency) .

Does anyone know what's causing this??

[Marie.O]

You might want to look at CPU utilization of mysqld for a longer period of time, and see if you can find a pattern.

[hari]

I'd also try to use the log_slow_queries setting in my.cnf

[dextaslab]

Possibly try installing mrtg, or collectd to graph cpu usage, that way you'll be able to better determine when it happens and possibly why. What CPU/RAM does the affected system have?

[JaseP]

The system is an Intel dual core with 4GB RAM...