On 03/17/2009 05:19 PM, Agile Aspect wrote: <snip>
My first guess is maybe it's a problem with irqbalance but it wouldn't explain the crashes.
Interestingly, I stopped irqbalance yesterday afternoon and it ran all night just fine.
It's possible you have a misbehaving CPU - maybe one which is getting hot since it crashing. Bad thermo grease perhaps.
Can you look at the temperatures?
I've reinstalled using the x86_64 distro and kernel 2.6.18-92.1.22.el5.centos.plus. gkrellm reports 104F cpu temp; seems to be accurate - the heat tubes on the cpu cooler are just warm.