Kain, Becki (B.) wrote:

Thanks. I have a user who is very upset about this web server. Other ideas for figuring out what is pegging the pu's?

it really doesn't matter if one CPU is pegged or not... as long as there are clock-cycles available. Some software is "bursty"... the nature of the problem causes the software to run flat out, and will get lucky that nothing else is schedule-able...and will get scheduled AGAIN...and thus momentarily peg CPU usage.... That's not bad...it's GOOD.. you're getting maximum CPU throughput. The question is...how is the system RESPONDING???? if you want to see which processes are hitting the CPU the hardest...just look at the list of processes that start on line 5 or so of top's display... The processes at the top of the list are the ones using the highest percentage of CPU time during the most recent polling period (default 1 second)

-----Original Message----- From: Aaron Kulkis [mailto:akulkis00@hotpop.com] Sent: Friday, January 11, 2008 2:18 PM To: opensuse Subject: Re: [opensuse] Top/lsof

...

I read the man page of top but it's not helpful for this question. When top tells me cpu0 is being used 50.0% by sys, cpu1 is being used 65.7% by system space, how do I break down what processes are making up

Kain, Becki (B.) wrote: that

...

65.7%, in system space, of the cpu? And in lsof, how do you tell which cpu a process id's threads are tied to?

One, you can't Two, it doesn't matter much.

"sys" merely correlates to time spent executing system-calls (i.e. kernal calls), whereas "user" correlates to the time spent executing code outside of the kernel.

As far as lsof... try this:

ps -ef | grep lsof

once you get the PID of the lsof process(es), you can grep for them:

ps -ef | grep _PID_of_lsof_goes_here

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org