On Dec 18, 2008, at 12:59 PM, Matt <lm7812@xxxxxxxxx> wrote:
>> That's why I asked what kind of Controler the board had on it in a
>> previous
>> post to you and stated ram was not the suspect problem. IMO if you
>> keeped
>> the dual core proc and just switched to ICH7 Board you would have
>> saved
>> money. Your utilization rate would probly stayed the same or no
>> higher than
>> %30. Just to keep things in balance you will probly want to try the
>> cfq
>> schedular with a high user load so every thing gets it fair share in
>> time_wait. Some people will contradict that it's about making the
>> users
>> happy. When access time for one user takes longer than another then
>> the
>> complaints start coming in. I would like to know the Proc
>> Utilization per
>> core or are you running it in Single Core?
>
> top - 11:53:39 up 2 days, 9:47, 1 user, load average: 8.27,
> 13.66, 29.82
> Tasks: 188 total, 1 running, 183 sleeping, 0 stopped, 4 zombie
> Cpu0 : 10.6% us, 2.3% sy, 0.0% ni, 87.0% id, 0.0% wa, 0.0% hi,
> 0.0% si
> Cpu1 : 8.0% us, 2.0% sy, 0.0% ni, 88.3% id, 1.7% wa, 0.0% hi,
> 0.0% si
> Cpu2 : 11.6% us, 2.0% sy, 0.0% ni, 86.4% id, 0.0% wa, 0.0% hi,
> 0.0% si
> Cpu3 : 6.0% us, 0.7% sy, 0.0% ni, 92.4% id, 1.0% wa, 0.0% hi,
> 0.0% si
> Mem: 4151316k total, 3438536k used, 712780k free, 428520k
> buffers
> Swap: 2031608k total, 8k used, 2031600k free, 1839452k
> cached
Yup, that puppy is all iowait, see how the cpu idle is high and CPU
wait is low, but load avg is high.
Probably lots of random ops going on there and SATA drive can't keep up.
If you have a UPS there you can try to enable the drive's write cache
which should also help some.
-Ross
>
>
> I do see on occasion CPU load jump up to 30-60% accross the board but
> that is rare. Mostly it looks like above.
>
> I will try switching back too CFQ this evening to see what that does
> for 24 hours. I suspected it was either the SATA controller or the
> scheduler that fixed/helped things. With the quad-core I figure that
> at least I will never have to worry about the CPU being a bottle neck.
> I still see the load average jump up at peak times to as much as 60
> percent but its a rare event now. When I did a grep on a 450Mbyte
> file it jumped to 90 earlier.
>
> Matt
> _______________________________________________
> CentOS mailing list
> CentOS@xxxxxxxxxx
> http://lists.centos.org/mailman/listinfo/centos
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
