High CPU Util

[yc_2]

Hi,

If there is a process (Oracle finance) showing in glance is 100% CPU util for 1 of the 4 CPU. Is there a way to lower the reading or it is normal.

The swapinfo PCT used was 40% to 60% before and I have increased the space. The reading now reads 12%.

Machine running N-class 440Hz with HP-Ux 11.00.

Thanks in advance,
YC

[Stefan Farrelly]

A process heavy on CPU can only ever use a max of 100% of one of the CPU's. Nothing much you can do about that. If the application were written a little better it would use multiple processes which share the CPU load over multiple CPUs (one process per CPU), but thats Oracle for you.

If you have and USED swapspace then you have memory contention. We run all our servers here with 0 % USED - which means nothing is paged out from memory to disk. Whats the total MB (megabytes) used as report by swapinfo -mt ? Whatever that figure, thats how many Megabytes have been paged to disk. Not good - thats affecting your performance a lot. You should look at adding more memory or tuning Oracle to not use so much memory, then your performance will pickup a lot.

[eran maor]

Hi

i dont think that 100% cpu is good . ( i know that you have 4 cpu )

my advise is to

1. check the if the patches of the oracle is install .

2. try to increase the memory for the oracle .

3. check if the size of the bufer cashe in your system . i know that the best perform. is 10% bufer cashe for oracle .

hope it will help you

[Rita C Workman]

Leong,

I 100% agree with Stefan. I don't like CPU's getting hit that hard...but.....My first goal is always to try to run everything you can in memory. Swapspace (jsut my 2 cents) is only good for a dump file, and the final resort in the event of paging out. You need to look at what you can tune in parms with your configuration, you might check and see if you have swapmem turned on so the system can dynamically control swapping. Use your sar -w command to see how much your paging out....then based on your memory (possibly you need more memory) see what you can adjust to make sure you're using as much of it as you can.

Just my thought,
Rita

[yc_2]

Hi,

Thanks for the advises. Just to make sure I didn't convey the wrong message in my earlier message. The following is the reading I got:

swapinfo gives:
Mb Mb Mb PCT
TYPE AVAIL USED FREE USED
dev 4096 0 4096 0%
dev 17366 0 17366 0% reserve - 2197 -2197
total 21462 2197 19265 10%

glance in memory report gives:
Event Current Cumulative
Page Faults 950 5980
Page In 358 4424
Page Out 4 53

1) Is there a way to determine which process cause the heavy page in?
2) Any parameter I can tune to reduce it?


Thanks,
YC