Unusually High Load Average after Reboot

[Russ Walicki]

everyone,

i have a K570 4-way HP Server which serves as an Oracle database for our plant. Recently, I rebooted the server and after reboot whenever I do a top the Load Average on the box is anywhere from 3-4.5! This is very high and never seems to drop below this point. The funny thing is that on the weekends when the plant isn't working and no backups or intensive jobs are scheduled to run the Load Avg is still in the same range. Any ideas why can't i get it to drop below this range, especially since before I rebooted the Load Avg was at 0.5 - 1.5?

thanks in advance,

Bob Morey

[steven Burgess_2]

Hi

Have you checked for any processes that appear to be utilising the cpu ? This would put other processes in a queue, causing the load to increase. Have you attached any new hardware recently such as a tape drive or a cdrom unit ?

HTH

Steve

[Patrick Wallek]

Did you change any kernel parameters before you rebooted?

I would check the 'timeslice'parameter and make sure it is at 10 and not at 1.

[Steven E. Protter]

run top, see what your top process is.

If you have automated find commands that look for files to delete and whatnot, they can really run up the system load quickly.

If the top process is oracle related, you need to look at what you changed.

With oracle its important(I think you already know this) to have more than adequate size in your message queues and shmmax parameter.

Oracle is a big user of shared memory and message queues. apache, and other things are also big users of that capacity and could be pinching oracle's capacity and causing bottlenecks.

glance even if not licensed can be used for 60 days trial and can also help diagnosis.

I'm posting in a prior thread of I responded to. It's got a variable term sar data collection script in it that really helps identify problem processes and resource bottlenecks.

My first post in the link below has the script as an attachement.
Link:
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0xfc5035a43b46d71190080090279cd0f9,00.html
SEP

[Krishna Prasad]

Also, since you just rebooted make sure all your CPU's are active - It's possible one could have been marked bad during the reboot.

Also check them amount of memory for the same reasons.

[Russ Walicki]

Thanks to all who responded. It looks as if on startup we had a very large number of 'find' processes which attempt to do some clean up and data collection from other machines to load into Oracle. They took forever after reboot and once completed the load_avg is back to normal.

thanks for all your help.

Bob