In article <4rtnsa$e2@server2.codetel.net.do>, Tom Hay
<nomad@codetel.net.do> writes
>Yes, I've seen it too, on Version 7.10D1 on a uniprocessor HP-UX, and
>I don't like it either. The machine virtually stops for around 30
>seconds. Can't offer any explanation, though ... sorry ! Anybody else
>?
>
>Greg Moye <msmcl301.gmoye01@eds.com> wrote:
>
>>We have 7.13.UC2 running on Solaris 2.4 w/ 2 Sparc CPU's. I occasionally
>>see both CPU VP's chewing CPU like crazy, yet there's few if any Buffer
>>Reads or any other evidence of productive work going on for users. Then
>>they'll quiet down for awhile, again for no apparent reason. There's
>>not a lot of real user activity on the system.
>
>>"top" shows about 1/3 of time as user, 2/3 as kernel.
>
>>Response to our client app seems OK and correct, can't find anything else
>>apparently wrong. It still smells kinda fishy though. Anybody seen
>>this or have any idea what the CPUVP's are doing??
>>--
>>Greg Moye
>>Electronic Data Systems Corp. - Plano TX
>
>
have you tried running all the onstat options to see what is
happening? Long spins? checkpoints?
onstat -g act - which threads are active.
onstat -g rea - which threads are ready to run.
onstat -g sch - long spins
onstat -D - any disk I/O
onstat -B - page cleaning
onstat -F - page writes
go into onmonitor as informix and run Status->Profile and hold down
the return key - which values change?
E-mail me if you want more info.
--
David Williams