Please help !! Strange behavior on HPUX 11i (vmstat)

[Qi Jian Gang]

Hi there,

I have a HP box rp8400 running HP-UX 11i, with
8 CPUs, 12GB memory, 4.0GB primary swap and 9.8GB seconday swap space.

The "vmstat 2" output really bewilder me.
I've encountered 2 scenarios whereby the
output of the command really strange.

Scenario 1:
The column sr under subject page are of 3 digits sometimes even 4 digits, while the fr
and po column remains 0, this kind of scenario
will last for at least more than 10 minutes.
Why the kernel scan the memory so frequently
while do not steal pages or page out some memory ? Under what circumstances will the kernel scan memory so fast but doing nothing ?


Scenario 2:
The re and at column under subject page are
remain high value. re is 15??, at is 25??
it'll last very long time, while other column
under page are 0.
What activity the memory are involved ?

Please help,,, I am so wondered :-()

Thanks in advance
regards
Jiangang

[Stefan Farrelly]

It sounds like what ever application you are running is causing these strange statistics. What applications do you have running ?

you may need to use glance/gpm to monitor in detail the heaviest cpu processes to see which ones are causing this.

[Qi Jian Gang]

Hi Stephen,

The system is running oracle9.2, and the
application using oracle is Wall Street 4.0.

I've used glance to monitor the system, but
seems the memory got no problem,,, except the
sr issue.

Regards
Jiangang

[Steven E. Protter]

I'd be the output of vmstat over a longer time period.

You might try using the attached script to collect some data, maybe modify it and add a line for vmstat data.

Take note of what is going on during the data collection period. I like the way your box is configured.

SEP