Checking disk util problem

[Heath Ramos]

I am running HP-UX 11.0 and my system is really slow right now.

When I run Glance, it says that disk util is 100%.

What could cause that and what should I be checking for?

[Con O'Kelly]

Hi Heath

The Disk Util metric in glance is not reallly a very accurate measurement of potential disk bottlenecks. It is only a measurement of one disk and the queue at that point in time.

You really need to use the motif version of glance - /opt/perf/bin/gpm. Then look at the IO by Disk Report under the "Reports -> Disk Info -> IO by Disk" menu.

Sort disks by Average Service time and/or request queue. If you are seeing high service times and/or large queues then you liklely have a disk bottleneck.

Cheers
Con

[Heath Ramos]

That menu shows the following:

8/4.10.0 Req Queue: 0
Disk%: 91.2
Phys IO RT: 680.1

The other disks were low.

[Con O'Kelly]

Hi Heath

I'd advise looking at service times. I/O bottlnecks can hard to track down, especailly if you are using storage arrays (eg xp, EMC etc).
Once you've identified the disk ith high service times then look at what filesystem(s) they are in.

Use 'gpm' instead of glance as you can get more detailed info and the flexibility to sort infor etc.

Attaching a performnace guide which may help your trouble-shooting.

Cheers
Con