Which Processes causing High disk utilization

[arun khatwani]

I have got following request from client,

The Disk Utilization of the nlxsl167 has increased to about 100% since August 2nd (

Is it possible to find which process(es) causes this high disk utilization?

Can anybody help me to sort it out, Is is possible to find out which preocesses were using high disk utilization on 2nd august.

[Steven E. Protter]

Shalom,

Gather some general data:
http://www.hpux.ws/?p=6

Note these utilities are sar based and not good for directly identifying the process. They do identify the resource in question. If its the disk you stick your oracle data on you are close to identifying the process.

glance/gpm has a good tool set for identifying the process doing the deed. There is a 60 day trial version on the Core OS DVD

SEP

[Nido]

Arun,

Current Disk utilization can be found by running glance.

You need to check perfview performance data to get the historical data, launch pv in graphical mode and dig-down to get the desired performance data, export the data in excel format if you wish to or run extract from the command line.

You will find alot of threads talking about pv/extract/

Cheers!!

[R.K. #]

Hello Arun,

You can use:
# sar -d 5 10 -- to get the stats for all the disks

If you have identified the disks with high utilization, then look for what kind of LV they are associated with:
# strings /etc/lvmtab -- to identify the volume group associated with the disks.
# lvdisplay -v /dev/vgXX/lvolX -- where x represents the lvol name. This will tell you what disks are associated with the logical volume.
# bdf -- to see if this volume groups files sytems are full ( > 85%)
# more /etc/fstab -- to determine the file system type assiciated with the lvol/mountpoint

You can try doing following to improve disk I/O:
1. Reduce the volume of data on the disk to less than 90%
2. Stripe the data across disks to improve I/O speed
3. If you are using Online JFS , run 'fsadm' to defragment the extents.

Hope this helps..

[Alzhy]

Without glance it will be a bit tad difficult -- but not entirely impossible.

Do this sir:

sar -d 5 10 (it will give you your busiest disk or disks - look for high blk/s or %busy)

find what LVM VG or VxVM DG the disk or disks belong to (vgdisplay or vxprint)

For each logical volume (lvm) or volume (vxvm) -- find the coresponding mount points (bdf)

Do an fuser for each filesystem -- it will give you the process IDs of active users on the filesystem

Pretty convuluted but it can help you hunt down the culprit/busy process.

However -- 90% of the time, the IO hoggers would be your busiest/long running process which you can find using "top" whic should be built-in to your HP-UX OE.

HTH

[Mel Burslan]

>Is is possible to find out which preocesses were using high disk utilization on 2nd august

If you need the analysis of historical data, you need to have been logging the performance data using measureware or maybe with some other tool your enterprise environment is using for performance analysis.

ls -l /var/opt/perf/datafiles/log*

if you see large files with today's timestamp, then you might be in luck. You can generate an output using extract utility. Unfortunately, I did not and do not use this utility to help you there to tell the exact syntax of it. You need to do some reading on the man pages for extract to see how you can extract the information you are looking for.

Good luck