Disk I/O in hpux 11i.

[Shashi Kiran_1]

How can I check the Disk I/O per process.

[Francis_12]

Hello,

The best is to use Glance for a real-time troubleshooting (by selection the process from the process list and listing its specific I/Os) and Measureware if you want to get a complete history.

Hope this helps, Bye.

Francis DERDEYN - HP-UX ASCE.

[Bruno Ganino]

Use "ps -ef" for process active, or "ioscan" command.

[Hari Kumar]

#iostat -t
#sar -d 5 100 ( your own choice of intervals)
Check the GLANCE pack also here ---

http://www.openview.hp.com/products/gppak2k/index.html

[Francis_12]

Hello back again,

To my knowledge there is no other way than using Glance and/or MeasureWare to clearly identify which process is doing which I/O.

You can nevertheless install Glance for a trial period of 30 days.

Hope this helps, Bye.

Francis DERDEYN - HP-UX ASCE.

[Shashi Kiran_1]

Hi
Many thanks to those who replied and passed on useful suggestions.
Please let me know how can I check Disk I/O per process using Glance.

Regards,

[Francis_12]

Hello,

Easy. Just run 'glance'

Check the process PID.

Press 'S' to select and enter the pid.

You will directly see the logical read/writes and physical read/writes for that specific process.

At that point, if you press 'F4' you will the list of opened files for that process.

Hope this helps, Bye.

Francis DERDEYN - HP-UX ASCE.

[Steven E. Protter]

See the disk and buffer sections of the attached script.

Also check into the add in program lsof.

SEP

[Shashi Kiran_1]

Hi All,
We are using hpux11i and when we try to login 300 concurrent users (using load runner) root disk I/O goes 100% !!! Why?
Regards,

[Patrick Wallek]

It is possible that when you get that many users logged in your system is starting to page out (ie. swap).

Some things to check when this happens.

1) Via Glance, or GPM, check and see how much cpu the 'vhand' process is using. If it is using a lot this is indicative of paging activity.

2) 'swapinfo -tam' to see if you are using any of your swap. Your default swap /dev/vg00/lvol2, so you could see some activity here.

[Tim D Fulford]

Hi

Further to Patrick's comment. On our boxes we saw system memory go up from 100-200MB in 11.0 to 1.1GB in 11i. For this reason we have increased swap & physical memory in all the 11i boxes.

Tim

[Michael Steele_2]

SEP - can't read your attachements.

More than likely you have a file system that everybody's hitting. So please attach:

# sar -d 5 5

I realize a lot of glance users have a bias towards glance, but the GUI often never agress with the TUI and sar never has an issue.

If you have to use glance then use the glance advisor which by passes the GUI and works just fine.

Again, please attach 'sar -d 5 5' as well as these others:

sar -d 5 5
sar -v 5 5
sar -u 5 5
sar -b 5 5
vmstat 5 5
swapinfo -tam

Thanks!