System Performance

[Brian K. Arnholt]

I have an application that contains several individual modules that run on HP C240s running HP-UX 11.00. Over time (~ 1 month) I notice that system performance greatly diminishes until one of the modules needs to be stopped and restarted and then the system performance returns to an acceptable level. Also, system reboots also free up resources.
Are there any diagnostic tools that can help determine the problem?

[Alan Riggs]

I hesiate to draw a conclusion based on scant onformation, but what you describe sounds characeristic of a memory leak. It is possible that you are experiencing RAM pressure as the memory requirements of your application grow over time. Good tools to track down memory leaks are

glance
UNIX95= ps -e -o "user,pid,ppid,sz,vsz,comm" | sort -rn +4|more
vmstat

[Sanjay_6]

Hi Brian,

Look at the thread below.

http://us-support2.external.hp.com/iv/bin/doc.pl/sid=c70d82d8044299a687/screen=ivHome/?NODEID=English_SHW::WW_SW_UX_SYSADM_EN_E/Q1.7&WARP=1

Also do a search in forum for system performance and you will get a lot of earlier postings. They may also help you.

hope this helps.

thanks

[Uday_S_Ankolekar]

Hi,

Use this command that will show you who is eating up resource (I got this from Forum)
UNIX95= ps -e -o ruser,vsz,pid,args | sort -rnk2 | more
Also you can monitor with Glance for the realtime status.
GoodLuck
-USA..

[James R. Ferguson]

Hi Brian:

Take a look at your system with 'glance'. If you don't have a permanent license for it, you should, but in the mean time you can load a trial version from one of the Application CDROMs. 'glance' has built-in help and 'man' pages, too.

In the case you describe, you might also want to look for potential memory leaks. Track your application over time by doing this:

# UNIX95= ps -e -o "user,vsz,pid,ppid,args"|sort -rnk2|more

You can also use the old standby tools like 'sar' and 'vmstat'.

Regards!

...JRF...

[Roger Baptiste]

Brian,

Firstly, you would need
to figure out where is
the resource overload?
Is it CPU-bottleneck? or
Memory? or I/O? or swap?
To find this, use glance tool.
You can run sar command
to get a quick view on how
the CPU time is divided.

Definitely, more info is
needed to point out directions
to the source of the problem.

Glance is a good starting point. Also, look at swapinfo to make sure
you are not running out of swap.

-raj

[Brian K. Arnholt]

Thanks all who replied. I discovered that I have two processes that grow over time. When initially started they are about 1 MB each. Over a month, one gets to about 25 MB, the other gets to about 100 MB. Upon stopping and restarted the processes, they return to a "normal size". Now I have to figure out why they are growing and how to fix it. Increasing swap space seems to be the easiest fix. These processes use a rpc method to transfer data from other systems.