load average in HP Ux

[Adithyan]

Hi,

In the uptime command the load average is showing as 68 67 68.

What is the command to watch the load average for certain period of time ?

Thanks.

[Peter Godron]

Hi,
check out the sar command "man sar"
for example to check cpu
sar -u 5 5

Please read:
http://66.34.90.71/ITRCForumsEtiquette/after.html

Please also read:
http://forums1.itrc.hp.com/service/forums/helptips.do?#33
on how to reward any useful answers given to your questions.

So far you have not awarded any points !

[Steven E. Protter]

Shalom,

The numbers indicate a lot of processs waiting.

This may not be a problem. It could be a lot of sleeping processes caused by bad application design.

If the system is performing normally, no action need be taken.

To find performance problems there are a few ways:

http://www.hpux.ws/system.perf.sh
User complaints.

Both work well.

Please read Peter's links and follow his advice.

SEP

[Yogeeraj_1]

hi Adithyan,

You can also use measureware to get historical performance from collected data.

For more information, see man mwa

mwa - Performance tool script for starting and stopping data collection and alarms


hope this helps too!

kind regards
yogeeraj

[spex]

Hello,

'sar' does not capture load average (CPU run queue) data, so you will have to look elsewhere. If you are willing to pay the licensing fee, MeasureWare is an excellent product. Alternatively, you could just set up the following cron job, which captures the 5-min. average:

0,5,10,15,20,25,30,35,40,45,50,55 * * * * uptime | awk '{gsub(/,/,"");print $1,$11}' 1> /var/opt/hist/load_avg.$(date +%Y%m%d).out

PCS

[Bill Hassell]

The load average does not describe the measurement correctly. It is the average size of the run queue which is the number of processes that are currently running or waiting for a CPU. If your system has 64 CPUs, then this is normal and represents a full load. If you have only 2 CPUs, then your system is massively undersized (3,000% load) for the workload. This of course assumes that your applications are working properly and are designed for efficiency.

There is a lot more to know about this situation: size of RAM, number of CPUs, type of application being run, number of network users, etc. And of course, does the system seem to respond quickly? It is certainly possible that your programs perform extremely short tasks. In this case, the system overhead will be quite high (more than 30%).

[ln_unix]

Hi Adithyan,

You can use 1)top ,or
2)glance
to show the load averages.
Basically these are the tools used in hp-ux for performance monitoring.
Native command is sar.

you can do man top , man glance , man sar
to get detailed information about these.
ok.

Regards,

Lokesh....