CPU utilization

[TMcB]

Hi folks,

On one of my HPUX servers, glance plus is intsalled and gives me the avarage and high CPU utilization.

Can anyone advise me as to how can I find this out on all my other servers (none of which have glance plus)

Thanks for your help

[A. Clay Stephenson · Accepted solution]

Hi Terry:

You can use the old UNIX standby's - sar and sadc.

sadc (System activity data collector) can be used to take snapshots on a fixed interval. This does not mean that you will capture peak just what ever the metric happens to be at that point in time.

Man sar and sadc for details.

[A. Clay Stephenson]

Hi again Terry:

Plan B - you can install the 30-day trial version of Glance on your other boxes from the Application CD's.

[Uday_S_Ankolekar]

Hi,

You can use sar command. sar -u 5 10

You can also use top cammand

man sar for details.

Goodluck..

-USA...

[Carsten Krege]

sar should do it for you. E.g.

# sar 5 5
measures 5 times the CPU utilization for 5 second each (good for average)

# sar 1 5
measures 5 times the CPU utilization for 1 second each (good for peak)

Alternatively you can use top(1m) or could install the trial license of Glance on your other unix boxes (expires after some weeks).

Carsten

[Santosh Nair_1]

Top is a really nice utility or you can get yamm (one of my favorites) from the HP porting center:

http://hpux.cs.utah.edu/hppd/hpux/Sysadmin/yamm-2.6.0

-Santosh

[Deshpande Prashant]

Hi Terry
For online monitoring you can use
#sar -u 1 5 or #top

If possible install glance on rest systems from app. CD.
You can capture/store history using measureware software or using sadc (/usr/lbin/sa1 and /usr/lbin/sa2 commands) in cronjobs.

Thanks.
Prashant

[TMcB]

Hello again,

I have been able to find out what I need from using sar, but I would now like to be able to run this as a cron job and monitor the systems for 24hours.
I have tried the command /usr/lbin/sa/sa1 and get thef following message :
sadc: Can't create/open data file: No such file or directory

Also, for 24hour monitoring, what would be the best intervalsto add in the cron job

Thanks for your help.

[A. Clay Stephenson]

HI:

You simply need to create the directrory /var/adm/sa assuming that is where the sa1 script has defined the datafiles directory.

I typically use a 20 min sampling interval.

[TMcB]

Thanks - cron job is now running.
This has created a file /var/adm/sa/sa17,
which is a binary file.
How do I then get the information I want from this file

Again, thankyou.

[Volker Borowski]

Hi Terry,

if you want a surveiliance for several systems and may be even statistics later, you might consider to try

http://www.bb4.com

Several monitoring agents for this tool appear at http://www.deadcat.net

Hope this helps
Volker

[Uday_S_Ankolekar]

Hi,

You can also try sarcheck utility for a month free.
www.sarcheck.com for more details.
-USA..

[A. Clay Stephenson]

To now get the data from the sa* files, you simply do a sar, e.g. sar -u will display CPU utilization using today's statistics files. You can specify a -f option to choose another day's data.

Any options that sar supports with a number of samples/sampling period set of arguments will not work instrad on today's file by default.


Man sar for all the possible options.

Clay