[SunHELP] clustered 280: high percentage of usr cpu time

[michael.petereit at materna.de]

Tige,

maybe I made a mistake while describing the situation:
top show me for example:
CPU states: 0.3% idle, 85.6% user,  4.3% kernel,  0.0% iowait,  0.0% swap

If I count all percentage shown in the top list below I do not get 100%.
Even the first process in the list eats up not more then 4% cpu time. 

Maybe top adds waittime to the usr process ?

Michael

-----Original Message-----
From: sunhelp-bounces at sunhelp.org [mailto:sunhelp-bounces at sunhelp.org]On
Behalf Of Tige D. Chastain
Sent: Wednesday, March 17, 2004 6:40 AM
To: sunhelp at sunhelp.org
Subject: Re: [SunHELP] clustered 280: high percentage of usr cpu time


Michael,

You could always attach a truss process to what you think is eating CPU
and see what it is doing.  What does top show you?  What processes are
getting CPU time the most?  

I administrate two V880s that are clustered and they didn't have the
same issue you are seeing.  That said, I was not running the
Availability Suite that you reference.  

Tige

On Wed, 2004-03-17 at 00:16, michael.petereit at materna.de wrote:
> Hi,
> 
> I have two 280s runnig in a cluster with two 3310 storages.
> Since the two machines became a cluster (formerly they acted as single
> machines) "prstat -a" and "sar" show me a high amount of usr time on the
cpu
> and very low idle time even there is nothing running on it. Every
ressource
> group is offline and the system should have idle time of 90%.
> The only new thing is the "availability suite" of SUN.
> I assume this software is eating up most of the cpu time on this cluster
> system.
> Am I right ?
> 
> Regards,
> Michael
> _______________________________________________
> SunHELP maillist  -  SunHELP at sunhelp.org
> http://www.sunhelp.org/mailman/listinfo/sunhelp
_______________________________________________
SunHELP maillist  -  SunHELP at sunhelp.org
http://www.sunhelp.org/mailman/listinfo/sunhelp