[GE users] high CPU load for sge_qmaster

Stephan Grell - Sun Germany - SSG - Software Engineer stephan.grell at sun.com
Mon Apr 25 16:02:01 BST 2005


Hi,

can you take a look at the qmaster using qping when it is really
busy? Would be nice to know, if there is a lot of communication
going on.

A output of the scheduler profiling would also be very very helpfull.
It would be enough to have 2 or 3 scheduling runs in the profiling
log.

Cheers,
Stephan

Kees Verstoep wrote:

>Hi,
>
>I am currenly working with an SGE-6.0u3 setup for a cluster with
>40 compute nodes.  It is running fine, but what I often see a while
>after the sge_qmaster has started, it will come into a mode where
>it constantly takes about 45% CPU time.  This is even the case
>when there are no jobs at all in the queue.  Looking with strace
>reveals that sge_qmaster is constantly doing gettimeofday() calls.
>Other than the high CPU overhead on the head node due to this,
>it is running quite fine.  After restarting the sge_qmaster again,
>I see the same pattern: it has low load for a while (even when
>it has to start jobs), but then goes into this "polling" mode again.
>
>Anyone seen this behaviour as well?  This is on a dual PIII-1GHz
>system running RedHat Enterprise Linux version 3.4.
>
>Thanks!
>Kees Verstoep
>
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
>For additional commands, e-mail: users-help at gridengine.sunsource.net
>
>  
>


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list