[GE users] SGE imposing cpu-limits on jobs in queue.

Jesper Krogh jesper at krogh.cc
Thu Jan 24 16:28:29 GMT 2008


    [ The following text is in the "iso-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]


The same job running in the all.q consumes 100% in CPU time.

Jesper

> SGE should just execute the job or suspend the job, but I don't think
> SGE can make the job only consume 50% of the CPU time (even if you set
> "reprioritize" in sge_conf(5)), the OS process scheduler still tries
> to run something and not let the machine idle.
>
> Do you get 100% CPU utilization when you run the jobs outside of SGE??
>
>
> Rayson
>
>
>
>
> On 1/24/08, Jesper Krogh <jesper at krogh.cc> wrote:
>
>> We have 3 queues..
>>
>>
>> longjobs which is subordinate to both all.q and shortjobs shortjobs
>> which is subordinate to all.q all.q for interactive jobs
>>
>> When nothing else is running in the cluster, a jobs running in the
>> longjobs queue is getting a cpu-limit imposed (dual-cpu nodes with 2
>> jobs have a load aound 1.5 and 50% idle in top).
>>
>> Where should I go scout for this?
>>
>>
>> Jesper
>> --
>> Jesper Krogh
>>
>>
>>
>> ---------------------------------------------------------------------
>> 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
>
>
>


-- 
Jesper Krogh


---------------------------------------------------------------------
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