[gridengine users] jobs in qw stats but have enough empty slots

Reuti reuti at staff.uni-marburg.de
Mon May 27 22:03:22 UTC 2013


Hi,

Am 27.05.2013 um 15:53 schrieb Fan Dong:

> Hope someone can help with this.  We submitted hundreds of jobs using something similiar to qsub -pe my_pe 4 my_job.sh.  We found that there is always a nodes with 8 slots empty at any time that we checked.   A screenshot is pasted here, either comp03 or comp04 is idle while there are bunch of jobs waiting in the queue.  Ideally both comp03 and comp04 should have 2 tasks running all the time.  Given our setup we expect 6 jobs running simultaneously but there are only 4 jobs instead.  These are long-run jobs that each of them may last 4 hours so we checked the queue plenty of times and find this behaviour.
> 
> Can someone shed some light on this?
> 
> ---------------------------------------------------------------------------------
> npairs.q at comp01.neuroinfo.rri  BIP   0/4/6          2.45     linux-x64     
>    1344 0.55500 npairs_run anita        r     05/27/2013 01:20:46     4        
> ---------------------------------------------------------------------------------
> npairs.q at comp02.neuroinfo.rri  BIP   0/4/6          2.71     linux-x64     
>    1343 0.55500 npairs_run anita        r     05/27/2013 00:37:16     4        

The above machines have only 6 slots. What is the definition of your PE - maybe they don't fit in the allocation rule?

$ qconf -sp my_pe

-- Reuti


> ---------------------------------------------------------------------------------
> npairs.q at comp03.neuroinfo.rri  BIP   0/8/8          5.46     linux-x64     
>    1345 0.55500 npairs_run anita        r     05/27/2013 02:17:01     4        
>    1346 0.55500 npairs_run anita        r     05/27/2013 03:11:31     4        
> ---------------------------------------------------------------------------------
> npairs.q at comp04.neuroinfo.rri  BIP   0/0/8          0.02     linux-x64     
> 
> ############################################################################
>  - PENDING JOBS - PENDING JOBS - PENDING JOBS - PENDING JOBS - PENDING JOBS
> ############################################################################
>    1347 0.55500 npairs_run anita        qw    05/23/2013 16:37:25     4        
>    1348 0.55500 npairs_run anita        qw    05/23/2013 16:37:25     4        
>    1349 0.55500 npairs_run anita        qw    05/23/2013 16:37:25     4        
>    1350 0.55500 npairs_run anita        qw    05/23/2013 16:37:25     4        
>    1351 0.55500 npairs_run anita        qw    05/23/2013 16:37:25     4        
>    1352 0.55500 npairs_run anita        qw    05/23/2013 16:37:25     4        
>    1353 0.55500 npairs_run anita        qw    05/23/2013 16:37:25     4        
>    1354 0.55500 npairs_run anita        qw    05/23/2013 16:37:25     4        
> 
> 
> 
> On 05/17/2013 08:00 AM, users-request at gridengine.org wrote:
>> Send users mailing list submissions to
>> 	
>> users at gridengine.org
>> 
>> 
>> To subscribe or unsubscribe via the World Wide Web, visit
>> 	
>> https://gridengine.org/mailman/listinfo/users
>> 
>> or, via email, send a message with subject or body 'help' to
>> 	
>> users-request at gridengine.org
>> 
>> 
>> You can reach the person managing the list at
>> 	
>> users-owner at gridengine.org
>> 
>> 
>> When replying, please edit your Subject line so it is more specific
>> than "Re: Contents of users digest..."
>> 
>> 
>> 
>> Today's Topics:
>> 
>>    1. Where do the factors for np_load_short come from?
>>       (Tim Landscheidt)
>>    2. Re: Where do the factors for np_load_short come	from? (Reuti)
>>    3. Re: Where do the factors for np_load_short come	from?
>>       (Tim Landscheidt)
>>    4. Re: Where do the factors for np_load_short come	from? (Reuti)
>> 
>> 
>> 
>> _______________________________________________
>> users mailing list
>> 
>> users at gridengine.org
>> https://gridengine.org/mailman/listinfo/users
> 
> _______________________________________________
> users mailing list
> users at gridengine.org
> https://gridengine.org/mailman/listinfo/users





More information about the users mailing list