[GE users] Queue definition for 4 Queue

Stephan Grell - Sun Germany - SSG - Software Engineer stephan.grell at sun.com
Wed Jan 11 08:43:02 GMT 2006


    [ 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. ]

Hi Sebastian,

based on your description, you have 2 myslots on each host. Every
running job
requests myslots, regardless of the queue it is running in. When you do
a qstat -F myslots
you will most likely see, that all myslots are consumed or that there
are not
enough left to run your lam2 job.
Though, I would assume, that you setup is working. If you use all your
queues
to mimic job priorities, you might want to look into the queue
subordinate concept.

Cheers,
Stephan

Sebastian wrote On 01/10/06 13:32,:

>I tried a little bit with complex values and defined this:
>
>for each execution host the consumable attribute myslots1 and myslots2 with
>the value 2.
>
>then i defined two complexes myslots1 and myslots2 which are consumable and
>requestable with default=1.
>
>i gave the queues asap.q, high.q and normal.q the complex myslots1=2 and the
>test.q the complex myslots2=2.
>
>I thougt this can really work, but it doesn't
>There is a error "cannot run because available slots combined under PE lam2
>are not in range of job"
>
>I just have 2 lam's, all other jobs are running in lam1, why did i get this
>error for a other PE named lam2?
>
>Kind Regards, Sebastian
>
>  
>
>>--- Ursprüngliche Nachricht ---
>>Von: "Sebastian" <ChiefShadow at gmx.de>
>>An: users at gridengine.sunsource.net
>>Betreff: [GE users] Queue definition for 4 Queue
>>Datum: Tue, 10 Jan 2006 09:28:59 +0100 (MET)
>>
>>
>>Hello,
>>
>>we use acutally 10 dual-core nodes, that mean 20 cores.
>>We run our jobs with LS-DYNA and LAM/MPI, so i just configured Reuti's
>>Integration, which works.
>>
>>We almost want only have one job per core, and because of performance we
>>use
>>the allocation_rule $fill_up, which works.
>>
>>Now we made 4 Queues, which are configured like these:
>>
>>queue_name=test.q   Job's Nice=-15   PE=lam2
>>queue_name=asap.q   Job's Nice=-10   PE=lam1
>>queue_name=high.q   Job's Nice=1     PE=lam1
>>queue_name=normal.q Job's Nice=15    PE=lam1
>>
>>Both lam-PEs have 20 Slots.
>>
>>Now i think, the queues asap.q,high.q and normal.q share all available
>>core's and the test.q can run closed.
>>
>>But when i submit two jobs, one to high.q and another to normal.q, they
>>run
>>but on one node there run 4 LS-DYNA instance's, that mean 2 jobs per core.
>>
>>I think i have to define complex values in the queues asap.q,high.q and
>>normal.q or something like that, that the queues work interactive and
>>share
>>there infos about the used nodes, or how can i make this work?
>>
>>Thanks, Kind Regards,
>>
>>Sebastian
>>
>>-- 
>>Lust, ein paar Euro nebenbei zu verdienen? Ohne Kosten, ohne Risiko!
>>Satte Provisionen für GMX Partner: http://www.gmx.net/de/go/partner
>>
>>---------------------------------------------------------------------
>>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