Hi,
default_duration set to INFINITY.
N.
----- Message from Reuti <[email protected]> ---------
Date: Tue, 12 Mar 2013 10:18:54 +0100
From: Reuti <[email protected]>
Subject: Re: [gridengine users] Parallel and single-core jobs - same queue
To: [email protected]
Cc: [email protected]
Hi,
Am 12.03.2013 um 04:42 schrieb Nizar Abed:
What could be the reason preventing it from working?
Now the situation, in the queue 900 cores used(single-core jobs),
20,000 waiting, all single-core, and first in the queue is a
parallel job requesting 200 slots, but it's not dispatched, it's
starving.
I changed scheduler's max_reservation(qconf -msconf) to 400, and
added '-R y' to qsub.
what is the value of "default_duration" in the scheduler
configuration? The reservation will work best, if suitable runtimes
are specified at submission time for the job to avoid false
backfilling.
-- Reuti
Could it be clashing with fair-share policy or any of these
scheduler parameters?
weight_tickets_functional 10000
weight_tickets_share 0
share_override_tickets TRUE
share_functional_shares TRUE
max_functional_jobs_to_schedule 200
report_pjob_tickets TRUE
max_pending_tasks_per_job 50
halflife_decay_list none
policy_hierarchy SFO
weight_ticket 1.000000
weight_waiting_time 0.000000
weight_deadline 3600000.000000
weight_urgency 0.100000
weight_priority 0.800000
Thanks,
N.
----- Message from Reuti <[email protected]> ---------
Date: Mon, 11 Mar 2013 13:22:39 +0100
From: Reuti <[email protected]>
Subject: Re: [gridengine users] Parallel and single-core jobs - same queue
To: [email protected]
Cc: [email protected]
Hi,
Am 10.03.2013 um 15:49 schrieb Nizar Abed:
Wondering what's the best way settings a queue for single-core
and parallel jobs.
Assume a 1000 cores queue, most of the time, 1000 single-core
jobs running, 20,000 single-core jobs in 'qw', and we have fair
share policy applied.
If a user submit's a parallel job, (say 200 slots), the job
remains first in 'qw'(fair share), but not dispatched(?)
What is the best and efficient way of setting up a 'mixed' queue?
http://www.gridengine.info/2006/05/31/resource-reservation-prevents-parallel-job-starvation/
-- Reuti
All the best,
N.
_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users
----- End message from Reuti <[email protected]> -----
_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users
----- End message from Reuti <[email protected]> -----
--
110 105 122 97 114 32 97 98 101 100
System Administrator
Hebrew University of Jerusalem, 91904
Edmond J. Safra Campus, Givat Ram, Jerusalem, Israel
Inst. of Life Science
Dept. of Neurobiology/Neurophysics
Interdisciplinary Center for Neural Computation - http://icnc.huji.ac.il
Edmon and Lily Safra Center for Brain Sciences - http://elsc.huji.ac.il
Mobile: 972-54-88-20941
Telephone: 972-2-65-84203
Video: vc.elsc.huji.ac.il
Fax: 972-2-65-86152
----------------------------------------------------------------
This message contains information which may be confidential and
privileged. Unless you are the intended addressee (or authorized to
receive for the intended addressee), you may not use, copy or disclose
to anyone the message or any information contained in the message. If
you have received the message in error, please advise the sender by
reply and delete the message . To the extent contractual
confidentiality obligations exist, this message and all information
transmitted with it are designated "Confidential".
_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users
