Hi,

You will need slot reservation for the parallel job. Mainly as a starting point:

$ qconf -ssconf
...
max_reservation                   20
default_duration                  8760:00:00

and submit the parallel jobs with "-R y". The best would be to supply all jobs 
(serial and parallel) with a sensible expected run time by "-l h_rt=…". This 
will both reserve slots for the large parallel job (which may reserve some 
slots which keep them idle until enough slots are collected) and backfilling, 
which will still allow serial (or smaller parallel jobs) to start on idle 
reserved cores if it's known for sure, that they will finish before the last 
necessary slot for the parallel job becomes free.

-- Reuti

http://gridengine.org/pipermail/users/2012-July/004090.html


> Am 24.10.2017 um 01:37 schrieb System Administrator 
> <[email protected]>:
> 
> Hi all, we have an issue here with jobs with higher priority and more slots 
> requested are being preempted by jobs with lower priority and less slots.  
> Here is an example of qstat output:
> 
> [email protected]  BIP   0/40/41        37.34    lx26-amd64
> 126841 0.50500 utgovl_can X       r     10/22/2017 04:52:08     8 342
> 126841 0.50500 utgovl_can X       r     10/22/2017 17:46:38     8 344
> 126841 0.50500 utgovl_can X       r     10/22/2017 20:38:53     8 346
> 126841 0.50500 utgovl_can X       r     10/23/2017 09:51:23     8 366
> 126841 0.50500 utgovl_can X       r     10/23/2017 12:58:38     8 368
> 
> ############################################################################
> - PENDING JOBS - PENDING JOBS - PENDING JOBS - PENDING JOBS - PENDING JOBS
> ############################################################################
> 127048 0.84914 P15c7dbbc3 X       qw    10/23/2017 09:51:09    16
> 127049 0.84914 Pb8bb9f392 X       qw    10/23/2017 09:51:09    16
> 127050 0.84914 Peefd90c57 X       qw    10/23/2017 09:51:09    16
> 127051 0.84914 P000c3fd3f X       qw    10/23/2017 09:51:09    16
> 127052 0.84914 Pe82253e45 X       qw    10/23/2017 09:51:09    16
> 127053 0.84914 P9b9212411 X       qw    10/23/2017 09:51:09    16
> 126841 0.50500 utgovl_can X       qw    10/14/2017 20:22:30     8 369-552:1
> 126842 0.00000 canu_canu1 X       hqw   10/14/2017 20:22:30     1
> 
> In this example, once one of the utgovl_can tasks finishes, it frees up 8 
> slots and another utgovl_can task is started.  What we need to happen is the 
> queue needs to wait until 16 slots are free so the higher priority jobs will 
> run.  How to accomplish that?
> Thanks
> _______________________________________________
> users mailing list
> [email protected]
> https://gridengine.org/mailman/listinfo/users


_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users

Reply via email to