Am 14.09.2012 um 20:17 schrieb David Erickson: > Hi all- > I've got a number of jobs queued up (with a few waiting on others) on > a queue representing access to a physical hardware resource, thus the > queue has just 1 slot: ie: > > job-ID prior name user state submit/start at > queue slots ja-task-ID > ----------------------------------------------------------------------------------------------------------------- > 3995 0.75000 v120on793T derickso r 09/14/2012 14:00:57 > all.q@cr02 1 > 3996 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 3997 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 3998 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 3999 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4000 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4001 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4002 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4003 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4004 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4005 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4006 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4007 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4008 0.75000 v120on793T derickso qw 09/14/2012 04:34:57 > 1 > 4010 0.49840 v120on793T derickso qw 09/14/2012 04:49:24 > 1 > 4011 0.49840 v120on793T derickso qw 09/14/2012 04:49:24 > 1 > 4012 0.49840 v120on793T derickso qw 09/14/2012 04:49:24 > 1 > 4013 0.49840 v120on793T derickso qw 09/14/2012 04:49:24 > 1 > 4014 0.49840 v120on793T derickso qw 09/14/2012 04:49:24 > 1 > 4015 0.49840 v120on793T derickso qw 09/14/2012 04:49:24 > 1 > 4016 0.49840 v120on793T derickso qw 09/14/2012 04:49:24 > 1 > 4017 0.49840 v120on793T derickso qw 09/14/2012 04:49:24 > 1 > 4018 0.00000 v120on793T derickso hqw 09/14/2012 04:49:24 > 1 > 4019 0.00000 v120on793T derickso hqw 09/14/2012 04:49:24 > 1 > 4020 0.00000 v120on793T derickso hqw 09/14/2012 04:49:24 > 1 > > There are a couple tasks I'd like to be able to accomplish, but am not > sure how to: > > 1) Halt the queue from running any more work AFTER the currently > running job completes
Please check `man qmod`: $ qmod -d foobar.q > 2) After halting the queue (1), insert job(s) that will run BEFORE all > existing jobs when resuming the queue This you have to force by: a) requesting a resource which has a high urgency attached for these jobs b) use `qalter -p N <job_id>` and attach a high positive value for the POSIX priority: `qalter -p 1024 <job_id>` and check the order in the list of waiting jobs. Maybe the weight for the POSIX priotity needs to be increased in the scheduler configuration. Then you will have to enable the queue in question again. Depending on the scheduler setup you can also look into: c) `qsub -js <job_share> ...` d) `qalter -ot <override_tickets> <job_id>` to change the order of jobs. -- Reuti _______________________________________________ users mailing list [email protected] https://gridengine.org/mailman/listinfo/users
