[gridengine users] Temporarily stop new job submissions?
skylar2 at u.washington.edu
Tue Jan 23 16:22:25 UTC 2018
I guess it depends on how one wants to do maintenance - we actually do it a
third way by installing an outage calendar in each queue to block out the
resources for the duration of the maintenance. Jobs that request runtime
will either finish before the maintenance, or be held until after the
maintenance. Jobs without runtime or runtime exceeding one week (we try to
schedule all maintenance at least a week in advance) are killed as needed
when the exec hosts are shutdown/rebooted, but our users are aware of the
risks of running long jobs.
We have redundant schedulers and a pair of login/submission systems, so
aside from a few seconds to failover schedulers, users can submit
throughout any maintenance.
The JSV method will actually prevent new submissions but all of the other
methods will allow submissions to take place as long as the qmaster process
is available. Aside from GE upgrades that require clearing out both running
and pending jobs, we don't use the JSV method in practice.
On Tue, Jan 23, 2018 at 04:53:17PM +0100, Reuti wrote:
> > Am 23.01.2018 um 16:39 schrieb Chester Langin <cl.research at siu.edu>:
> > Is there a way to temporarily reject new job submissions? We are going to be doing scheduled maintenance on the cluster internal network and we want to prevent new jobs from being accepted starting a couple of days prior to this scheduled maintenance. After the maintenance, we want to return the scheduler to accept new jobs, again, as before. Is there a quick and easy way to do this?
> Just add yourself to the user_lists in SGE's configuration, essentially all others will be blocked.
> -- Reuti
> > --Chet Langin, SIU
> > _______________________________________________
> > users mailing list
> > users at gridengine.org
> > https://gridengine.org/mailman/listinfo/users
> users mailing list
> users at gridengine.org
-- Skylar Thompson (skylar2 at u.washington.edu)
-- Genome Sciences Department, System Administrator
-- Foege Building S046, (206)-685-7354
-- University of Washington School of Medicine
More information about the users