>> That I have seen too, but I doubt this is the case now. Namely the level is
>> maintained and jobs are scheduled and run. Just not beyond a certain
>> threshhold and the error is reservation issue. If I stop maui and start
>> pbs_sched, then it will happily schedule jobs. Moving back to maui it'll let
>> the level drop back to its comfort zone and then start scheduling jobs
>> again. The scenario you describe usually leads to full blocking and you see
>> just job count dropping, not a flow of jobs. Also, we use nodes with 27-36
>> job slots so a single job can't really block a node.
>
> I agree, it could be possible just one of a many maui limitations,
> it's a free product, not supported by any conpany at any agreement level.
> So, in case you did the check and not found any jobs in the "W" state
> with node assigned to them, I just give up for that problem.
> One dumb question is: the site purchase a plenty amount of h/w,
> now it is the time to think about to buy s/w too.
> Sure, I'm not sponsored by any s/w company and use free s/w only,
> (P.S.) and ohh! for the time being certainly.
I would consider it a maui limitation if I hadn't seen about a month ago a
state where maui had scheduled 4930 jobs (would have more, but more weren't
coming at that time). Therefore the 3500 or 4200 level seems arbitrary and more
a dynamic state that causes trouble for maui.
With regard to commercial software I'm somewhat limited in choices as it has to
work with EMI middleware. Replacing maui with moab would be an option, but from
what I've heard from people who have purchased moab, the core is the same and
the problems are similar so none of them plan to extend their subscription and
move back to maui or even better we're all lobbying for EMI to support SLURM
instead.
However right now the error messages show that his is a maui problem with
reservations. If noone has any ideas I'll probably try to move to 3.3.1 though
it's not available as RPM from EMI so I'll have to custom compile it and I'd
rather not have custom built tools as that's not a sustainable method of
deployment in the long run. But as I'm guessing the 3.3 maui client tools can
talk to 3.3.1 maui server, then I don't need to swap it out anywhere beyond the
scheduling server.
Mario Kadastik, PhD
Researcher
---
"Physics is like sex, sure it may have practical reasons, but that's not why
we do it"
-- Richard P. Feynman
_______________________________________________
mauiusers mailing list
[email protected]
http://www.supercluster.org/mailman/listinfo/mauiusers
