Re: [Condor-users] Looking for suggestions for inhomogeneous pool

[Steffen Grunewald <Steffen.Grunewald@xxxxxxxxxx>]

On Tue, Nov 22, 2011 at 07:15:03AM -0500, Ben Cotton wrote:
> On Tue, Nov 22, 2011 at 4:09 AM, Steffen Grunewald
> <Steffen.Grunewald@xxxxxxxxxx> wrote:
> 
> > Is there a means to overrule the current NEGOTIATOR_PRE_JOB_RANK setting
> > for parallel universe jobs only? Another term in the expression, like
> >  1000000000*(Target.JobUniverse == 11)
> 
> Perhaps using ifThenElse? For example:
> 
> NORMAL_PRE_JOB_RANK = 1000000 - 100 * Memory - $(SWEEP_DIFF)
> PARALLEL_PRE_JOB_RANK =  1000000000
> NEGOTIATOR_PRE_JOB_RANK = ifThenElse( (Target.JobUniverse==11),
> $(PARALLEL_PRE_JOB_RANK), $(NORMAL_PRE_JOB_RANK) )

I still can't see whether this makes a difference (the part detecting parallel
jobs should be stronger than the memory part).

But I always get confused by the Target/My prefixes, and again this time I
suspect I got it wrong - parallel universe jobs, with a

NEGOTIATOR_PRE_JOB_RANK=1000000000 + 1000000000 * (TARGET.JobUniverse == 11) * TotalCpus - 1000 * Memory

appear to be scheduled the same way as vanilla jobs. (The intention is to
schedule small machines first and keep the big ones for parallel runs, thus
populating them "bottom up" with vanilla and "top down" wich parallel jobs.
Obviously the naive approach doesn't work.)

This is certainly a PEBKAC but I cannot find any documentation how to proceed
(e.g. how do I debug the P_J_R being used by the negotiator which matching
parallel jobs?)

Cheers,
 S

-- 
Steffen Grunewald * MPI Grav.Phys.(AEI) * Am Mühlenberg 1, D-14476 Potsdam
Cluster Admin * --------------------------------- * http://www.aei.mpg.de/
* e-mail: steffen.grunewald(*)aei.mpg.de * +49-331-567-{fon:7274,fax:7298}