Re: [Condor-devel] Per machine resource limits in 7.6/7.8

[Matthew Farrellee <matt@xxxxxxxxxx>]

Something like Rank on TotalCustomResource < CustomResource? Similar to 
TotalMemory, or maybe it's TotalSlotMemory, compared to Memory.

Though, those attributes would be on a slot ad and matchmaking would 
suffer from delayed advertising problem - only the slot that is changing 
state (consuming or releasing the CustomResource) would have current 
values for the attributes. Delay, of course, configurable with 
UPDATE_INTERVAL.

Best,


matt

On 05/01/2012 10:52 PM, Lans Carstensen wrote:
> Does anyone have a good answer for the opposite case?  We've hit it
> and I'm wondering if anyone has a better way.
>
> Software X is licensed per-machine but floating, and you can run
> unlimited copies of X on that machine.  Ideally I'd rank the unclaimed
> partitionable slot higher for X jobs if a dynamic slot is already
> running X, and then I'd also have a constraint to cap me at the
> maximum number of machines.
>
> The only solution for some of these software licensing optimization
> cases appears to be to adjust the limits up from worst case via a
> custom external observation daemon.  Anyone else dealing with that?
>
> -- Lans Carstensen
>
> On Tue, May 1, 2012 at 3:23 PM, Tim St Clair<tstclair@xxxxxxxxxx>  wrote:
> > IMHO local resource limits solves this problem nicely, which is eje's patch on master. Pssst - There's a fedora build for that ;-)
> >
> > Cheers,
> > Tim
> >
> > ----- Original Message -----
> > > From: "Dan Bradley"<dan@xxxxxxxxxxxx>
> > > To: condor-devel@xxxxxxxxxxx
> > > Sent: Tuesday, May 1, 2012 4:12:50 PM
> > > Subject: Re: [Condor-devel] Per machine resource limits in 7.6/7.8
> > >
> > >
> > >
> > > On 5/1/12 4:01 PM, Todd Tannenbaum wrote:
> > > > On 5/1/2012 3:52 PM, Alan De Smet wrote:
> > > > > I had an interesting chat with a Condor Week attendee who has
> > > > > some challenges that I'm not sure how to handle.  They have
> > > > > various limited resources (software licenses) , but the limits
> > > > > aren't global, they're per machine, so the concurrency limits
> > > > > aren't a good fit.  Furthermore a job might claim multiple
> > > > > identical resources at once, in much the same way a single job
> > > > > might claim multiple cores.
> > > >
> > > > Maybe something like so in the job submit file:
> > > >
> > > >    concurrency_limits = licenseX_$$(Machine), licenseY_$$(Machine)
> > > >
> > > > Actually, the above likely will not work because $$() is not
> > > > expanded
> > > > in the matchmaker... but if concurrency_limits is evaluated,
> > > > perhaps
> > > > you could do the above idea but via ClassAd string cat functions.
> > > >
> > > > Todd
> > >
> > > The job's concurrency limit string is read by the matchmaker before
> > > considering any machine, so trying to refer to machine attributes in
> > > the
> > > concurrency limit string just won't work.
> > >
> > > --Dan
> > >
> > >
> > > > > Custom resource limits per eje's proposal
> > > > > (https://condor-wiki.cs.wisc.edu/index.cgi/tktview?tn=2905) seem
> > > > > like a good fit, and he was open to the idea.  However, that's
> > > > > 7.9 territory, and they'd prefer to not use a development
> > > > > release.
> > > > >
> > > > > Is there a good solution for 7.6 or 7.8?  All I've got in having
> > > > > slots START expressions total up usage by other slots and testing
> > > > > against the limit, which seems clumsy.
> > > _______________________________________________
> > > Condor-devel mailing list
> > > Condor-devel@xxxxxxxxxxx
> > > https://lists.cs.wisc.edu/mailman/listinfo/condor-devel
> > _______________________________________________
> > Condor-devel mailing list
> > Condor-devel@xxxxxxxxxxx
> > https://lists.cs.wisc.edu/mailman/listinfo/condor-devel
> _______________________________________________
> Condor-devel mailing list
> Condor-devel@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/condor-devel