Re: [HTCondor-users] affecting user priority

["Joan J. Piles" <jpiles@xxxxxxxxx>]

Hi,

We came across a similar issue. Our point of view is that SlotWeight 
should match the resources other users are deprived of, not only those 
actively used. That is, if we have a 4 cores, 8 Gb RAM machine, then a 
job requesing 8Gb RAM but only one core could as well request all the 4 
cores, since nobody else will be able to use the other three.

Thus, as our average is 2Gb / core, we allow for an extra 2Gb (because 
we also have smaller jobs that usually can fill the gaps), and then we 
match the ram with the cores requested:

EquivalentCpus = ( Memory - 2048 ) / 2048
SlotWeight = ifThenElse(Cpus > $(EquivalentCpus), Cpus, $(EquivalentCpus))

Joan

On 01/05/13 20:11, Matthew Farrellee wrote:
> On 05/01/2013 08:16 AM, Cotton, Benjamin J wrote:
> > On Tue, 2013-04-30 at 17:39 -0400, Rita wrote:
> > > Why do you have, SlotWeight="Cpus*1000+memory"?  What is the
> > > significance of 1000, is this the request_memory * 1000 or some other
> > > factor?
> > >
> > >
> > > Why not just have SlotWeight=Cpus+Memory ?
> >
> > It's a matter of scale. CPUs are measured in ones while Memory is
> > generally in thousands these days. There's no reason your second setting
> > wouldn't work, but the one Matt suggested is probably closer in spirit
> > to what you're asking for.
>
> I just cut&pasted the policy example Todd gave.
>
> Best,
>
>
> matt
> _______________________________________________
> HTCondor-users mailing list
> To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx 
> with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users
>
> The archives can be found at:
> https://lists.cs.wisc.edu/archive/htcondor-users/