Yes, you need to set the following:

SelectType=select/cons_res

to ensure that nodes can be shared amongst different jobs and then
choose one of the consumable resources types e.g.

SelectTypeParameters=CR_Core


Cheers,
Adam

On Fri, Apr 29, 2016 at 6:36 AM, Lachlan Musicman <[email protected]> wrote:
> I'm finding this a little confusing.
>
> We have a very simple script we are using to test/train staff how to use
> SLURM (16.05-pre2). They are moving from an old Torque/Maui system.
>
> I have a test partition set up,
>
> from slurm.conf
>
> NodeName=slurm-[01-02] CPUs=8 RealMemory=32000 Sockets=1 CoresPerSocket=4
> ThreadsPerCore=2 State=UNKNOWN
> PartitionName=debug Nodes=slurm-[01-02] Default=YES MaxTime=48:0:0
> DefaultTime=0:40:0 State=UP
>
> when we run:
>
> sbatch test.sh
>
> cat test.sh
> #!/bin/sh
> x=1; while [[ $x -eq 1 ]]; do echo $x; done
>
> It takes an entire node. If we run it >2 times, a queue forms until we kill
> the jobs. I would expect that each execution of test.sh would only require a
> single CPU on each node, so I could ostensibly run 16 of these without a
> queue forming?
>
> We tried "sbatch -c 1 test.sh" but this showed the same behaviour.
>
> What are we doing wrong? When I read on sbatch, it seems that -N, -n and -c
> are the three switches in particular to look at, but I am not seeing what I
> would expect.
>
> cheers
> L.
>
>
>
>
> ------
> The most dangerous phrase in the language is, "We've always done it this
> way."
>
> - Grace Hopper

Reply via email to