[slurm-users] EXTERNAL: Re: Memory per CPU

[Thomas M. Payerle]

I am not familiar with using Slurm with VMs, but do note that Slurm can
behave a bit "unexpectedly" with memory constraints due to the memory
consumed by OS, etc.
E.g., if I had a 16 core machine with 64 GB of RAM and requested 16 cores
with 4 GB/core, it would not fit on this machine because some of the RAM is
being used by the OS.
Whereas if I requested 3.75 GB/core it likely would fit.

As Michael Di Domenico, can you detail how you are concluding that 2 CPU
cores are assigned to the 16 GB job?  E.g. if your VM nodes are 2 core and
32 GB, then I would expect placing a 1 core 16 GB job will prevent another
such job being scheduled to that node.  I would suggest reducing your
default/max RAM per core a little bit so you have a GB or two left over for
the system, and things should work more as expected.

On Tue, Sep 29, 2020 at 11:22 AM Luecht, Jeff A <jeff.luecht at pnc.com> wrote:

> There are three pieces of information that may provide Useful:
>
> 1 - these are VMs and not physical servers
> 2 - the OS is RedHat 7.8
> 2 - As far as I can tell, hyperthreading is not enabled, but will check
> for sure
> 3 - when we ask for 15Gb memory - we will only get 1 CPU
>
> -----Original Message-----
> From: slurm-users [mailto:slurm-users-bounces at lists.schedmd.com] On
> Behalf Of Michael Di Domenico
> Sent: Tuesday, September 29, 2020 10:20 AM
> To: Slurm User Community List <slurm-users at lists.schedmd.com>
> Subject: EXTERNAL: Re: [slurm-users] Memory per CPU
>
> ** This email has been received from outside the organization – Think
> before clicking on links, opening attachments, or responding. **
>
> what leads you to believe that you're getting 2 CPU's instead of 1?
> 'scontrol show job <id>' would be a helpful first start.
>
> On Tue, Sep 29, 2020 at 9:56 AM Luecht, Jeff A <jeff.luecht at pnc.com>
> wrote:
> >
> > I am working on my first ever SLURM cluster build for use as a resource
> manager in a JupyterHub Development environment.  I have configured the
> cluster for SelectType of ‘select/con_res’ with DefMemPerCPU and
> MaxMemPerCPU of 16Gb.  The idea is to essentially provide for jobs that run
> in a 1 CPU/16Gb chunks.  This is a starting point for us.
> >
> >
> >
> > What I am seeing is that when users submit jobs and ask for memory only
> – in this case, 16Gb, SLURM actually allocates 2 CPUs, not 1 that I would
> expect.  Is my understanding of how this particular configuration works
> incorrect?
> >
> >
> > The contents of this email are the property of PNC. If it was not
> addressed to you, you have no legal right to read it. If you think you
> received it in error, please notify the sender. Do not forward or copy
> without permission of the sender. This message may be considered a
> commercial electronic message under Canadian law or this message may
> contain an advertisement of a product or service and thus may constitute a
> commercial electronic mail message under US law. You may unsubscribe at any
> time from receiving commercial electronic messages from PNC at
> http://pages.e.pnc.com/globalunsub/
> > PNC, 249 Fifth Avenue, Pittsburgh, PA 15222; pnc.com
> >
>
>
>
>
> The contents of this email are the property of PNC. If it was not
> addressed to you, you have no legal right to read it. If you think you
> received it in error, please notify the sender. Do not forward or copy
> without permission of the sender. This message may be considered a
> commercial electronic message under Canadian law or this message may
> contain an advertisement of a product or service and thus may constitute a
> commercial electronic mail message under US law. You may unsubscribe at any
> time from receiving commercial electronic messages from PNC at
> http://pages.e.pnc.com/globalunsub/
> PNC, 249 Fifth Avenue, Pittsburgh, PA 15222; pnc.com
>
>
>

-- 
Tom Payerle
DIT-ACIGS/Mid-Atlantic Crossroads        payerle at umd.edu
5825 University Research Park               (301) 405-6135
University of Maryland
College Park, MD 20740-3831
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.schedmd.com/pipermail/slurm-users/attachments/20200929/d6e94b49/attachment.htm>