[CP2K-user] [CP2K:19393] Help with running cp2k in parallel with Slurm

[Ava Rajh]

Hi,

I tested the provided docker. When running in the container it again works
great, outside the container  starting the job with mpirun -n works and
sets the number of processes, but starting with srun (either through bash
file or inline) produces the following error:

--------------------------------------------------------------------------
The application appears to have been direct launched using "srun",
but OMPI was not built with SLURM's PMI support and therefore cannot
execute. There are several options for building PMI support under
SLURM, depending upon the SLURM version you are using:

  version 16.05 or later: you can use SLURM's PMIx support. This
  requires that you configure and build SLURM --with-pmix.

  Versions earlier than 16.05: you must use either SLURM's PMI-1 or
  PMI-2 support. SLURM builds PMI-1 by default, or you can manually
  install PMI-2. You must then build Open MPI using --with-pmi pointing
  to the SLURM PMI library location.

Please configure as appropriate and try again.
--------------------------------------------------------------------------

So my guess would be that I should configure my version of OpenMPI to work
with slurm?

thanks for all the help, Ava

V V čet., 19. okt. 2023 ob 12:46 je oseba Krack Matthias <
matthias.krack at psi.ch> napisala:

> Hi Ava
>
>
>
> Thanks for testing. Indeed, the .sif file does not seem to work with the
> host MPI properly. These .sif files were directly built with apptainer.
> This was a test and we will not provide such containers in the future any
> longer. Instead, I suggest the download of a cp2k production docker
> container. Could you test such a docker container with apptainer? You can
> download it with the command
>
> apptainer pull docker://mkrack/cp2k:2023.2_mpich_generic_psmp
>
> or
>
> apptainer pull docker://mkrack/cp2k:2023.2_openmpi_generic_psmp
>
> depending on your host MPI version.
>
> A list of the available docker containers can be found here
> <https://hub.docker.com/repository/docker/mkrack/cp2k/tags?page=1&ordering=last_updated>
> .
>
>
>
> Best
>
>
>
> Matthias
>
>
>
> *From: *cp2k at googlegroups.com <cp2k at googlegroups.com> on behalf of Ava
> Rajh <ava.rajh at gmail.com>
> *Date: *Thursday, 19 October 2023 at 12:17
> *To: *cp2k <cp2k at googlegroups.com>
> *Subject: *Re: [CP2K:19391] Help with running cp2k in parallel with Slurm
>
>
>
> Sorry for the confusion, I read the output wrong,
>
> and if I put the ntasks in the header of the batch file I get the same
> result as running just with srun, so the number of tasks determines the
> number of programs running with one MPI rank each.
>
>
>
> On Thursday, October 19, 2023 at 10:20:39 AM UTC+2 Ava Rajh wrote:
>
>
>
> Hi Matthias,
>
>
>
> Thank you for the reply
>
>
>
> I tried both, first just with the srun command and as part of the #SBATCH
> file.
>
> If I run the "srun" command and specify the number of tasks I get the
> before mentioned multiple copies of the same program with 1 MPI rank each.
> The same happens if I run the command outside the container without the
> srun and specify the number of tasks with just: mpiexec -n 4
>
>
>
> and if I put the ntasks in the header of the batch file, then I just get
> one copy of the program with 1 MPI rank, no matter how many -ntasks are
> defined.
>
>
>
> Any idea about how to start my calculation would be very helpful, thank
> you.
>
> kind regards, Ava
>
>
>
>
>
> On Wednesday, October 18, 2023 at 7:21:22 PM UTC+2 Krack Matthias wrote:
>
> Hi Ava
>
>
>
> Do you run the “srun” command as part of a SLURM batch job file with a
> #SBATCH header section or interactively?
>
> Your guess is right, the –ntasks flag defines the number of MPI ranks.
>
>
>
> Best
>
>
>
> Matthias
>
>
>
> *From: *cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Ava
> Rajh <ava.... at gmail.com>
> *Date: *Wednesday, 18 October 2023 at 14:17
> *To: *cp2k <cp... at googlegroups.com>
> *Subject: *[CP2K:19379] Help with running cp2k in parallel with Slurm
>
> Dear all,
>
> I am trying to run cp2k on our HPC cluster and I am new in doing any kind
> of parallel computing and work on a cluster, so I would appreciate if some
> help and I apologize if I am missing something obvious.
>
> I am trying to use Cp2k in combination with apptainer  and I followed
> instructions at https://github.com/cp2k/cp2k/tree/master/tools/apptainer
>
>
>
> I have a .sif file in my work directory and if I work within it (Running
> MPI within the container), everything works perfectly and I can set the
> number of MPI threads.
>
>
>
> But when trying to run it through slurm, I can't seem to be able to set
> the number of MPI processes per node. If i start a command like:
>
> srun --ntasks=2 apptainer run -B $PWD cp2k-2023.2_mpich_generic_psmp.sif
> cp2k -i H2O-32.in
>
> it just starts 2 instances of the program that run at the same time, and
> for each, the total number of message passing processes is 1. I am able to
> set and change the number of OpenMP threads though.
>
> So my question would be first, am I wrongly assuming that --ntasks X
> should correspond to the number of MPI threads? And if I am, how would I
> set it.
>
>
>
> Please let me know if I need to provide any more information to diagnose
> the issue.
>
> Thank you very much for the help and kind regards,
>
> Ava Rajh
>
> --
> You received this message because you are subscribed to the Google Groups
> "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to cp2k+uns... at googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/cp2k/2393b350-7de0-4599-97e0-4f52ffec0c5en%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/2393b350-7de0-4599-97e0-4f52ffec0c5en%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>
> --
> You received this message because you are subscribed to the Google Groups
> "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to cp2k+unsubscribe at googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/cp2k/1a89bd22-860c-4922-a2b6-bd20232ca770n%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/1a89bd22-860c-4922-a2b6-bd20232ca770n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>
> --
> You received this message because you are subscribed to a topic in the
> Google Groups "cp2k" group.
> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/cp2k/or6vb6En6aU/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> cp2k+unsubscribe at googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/cp2k/ZRAP278MB0827D6DA09823659200590A5F4D4A%40ZRAP278MB0827.CHEP278.PROD.OUTLOOK.COM
> <https://groups.google.com/d/msgid/cp2k/ZRAP278MB0827D6DA09823659200590A5F4D4A%40ZRAP278MB0827.CHEP278.PROD.OUTLOOK.COM?utm_medium=email&utm_source=footer>
> .
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/CADBW8Us_6gVF8ABCEaUF%2B9VpK_84T8%3DNcU%2BLONq491L9DkwNVw%40mail.gmail.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20231019/0211c37e/attachment-0001.htm>