[CP2K-user] [CP2K:14777] CP2K 8.1: Runtime error running on more than one node

[Tiziano Müller]

Hi Mauro,

there are some things you can try:

* build a CP2K without COSMA, ELPA, SIRIUS, SpFFT and see whether you 
can reproduce it with that (simply create an arch-file as a copy of the 
toolchain-generated ones with the respective -D..., -l..., -I... flags 
removed)
* check the toolchain build logs of ELPA and COSMA (cmake.log/make.log) 
to verify that the correct MPI implementation was picked up during 
compilation, see also [1]

My guess since other (less complex) software works is that either there 
is/was a mismatch of OpenMPI implementations and/or that one of the 
dependencies tries to do some advanced MPI ops (RMA, I/O) which is not 
configured on your cluster (CP2K has RMA off by default).

Best,
Tiziano

[1] https://github.com/cp2k/cp2k/issues/1351

On 2/15/21 10:07 PM, Mauro Sgroi wrote:
> Dear Alin,
> thanks a lot for the reply.
> With those mpi we are running smoothly Quantum Espresso and other codes.
> So it seems strange to find this error.
> Best regards,
> Mauro Sgroi.
> Il giorno lunedì 15 febbraio 2021 alle 18:43:31 UTC+1 al... at gmail.com 
> ha scritto:
> 
>     Hi Mauro,
> 
>     seen this before with other codes... the issue invariable ended up due
>     to openmpi not playing nicely with the queuing system.
> 
>     if you want to see if a simple MPI hello world using the above mpi
>     actually works....
> 
>     Regards,
>     Alin
>     Without Questions there are no Answers!
>     ______________________________________________________________________
>     Dr. Alin Marin ELENA
>     http://alin.elena.space/ <http://alin.elena.space/>
>     ______________________________________________________________________
> 
>     On Mon, 15 Feb 2021 at 17:25, Mauro Sgroi
>     <mauro... at gmail.com> wrote:
>      >
>      > Dear Developers,
>      > I'm writing to ask you help for setting CP2K on a HPC server.
>      > Our IT compiled the code using gcc-8.3.0, openmpi 4.1.0 and the
>     toolchain:
>      >
>      > cmake-3.18.5 cosma-2.2.0 elpa-2020.05.001 fftw-3.3.8 gsl-2.6
>     hdf5-1.12.0 libint-v2.6.0-cp2k-lmax-5 libvdwxc-0.4.0 libvori-201229
>     libxc-4.3.4 libxsmm-1.16.1 openblas-0.3.10 scalapack-2.1.0
>     sirius-7.0.0 SpFFT-0.9.13 spglib-1.16.0
>      >
>      > The obtained code works fine on a single node of our server. When
>     we launch it with the command:
>      > mpiexec --prefix ${MPI_INST_DIR} -x PATH -x LD_LIBRARY_PATH -n
>     $CORES -machinefile machine_file
>     ${INST_DIR}/cp2k-8.1/exe/local/cp2k.${SOLVER} -inp ${INPUT} 2>&1
>      >
>      > on more than one node we get the error:
>      >
>      > [btl_openib_component.c:1699:init_one_device] error obtaining
>     device attributes for mlx5_0 errno says Protocol not supported
>      >
>      > Could you please give us some advice?
>      >
>      > Thanks a lot in advance and best regards,
>      >
>      > Mauro Sgroi.
>      > Centro Ricerche FIAT.
>      > Italy.
>      >
>      > --
>      > 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 cp... at googlegroups.com.
>      > To view this discussion on the web visit
>     https://groups.google.com/d/msgid/cp2k/cc9bb9af-4e72-4667-ad33-ef04edf7a70bn%40googlegroups.com
>     <https://groups.google.com/d/msgid/cp2k/cc9bb9af-4e72-4667-ad33-ef04edf7a70bn%40googlegroups.com>.
> 
> 
> -- 
> 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 cp... at googlegroups.com 
> <mailto:cp... at googlegroups.com>.
> To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cp2k/7b611401-f960-464c-9394-0b3add2aad6dn%40googlegroups.com 
> <https://groups.google.com/d/msgid/cp2k/7b611401-f960-464c-9394-0b3add2aad6dn%40googlegroups.com?utm_medium=email&utm_source=footer>.

-- 
Tiziano Müller
University of Zurich
Department of Chemistry
Winterthurerstrasse 190
CH-8057 Zürich

Tel: +41 44 63 54234
www.chem.uzh.ch
tiziano... at chem.uzh.ch