Molcas Forum

Support and discussions for Molcas and OpenMolcas users and developers

You are not logged in.

Announcement

Welcome to the Molcas forum.

Please note: The forum's URL has changed. The new URL is: https://molcasforum.univie.ac.at. Please update your bookmarks!

You can choose an avatar and change the default style by going to "Profile" → "Personality" or "Display".

#1 2022-09-15 00:36:38

exm114514
Member
Registered: 2022-09-08
Posts: 2

Job gets killed randomly

Dear developers and users,

I just installed OpenMolcas on my machine, using Ubuntu 20.04 on WSL2. Most of the prerequisites are installed by the package manager, while the OpenBLAS and GA are compiled exactly as on page 18 of the manual.

# OpenBLAS
tar zxvf OpenBLAS-v0.2.15.tar.gz
cd OpenBLAS-0.2.15/
make USE_OPENMP=1 NO_LAPACK=0 INTERFACE64=1 BINARY=64 DYNAMIC_ARCH=1 \
libs netlib shared
[sudo] make PREFIX=/opt/openblas-lapack-ilp64 install
# GA
tar zxvf /path/to/ga-5-4b.tgz
cd ga-5-4b/
./configure --enable-i8 --with-blas8 --with-lapack8 --with-scalapack8 \
--prefix=/opt/ga54b-ilp64.OpenMPI
make
[sudo] make install
# Molcas
tar zxvf molcas.tgz
cd molcas
mkdir build && cd build/
export GA=/opt/ga54b-ilp64.OpenMPI
export OPENBLASROOT=/opt/openblas-lapack-ilp64
CC=mpicc FC=mpifort cmake -DMPI=ON -DGA=ON -DLINALG=OpenBLAS ../
make

Running single-point calculations seems to be fine, but when I was trying to do an MEP job, it was fine at the beginning. After a while, the .log file, or anything in the directory was no more updated. Like, the output file suddenly stopped being updated at a random point, without any notice in the .log or .err file. In the command line, the OpenMolcas seems to be running but nothing was updated, after some point. Before I installed openmolcas on my computer, I was running openmolcas on clusters, and exactly the same issue happens. Here's the output, configuration info is exactly the same as when I was running openmolcas on clusters, except for the host name.

           *************************************************
           * pymolcas version py2.23                       *
           *   build d53e6f8e18683222a6e02cb84ce98eb3      *
           *   (after the EMIL interpreter by V. Veryazov) *
           *************************************************

configuration info
------------------
Host name: DESKTOP-QM4JF7M (Linux)
C Compiler ID: GNU
C flags: -std=gnu99
Fortran Compiler ID: GNU
Fortran flags: -fno-aggressive-loop-optimizations -cpp -fdefault-integer-8 -fmax-stack-var-size=1048576
Definitions: _MOLCAS_;_I8_;_LINUX_;_MOLCAS_MPP_;_GA_
Parallel: on (GA=on)


   ----------------------------------------------------------------
  |
  |              Project: 45_1008_mep_s1fc_ccdz
  |       Submitted from: /home/neko/projects/45/mep/fc_1008_ccdz
  |         Scratch area: /home/neko/scratch/45_1008_mep_s1fc_ccdz
  |      Save outputs to: /home/neko/projects/45/mep/fc_1008_ccdz
  |               Molcas: /home/neko/OpenMolcas
  |
  | Scratch area is NOT empty
  |
  |       MOLCAS_DRIVER = /home/neko/OpenMolcas/pymolcas
  | MOLCAS_KEEP_WORKDIR = YES
  |  MOLCAS_NEW_WORKDIR = NO
  |       MOLCAS_NPROCS = 8
  |        MOLCAS_PRINT = 3
  |       MOLCAS_SOURCE = /home/neko/OpenMolcas
  |    MOLCAS_STRUCTURE = 0
  |
   ----------------------------------------------------------------

++ ---------   Input file   ---------

&GATEWAY
        COORD=  fc.xyz
        BASIS=  cc-pvdz
        GROUP=  c1
        ACCD BASIS
        RICD

>>> EXPORT MOLCAS_MAXITER = 200

>>> DO WHILE

  &SEWARD

  >>> IF (ITER == 1)

    &RASSCF
            FILEORB=    fc_1008.RasOrb
            SPIN=           1
            NActEl=         10 0 0
            Inactive=       24
            RAS2=           8
            CIRoot=         4 4 1

  >>> END IF

  &RASSCF
          SPIN=           1
          NActEl=         10 0 0
          Inactive=       24
          RAS2=           8
          CIRoot=         4 4 1
          RlxRoot=    2

  &ALASKA

  &SLAPAF
          CARTESIAN
          MEP-search
          NMEP =          50
          MEPStep=        0.02

>>> END DO

-- ----------------------------------

--- Start Module: gateway at Wed Sep 14 19:18:49 2022 ---

I was wondering if it dues to the I/O issue of the scratch, since scratches are duplicated for each of the mpi processes, but I'm not very sure about it. Also, I see the fno-aggressive  flag when compiling the GA, it is the default and I have no idea how to turn it off, but could that be a problem?

Any advice would be greatly appreciated.

Last edited by exm114514 (2022-09-15 00:51:26)

Offline

#2 2022-09-15 09:57:58

Ignacio
Administrator
From: Uppsala
Registered: 2015-11-03
Posts: 1,003

Re: Job gets killed randomly

You could try to see if there's something relevant in the log files of the slave processes. They should be in the "tmp_X" directories inside the WorkDir, named StdOut or something similar.

And running in serial is probably more stable.

Offline

#3 2022-09-20 03:05:18

exm114514
Member
Registered: 2022-09-08
Posts: 2

Re: Job gets killed randomly

Like everything else, the StdOut suddenly stopped being updated and stuck in the middle of the job. But as you said, running in serial is more stable so I'm going to stay at that level for this moment. Thanks for the advice!

Offline

Board footer

Powered by FluxBB 1.5.11

Last refresh: Today 19:37:50