PreCICE spack installation on HPC fails with python3 related error

Installing preCICE
, ,
1

Dear preCICE Community,

I am attempting to install preCICE on an HPC system using Spack, following the guidelines provided on the HPC webpage. However, the installation fails with the following error:

  >> 163    CMake Error at /ari/users/ucoskun/apps/spack/opt/spack/linux-rhel8-
            zen2/gcc-12.3.0/cmake-3.30.5-s6eovwbmevqy55yyon3pp6q77nlodpg5/share
            /cmake-3.30/Modules/FindPackageHandleStandardArgs.cmake:233 (messag
            e):
     164      Could NOT find Python3 (missing: Interpreter)
     165    
     166          Reason given by package:
     167              Interpreter: Cannot run the interpreter "/ari/progs/Pytho
            n/Python-3.12.4-openmpi-5.0.3-gcc-11.4.0/bin/python3"
     168    
     169    Call Stack (most recent call first):

Here is a summary of the steps I have taken:

  • I loaded a Python module (seemed most recent version) using:
module load Python/Python-3.12.4-openmpi-5.0.3-gcc-11.4.0
  • Installed spack to my home directory without any error.
  • Checked Python versions:
python3 -V:
Python 3.12.4
python -V:
-bash: python: command not found
  • Spack find compilers and spack compilers commands resulted as below:
spack compilers
==> Available compilers
-- gcc rhel8-x86_64 ---------------------------------------------
gcc@8.5.0  gcc@12.3.0  gcc@11.4.
  • Checked the precice with spack info precice
    The preferred version was 3.1.2, and the variants section appeared appropriate (probably not though).
  • I ran spack install precice

Despite these steps, the installation fails with the aforementioned Python3 interpreter error. I would appreciate any suggestions or guidance to resolve this issue. Please note that I am relatively new to this process and may need detailed instructions.

2

Hi,

Try detecting the python installation as an external package in spack.

spack external find python

You should then see it as an installed package using

spack find python

I normally use this for python and a row of standard tools such as make, cmake, pkg-config etc.

Kind regards
Frédéric

3

Dear Frédéric,

Thank you for your quick reply!

I’ve successfully (magically) installed “everything” (please pardon my enthusiasm). I managed to run the heat-exchanger tutorial on the HPC without errors—however, there’s a catch!

After copying your response to ChatGPT and discussing it for a while, this was the solution for me:

  • Avoid using any previously installed modules from the HPC where possible (including Python).
  • Use Spack (where possible) and ensure all dependencies are built with the same GCC version.
  • The most challenging part was adjusting the Makefile for the CalculiX preCICE adapter, but the rest was fairly straightforward.

I ran the heat exchanger tutorial using a SLURM script, assigning 50 cores to each fluid participant and 28 cores to the solid participant, making a total of 128 cores. (Just to check the speed-up)

However:

I realized that I had forgotten to add the -parallel option to the run.sh scripts for OpenFOAM. Naturally, I re-ran the simulation with -parallel, but this time, each time step took significantly longer—about 30 times longer than the serial run.

Therefore, I wanted to ask the community:

  • Is this the correct approach for running a coupled simulation on an HPC using SLURM?
  • Could you (or anyone) kindly review my SLURM script below and share your thoughts?

Kind regards,
Umut

SLURM script:

#!/bin/bash
#SBATCH -A account
#SBATCH -n 128
#SBATCH -p queue
# ------------------------------
#module load ek-moduller-easybuild
#module load arpack-ng/3.9.0-foss-2023a
#spack load openmpi@5.0.6 netlib-lapack@3.12.1 openblas@0.3.29 yaml-cpp@0.8.0 openfoam@2312 precice@3.1.2 petsc@3.22.3
#export PATH=$HOME/apps/calculix-adapter-master/bin:$PATHls
#export LD_LIBRARY_PATH=$HOME/apps/spack/opt/spack/linux-rhel8-zen2/gcc-12.3.0/precice-3.1.2-ukxuqmx2goykhc5c4tyw3huhawqxokwo/lib64:$LD_LIBRARY_PATH
#export LD_LIBRARY_PATH=$HOME/apps/spack/opt/spack/linux-rhel8-zen2/gcc-12.3.0/yaml-cpp-0.8.0-fypxvjn4bec57zb4rq3pb4aqp62vlog7/lib64:$LD_LIBRARY_PATH
#export LD_LIBRARY_PATH=$HOME/apps/spack/opt/spack/linux-rhel8-zen2/gcc-12.3.0/netlib-lapack-3.12.1-6wv46i4ij6ilsumqpyw23hmrwpwi7b5q/lib64:$LD_LIBRARY_PATH
#export LD_LIBRARY_PATH=$HOME/apps/spack/opt/spack/linux-rhel8-zen2/gcc-12.3.0/petsc-3.22.3-f6fi4zethcvvsqnq6wv5fwxprhsphju7/lib:$LD_LIBRARY_PATH
# ------------------------------

caseFolder=$(pwd)
$caseFolder/clean-tutorial.sh > clean-tutorial.log

cd fluid-inner-openfoam
$caseFolder/fluid-inner-openfoam/run.sh -parallel &
cd ..

cd fluid-outer-openfoam
$caseFolder/fluid-outer-openfoam/run.sh -parallel &
cd ..

cd solid-calculix
$caseFolder/solid-calculix/run.sh &
cd ..

wait

I also updated two lines in solid-calculix/run.sh as below:

export OMP_NUM_THREADS=28
export CCX_NPROC_EQUATION_SOLVER=28
4

3 posts were split to a new topic: Running CCX and OF on HPC

5

I split the topic after the python problem disappeared.
I’ll mark this on as solved.