Run a python script on multiple nodes

[thomas-saigre]

I have a python script made to run on 96 cores, but when I run it on a single node, it takes too much memory and fails.
I tried the following slurm script to split it on 2 nodes :

#!/bin/bash
#SBATCH --job-name=get-output
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=48
#SBATCH --time=1:00:00
#SBATCH --output=log-output%j.log
#SBATCH --error=log-error%j.log

mpirun -np 96 python output.py

but in the execution, at the step on loading the mesh (already partitioned over 96 cores), it fails :

Traceback (most recent call last):
  File "/data/home/.../output.py", line 127, in <module>
    run("Eye-model/eye.cfg", "outputCOMP.csv")
  File "/data/home/.../output.py", line 38, in run
    mesh = feelpp.load(m, path, 0.1)
RuntimeError: MPI_Test: MPI_ERR_TRUNCATE: message truncated

is it an issue of parametrization of slurm ? Or should I also update the python code ?