[mag10100]

I compiled SU2-7.3.1 with following configuring options on our institute supercomputer to an FSI example. The supercomputer runs on Linux OS based on CentOS 7.x distribution.

-Dcustom-mpi=true -Denable-mkl=true -Dmkl_root=/home-ext/apps/spack/opt/spack/linux-centos7-cascadelake/intel-19.0.5.281/intel-mkl2020.4.304-fet6h2j2qeq5alsxjiw7fzjkweqorbjf/mkl/ -Denable-pywrapper=true




SU2-7.3.1 is compiled with following packages.
gcc 11.2.0
mpi4py 3.0.3
openmpi 4.1.2
python 3.9.10
swig 4.0.2

The compilation was done successfully.

Now, I am checking the code by running the following FSI example on the cluster.
https://github.com/su2code/Tutorials...si_python/Ma01


I checked this example in serial and parallel. I am getting following error.
ValueError: mpi4py.MPI.Status size changed, may indicate binary incompatibility. Expected 48 from C header, got 40 from PyObject


This will be a great help. I have been struggling to run this example on the cluster for so long. Any help will work for me.
Please help and suggest.


Thank you

[bigfootedrockmidget]

I think this is a mismatch between mpi and mpi4py. Does a regular mpi job without the python wrapper work (it runs correctly and scales with nr of cores)? Do you have different mpi4py versions available (maybe through modules)? If so, you could try with a different version.

[mag10100]

Does a regular mpi job without the python wrapper work (it runs correctly and scales with nr of cores)?
Yes, I was able to run only a CFD problem without FSI.


Do you have different mpi4py versions available (maybe through modules)? If so, you could try with a different version.[/QUOTE]
Yes, I installed SU2-7.5.1 with the following versions. Now, fsi is working.
gcc v12.2.0
mpi4py v3.1.4
python v3.10.8
openmpi v4.1.5
swig v4.1.1
numpy v1.24.0
scipy v1.5.4
rtree v1.0.1
petsc4py v3.16.0



But I am unable to restart the following example after 500 steps. It runs fine when I start it from scratch.
https://github.com/su2code/Tutorials...si_python/Ma01
It produces the following error.
-------------------------
File "/scratch/asemagan/exctbls/su2/su2-751_gcc1220py3108ompi415/bin/SU2_Nastran/pysu2_nastran.py", line 784, in __setRestart
if int(line[1])==(self.Config["RESTART_ITER"]-2):
KeyError: 'RESTART_ITER'
------------------------------------

[bigfootedrockmidget]

looks like it did not find the keyword restart_iter:

% Iteration number to begin unsteady restarts
RESTART_ITER= 0

[mag10100]

To restart FSI example, we need fluid.cfg, solid.cfg, fsi.cfg cofig. files with mesh and two restart files (for second-order dual-time stepping). What I observed to resolve this error, provided that transient steps for fluid are completed, means "TIME_TRESHOLD" steps have already been done.

Suppose, your flow restart files are as follow.
restart_flow_999.dat
restart_flow_1000.dat

Then you set following in fsi.cfg.
TIME_TRESHOLD= -1
RESTART_ITER= 1001
RESTART_SOL= YES

Then you set following in fluid.cfg.
RESTART_ITER= 1001
RESTART_SOL= YES

Then you set following in solid.cfg.
RESTART_SOL= NO

This is how the problem is resolved for me.