At least 1 SYRTHES coupling was defined for which
no communication with a SYRTHES instance is possible.
I am using the Salome CFD distribution (version 8.0.5) and tried to make a case as simple as possible : observe the natural convection for a copper-ish cylinder in a cylindrical air domain. The copper rod is heated at the base by a set flux and the air domain use free inlet/outlet. I’ll append the relevant files.
I used the CFD_Study GUI to create the case, and got folders generated for both Saturne and Syrthes. I proceeded to use the GUIs to create and set each respective case.
I added and edited the cs_user_coupling.c file to the Saturne SRC direction
I configured the coupling in Syrthes GUI
I checked the coupling_parameters.py file
And finally I ran the case with the results that ensue…
So, yes, I was using the wrong runcase.
After trying the right runcase (the one with coupling_parameters.py for those as challenged as me) I come across another problem. Here is an excerpt :
Coupling execution between:
o Code_Saturne [1 domain(s)];
o SYRTHES [1 domain(s)];
Result directory:
/home/cosmo/Documents/Simulation/TEST/salomeCFD/couplage4/coupling/RESU_COUPLING/20200207-1806
\
Preparing calculation data
/bin/sh: 1: /usr/bin/mpicc: not found
make: *** [exe] Error 127
Apparently it is searching for mpicc in usr/bin
But I installed everything through a conda environment and the relevant binaries are in the anaconda direction. How do I make it look in the right direction ? listing.txt (3.72 KB)
I am not sure, but if you installed binaries without recompiling the code, you might have issues with hard-coded paths, and worse, issues if the MPI library used by the compiled code does not match the one used.
For syrthes, the MPI path probably is set in your case in /home/cosmo/opt/SalomeCDF8_0_5/Salome_CFD-SCFD_V8_5_0-Code_Saturne-5_2-univ_d54f597c/tools/Code_syrthes-mpa_syrthes435qt5/arch/Linux_x86_64/bin/syrthes.profile.
You can try to edit this file to replace /usr with the anaconda path.
But if the MPI library does not match the one used for build, you might still have crashes later.
Salome CFD is a self building .run file so shouldn’t it pack its own mpi binaries ?
How can I recompile them in such situation ?
I have just really been executing the run file in the already prepared environment in the hope it would pick the environment binaries/libraries
This is more tricky with MPI, because the Salome build does not include the MPI libraries, as this would also require including the compilers (MPI is source portable, not binary portable). And including the compilers is very tricky.
So if the versions are not aligned, it will fail. If things work by just changing a path, you may be lucky, but otherwise the best solution is to recompile both code_saturne and Sythes from sources (using the SALOME build as an option for prerequisites suc as MED file support).
Thank you for your help, I guess I will create a topic in the installation part of the forum because I tried installing them from source before but face problems I can’t fix so far.