Learning Forum

[JTAsia]

I have been working to try to get a FSI simulation running in our universities HPC cluster. I have the coupling.sci file, mechanical and fluent data, .py script, and submission script. I have searched through the ANSYS forums, but don't seem to find any solutions that work. nHere is the error file I am met with:nI have followed the System Coupling tutorials as well as browsed the current forum posts, but nothing has worked so far. Any help would be appreciated. Most of the other posts on this forum say the solution is to add the PartitionParticipants command, but I have tried the various ways, but I have found no luck.nThese are my submission and py scripts, respectively:nnn

[munjalpu]

Hello, You have 'Participant-1' and 'participant-2' in last argument. Replace that with Solution and Solution 1. n

[JTAsia]

It still has the same error even after replacing itn

[munjalpu]

Hi mmadore,nThanks for your reply. I can share this with IT. But I am not sure if they will help.nAnother question, can i write my own shell file instead of using the 'sh run_mpi.sh' to run the run_mpi.sh.nLike ANSYS CFX or Fluent, I can use the shell file with the following command to run the simulaiton.nmodule load cfx/19.0 nexport nodes=`awk -vORS=, '{ print $1 }' $PBS_NODEFILE | sed 's/,$/ /'`ncfx5solve -def Name.def -start-method Intel MPI Distributed Parallel -par-dist $nodes nnCould you please provide the shell file to run the ANSYS Forte?Thanks.Yang

[JTAsia]

All three result in the same error. Only difference in the error file is showing the different partitionparticipants commands around the 11th line. nThe core count also matches my core count from slurm. nAdditionally, it has to be an issue with the partitionparticipants command because if I remove the Solve() command, it still results in the same .err filenAlso here is the updated pyscript just in case I wrote it wrong. I only posted a picture of the DistrubutedAllocateCores commands, but all three methods you mentioned previously resulted in the same error file. nHere is also the output file after submitting the job. The output file is also the same across all three methods. One thing I would note is that the PrintSetup() command does not appear in the output file unless it comes before the PartitionParticipants command in the python script file.

[munjalpu]

Hello, Your python script is correct. It might be the ANSYS installation on your cluster causing issues. We have multiple version of Ansys on our cluster. I checked my setup with newer version 2020, it gave similar error. Contact your cluster administer and figure out which modules of v20 are missing or go to older v19 in the meantime if you can and check. n

[JTAsia]

We should have ANSYS 2020R2 on the cluster. As far as I know nothing is missing. We don’t have any older versions anymore. Everything was updated from ANSYS2018 to 2020 at the start of this year. Are you saying that your version of ANSYS20 is giving you the same error? Would that be an issue with ANSYS or our cluster?n

[munjalpu]

In my case, it worked on V19. I went through theory guide to see if anything is missing for v20, but could not find anything for V20. However, I found out that some modules were not setup to work with V20 on our cluster. Apart from that, the script you have works on v19.n

[munjalpu]

One more thing to note is that, specify processors in slurm while running python file. So your last line should have -t16 flag ahead of -R. nSo your syntax should be: systemcoupling -t16 -R ''$RUN_FILE >FSI.out Maybe that will get rid of issue you have. n

[JTAsia]

Sadly that has not worked eithern