open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello everyone，

I trying to install Telemac on linux cluster which use qusb command to submit the job. Now I am confusing about which .cfg file that I should use to compile Telemac The .cfg file that I used post below, the question is I can just run for one node. If I use more nodes in parallel command there will be an error as the error.txt below. It make me crazy...

Apologize for my poor English.

Any help will be grateful.

Hi
The problem seems related to a problem of rights in the execution directory.
the error shows that Telemac has a problem to create the file named CONFIG or access it...
Hope this helps

Thank you c.coulet.
I tried again.
when I use:
mpiexec -x LD_LIBRARY_PATH -np $NP -machinefile $PBS_NODEFILE --mca btl self,sm,openib runcode.py telemac2d --mpi MYJ_test.cas --ncnode=2 --ncsize=128 -c gofortranHPC
in pbs file to submit the job.There seems no problem related to CONFIG.
But another probblem I don't know how to describe, please see the attachments files.

You should not run mpiexec on python script...
just run telemac.py and the python script will automatically launch the mpiexec command at the run step.

Dear c.coulet, Thank you again!
I remove the mpiexec command and just left:
telemac2d.py MYJ_test.cas --ncnode=2 --ncsize=128 -c gofortranHPC
it is strange, seems I just run on one node with 128 cores. (there are only 64 cores per node). And the speed is slower than with --ncnode=1 --ncsize=64.

Here is the running log.

In you report, we could see 2 nodes listed, but only one in mpi_hostfile!

You probably should adapt some points.
This depends on your local architecture and scheduler ...
Look at the HPC config to be sure to well distribute the computation.

Sorry c.coulet.. Could you tell me where can I find the HPC config.

Look in the edf.cfg...

I have found it. So do I need to find a cfg that fits my machine cluster to recompile?