I’m having problems on calling “mpirun” from inside a bash script on CentOS7. When I call mpirun from the shell command line all is OK, but when I try to automate the process and put the commands inside a bash script I get an error when trying to execute “mpirun -np 32 ./wrf.exe”:
[proxy:0:0@calima-wrf] HYDU_create_process (utils/launch/launch.c:74): execvp error on file –np (No such file or directory) =================================================================================== = BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES = PID 28110 RUNNING AT calima-wrf = EXIT CODE: 255 = CLEANING UP REMAINING PROCESSES = YOU CAN IGNORE THE BELOW CLEANUP MESSAGES ===================================================================================
Previously to “mpirun” call, in the script I do a “cd /path_that_contains_wrf.exe”, like when I execute mpirun from command line.
It seems that mpirun is trying to find “-np” as executable instead of “wrf.exe” so parallel execution doesn’t start, mpirun is not treating “-np” as a parameter as should be, a very strange issue. Environment variables are setted in the same way as when I run mpirun from the shell command line (using “source set_env.sh”), which sets LD_LIBRARY_PATH, PATH, etc… variables. Any idea of what could be happening?
The content of $LD_LIBRARY_PATH environment var:
(I’m using mpich-3.1.4 and wrf3.6.1 compiled using icc and ifort compilers).
Thanks in advance,