Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

[robert_letscher@unh_edu]

I'm attempting to port CESM 2.1.3 to my local cluster and run a CIAF compset. The model builds successfully but fails at runtime with the SIGSEGV error above. I am using a GNU compiler with netcdf-c v4.8, netcdf-fortran v4.8, and openmpi v5.1.0. I've attached the cesm.log file, cesm.bldlog, pio.bldlog as well as my config_machines and config_compiler xml files. See the last entry in both xml files for my machine named "marvin". I have run with DEBUG=TRUE but still can't figure out the traceback of errors.

 

[jedwards]

Traceback shows that you are failing in the call to mpi_init. Have you run an mpi_hello_world to confirm that
mpi is working on your system? Also check your ulimits and make sure that the stack value is set to unlimited or
the maximum allowed value. If nether of these work please consult with your system administators for further suggestions.

 

[taoliu_tech]

I'm attempting to port CESM 2.1.3 to my local cluster and run a CIAF compset. The model builds successfully but fails at runtime with the SIGSEGV error above. I am using a GNU compiler with netcdf-c v4.8, netcdf-fortran v4.8, and openmpi v5.1.0. I've attached the cesm.log file, cesm.bldlog, pio.bldlog as well as my config_machines and config_compiler xml files. See the last entry in both xml files for my machine named "marvin". I have run with DEBUG=TRUE but still can't figure out the traceback of errors.

Did you solve the error? I received the same error.