Issues with MPI while running an Earth System Model

65 Views Asked by At

My question here is related the post below Linking libraries

As mentioned in the above question, I have been trying to work on an earth system model (CESM) and I have consulted with their forum as well. But since the question above was not exactly a model-specific one, I was having trouble fixing it. I was finally able to port it to my local system using the answers to the question above. Another link associated with the above post is Porting cesm

The case that I tried to build in the previous question was a very simple and computationally less extensive one. Even the run was successful. So, I moved to other detailed cases. Note: Case means a component-set or compset that I want to run for my model. It is an earth system model and it has several components which are coupled and can be run in prognostic mode, data mode and stub mode. So different combinations of these components are compsets and some have been defined and tested by the model developers.

Now I am trying to run such a case which is a tested one and is available as an example from one of the official tutorials (The case name is I1850CLM50_004 and compset is called I1850Clm50Sp). The build was successful but the run is failing and it is pointing to an error regarding mpiexec

I had to run the following (Invoking the following 4 scripts are the common steps to be done for any model configuration)

./create_newcase --case ~/clm_tutorial_cases/I1850CLM50_004 --res f19_g17 --compset I1850Clm50Sp --machine ubuntuvmfed
./case.setup
./case.build
./case.submit

On the last step (i.e after submitting the case), I am getting the following in the log


mpiexec does not support recursive calls



mpiexec does not support recursive calls

**********************************************************
[ubuntuvmfed:15606] PMIX ERROR: ERROR in file ../../../../../../src/mca/common/dstore/dstore_base.c at line 1709
[ubuntuvmfed:15610] PMIX ERROR: ERROR in file ../../../../../../src/mca/common/dstore/dstore_base.c at line 1709
[ubuntuvmfed:15606] PMIX ERROR: ERROR in file ../../../../../../src/mca/common/dstore/dstore_base.c at line 1709
[ubuntuvmfed:15610] PMIX ERROR: ERROR in file ../../../../../../src/mca/common/dstore/dstore_base.c at line 1709

Could this be due to an error with my installation of MPI library or does this mean I have to run this in an HPC cluster only? Any suggestions would be appreciated.

0

There are 0 best solutions below