Avoid initializing MPI in regression tests

[AndrewGaspar]

Addresses an issue with hygiene when importing the h5py module.

Prevent mpi4py from initializing MPI.

Fixes #312

PR Summary

Importing h5py causes future MPI operations in the process or child process to fail (probably due to unpaired MPI_Init) for some implementations of MPI, such as openmpi 4.0.2 on Darwin Power9. Therefore, this change runs restart.py's TestCase.Analyse script in a child process to maintain hygiene in the test process. This change stops the mpi4py transitive dependency from initializing MPI when present.

A possible alternative solution would be to import h5py in the scope of the Analyse function. My concern with this is that future mpiexec commands would fail, and I don't know if we're guaranteed to never have any more. This solution is more complete, but less pretty.

PR Checklist

• Code passes cpplint
• New features are documented.
• Adds a test for any bugs fixed. Adds tests for new features.
• Code is formatted
• Changes are summarized in CHANGELOG.md (N/A - test fix)

[pgrete]

I'm not sure I ran across the same problem, but in a different context I "fixed" an issue by using the following initialization order:

import mpi4py
mpi4py.rc(initialize=False)
from mpi4py import MPI
import h5py
MPI.Init()

Would you mind testing that approach and see if this fixes it? (I don't have access to the system where it's nor working).

[AndrewGaspar]

@pgrete I'm glad somebody else ran into this before. :) Let me see... the thing is, the MPI initialization seemed to be coming out of the loading of a native library in h5py, so I'm worried mpi4py isn't involved here...

[AndrewGaspar]

That did work! I'll promote it to run_test.py.

I didn't seem to actually need to call MPI.Init, so I'm going to leave it out for now... It's possible we don't actually need to build h5py with mpi support? Probably shouldn't break if it was, though.

[Yurlungur]

That did work! I'll promote it to run_test.py.

I didn't seem to actually need to call MPI.Init, so I'm going to leave it out for now... It's possible we don't actually need to build h5py with mpi support? Probably shouldn't break if it was, though.

Just wrap the MPI import in a try-catch block, in case MPI4Py is not available.

I was going to suggest moving the script in the string into it's own file in scripts/python, since this comparison of two dump files is probably a fairly generic operation. We might as well genericize it and pull it out of the testing infrastructure. But looks like that's not needed.

[JoshuaSBrown]

@AndrewGaspar Thanks so much for fixing this, @pgrete thanks for the suggestion. This is great! And @Yurlungur thanks for your suggestions!

[Yurlungur]

Looks good. 👍

Thanks for digging into this.

[AndrewGaspar]

@pgrete I assume you're good with this change then?

[AndrewGaspar]

Phil signed off on the call.