Re: [OMPI users] Strange IO issues in MPI_Finalize

2013-03-18 Thread Jeff Squyres (jsquyres)
Brian -- While I was on a plane today, I took a whack at making OMPI behave better when you forget to MPI_File_close() a file. Can you try this patch (should apply cleanly to OMPI trunk, v1.6, or v1.7): https://svn.open-mpi.org/trac/ompi/changeset/28177 On Mar 18, 2013, at 12:42 PM, Jef

Re: [OMPI users] Strange IO issues in MPI_Finalize

2013-03-18 Thread Jeff Squyres (jsquyres)
I *believe* that this means that you didn't MPI_File_close a file. We're not giving a very helpful error message here (it's downright misleading, actually), but I'm pretty sure that this is the case. On Mar 6, 2013, at 10:28 AM, "Smith, Brian E." wrote: > HI all, > > I have some code that us

[OMPI users] Strange IO issues in MPI_Finalize

2013-03-06 Thread Smith, Brian E.
HI all, I have some code that uses parallel netCDF. I've run successfully on Titan (using the Cray MPICH derivative) and on my laptop (also running MPICH). However, when I run on one of our clusters running OMPI, the code barfs in MPI_Finalize() and doesn't write the complete/expected output fi