[MOAB-dev] Issues with mbparallelcomm_test in the 'bcast' mode on cosmea

[Dmitry Karpeev]

Running mbparallelcomm_test -3 3 1
~karpeev/fathom/moab/data/tjunc6/tjunc6RIB_16.h5m PARALLEL_PARTITION
fails for me with 4, 8, 12, 16 procs on cosmea.  The data dir should
be group readable for 'collab' and also contains output in
bcast_delete, read_delete, read_part and bcast subdirs.

Reading the same mesh works with 'bcast_delete' and 'read_delete'.
Here's the error (any idea what may be going on? Thanks!):
----------------------------------------------------------------------------------------------------------
Using MPI from /gfs/software/software/mvapich2/1.0-2008-02-06-intel-shlib
Running on 1 nodes
4 cores per node, one MPI process per core
for a total of 4 MPI processes
.................................................
PBS nodefile:
n018
n018
n018
n018
.................................................
mpd ring nodefile:
n018:4
.................................................
Running mpdboot ...
done
Running mpdtrace ...
n018
done
Running mpdringtest ...
time for 1 loops = 0.000305891036987 seconds
done
Using MPIEXEC_CMD=/gfs/software/software/mvapich2/1.0-2008-02-06-intel-shlib/bin/mpiexec
-n 4
Commencing parallel run tjunc6RIB_16 of executable
/home/karpeev/fathom/moab/unstable/build/test/parallel/mbparallelcomm_test
Couldn't read mesh; error message:
Failed in step PARALLEL RESOLVE_SHARED_ENTS


application called MPI_Abort(MPI_COMM_WORLD, 0) - process 0Finished
........................................
Running mpdallexit ...
done