On Apr 2, 2009, at 6:35 PM, Jerome BENOIT wrote:
It appeared that the file /etc/openmpi/openmpi-mca-params.conf on
node green was the only one
into the cluster to contain the line
btl_tcp_port_min_v4 = 49152
Great -- glad you found the issue!
Once the this line commented, the tests suggest below, and the
sbatch script previously emailed,
work.
Now, if I put the above line, namely,
btl_tcp_port_min_v4 = 49152
I believe that you have chosen a port number that is too high -- I,
too, can confirm that this breaks for me when I try it. But if I
choose a much lower port number (like 10000), it seems to work. Try
that in your setup on the command line:
mpirun --mca btl_tcp_port_min_v4 10000 ...etc.
Looks like we need to do better range checking / print a better error
message here; our poor output did little/nothing to help diagnose this
problem. Sorry about that! I've opened up ticket #1869 for this
issue (https://svn.open-mpi.org/trac/ompi/ticket/1869).
--
Jeff Squyres
Cisco Systems
