Rlaph,

Thanks for the reply.  I should be able to build my own.

Tim

Ralph Castain wrote:
Hmmm...well, there -used- to be a tool that was distributed with the 1.2 series for doing just that, but I don't see it in the 1.2.7 release. Not sure when or how that got dropped - probably fell through a crack.

Unfortunately, minus that tool, there is no clean way to shut this down. However, you can re-use the universe name if you simply go to your tmp directory, find the session directory for you, and just rm -rf the directory with that universe name.

So you want to look for something like "/tmp/openmpi-sessions-username@hostname_0/univ3" per your example below, and blow the "univ3" directory tree away.

Sorry it isn't simpler - trying to re-release with that tool is probably more trouble than it is worth now, especially given that the "seed" operation isn't used anymore beginning with the upcoming 1.3 release.

Ralph

On Sep 22, 2008, at 10:08 AM, Timothy Kaiser wrote:

Greetings,

I have a manager/worker application.  The
manager is called "t2a" and the workers "w2d"

I launch the manager and each worker with
its own mpiexec with n=1.  They connect using
various calls including MPI_Open_port,
MPI_Comm_accept, MPI_Comm_connect and
MPI_Intercomm_merge.

It works fine.


I am using the command:

orted --persistent --seed --scope public --universe univ3 --set-sid

to set up the universe and the mpiexec commands are:

mpiexec -np 1 --universe univ3 t2a

mpiexec -np 1 --universe univ3 w2d

mpiexec -np 1 --universe univ3 w2d

mpiexec -np 1 --universe univ3 w2d


Question:

What is the proper way to shutdown orted?
I have found that if I just kill orted then
I can't reuse the universe name.

Platforms and OpenMPI versions:

OS X  openmpi-1.2.7 or openmpi-1.2.6 (ethernet)

Rocks openmpi-1.2.6 (Infiniband)




Thanks!

Tim

--
------------------------------------------------------
Timothy H. Kaiser, Ph.D. tkai...@mines.edu  CSM::GECO
"Nobody made a greater mistake than he who did nothing
because he could only do a little" (Edmund Burke)


_______________________________________________
devel mailing list
de...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/devel

_______________________________________________
devel mailing list
de...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/devel



--
------------------------------------------------------
Timothy H. Kaiser, Ph.D. tkai...@mines.edu  CSM::GECO
"Nobody made a greater mistake than he who did nothing
because he could only do a little" (Edmund Burke)

Reply via email to