[GE users] mpich2 smpd jobs not ending

Reuti reuti at staff.uni-marburg.de
Fri Sep 28 09:58:38 BST 2007


Am 27.09.2007 um 23:07 schrieb Benjamin Singer:

> Couldn't get daemonless to work (mpich2_smpd_rsh--
> getting a bad file descriptor error), so I moved on to
> the smpd daemon based mpich2_smpd method. This is on
> mac os x.

both methods are working without SGE?

> I've got the mpich2_smpd pe almost working-- mpihello
> is running and saying hello from the nodes. But the
> jobs don't die, I need to qdel them. When I do the
> qdel everything cleans up nicely.
> Any idea why my jobs won't take the final step and
> end?

What is the final step? The mpirun, the jobscript or the PE  
stopscript. Can you investigate this by putting some echo commands in  
these scripts?

Although I'm typing this on a Mac, I have no cluster of Macs to test  
this. But I remember a similar problem with LAM/MPI under SGE on Macs  
on this list.

-- Reuti

To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net

More information about the gridengine-users mailing list