[GE users] qdel not deleting all mpi slave tasks

Jon Lockley Jon.Lockley at comlab.ox.ac.uk
Thu Jun 10 10:28:23 BST 2004


Which version of MPICH are you using? When we upgraded to 1.2.5 I had to
add the lines

MPICH_PROCESS_GROUP=no
export MPICH_PROCESS_GROUP

to the startmpi.sh script which seems to have helped. I think there was
some discussion on the list about newer versions of MPICH changing the
process group of the legs of the job? Maybe someone with a better memory
than mine can clarify...

Jon

On Thu, 10 Jun 2004, Lengyel, Florian wrote:

> Yes, I am using the tight integration template: this is how it's modified
> for my setup:
>
> pe_name          mpich
> queue_list       all
> slots            999
> user_lists       NONE
> xuser_lists      NONE
> start_proc_args  /usr/local/sge/mpi/startmpi.sh -catch_rsh $pe_hostfile
> stop_proc_args   /usr/local/sge/mpi/stopmpi.sh
> allocation_rule  $round_robin
> control_slaves   TRUE
> job_is_first_task FALSE
>

---------------------------------------------------------------------
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