[mpich-discuss] Error with mpich2, hydra, sge

Richard Jacobsen ramassa at ucdavis.edu
Fri Apr 8 13:37:15 CDT 2011


I the MPIR_Init_Thread error was because the job was changing out of the 
working directory.
After resolving that I had a few PE errors.  Here's the SGE PE that 
seems to work well with MPICH2 1.3.2p1/Hydra:

qconf -sp hydra2
pe_name            hydra2
slots              9999
user_lists         NONE
xuser_lists        NONE
start_proc_args    /bin/true
stop_proc_args     /bin/true
allocation_rule    $fill_up
control_slaves     TRUE
job_is_first_task  TRUE
urgency_slots      min
accounting_summary FALSE

Thanks for all of the help,
Richard


On 04/08/2011 09:20 AM, Pavan Balaji wrote:
>
> On 04/07/2011 06:23 PM, Richard Jacobsen wrote:
>> I also just tried cutting a few things out of this users job script, and
>> now the mpiexec is running.  I'm going to see if I can isolate what
>> caused the error.
>
> I'm not so certain this has anything to do with Hydra. Can you try 
> running the application without mpiexec? It will behave as if it was 
> launched with a single process, but will not use hydra at all. I'm 
> guessing that it will abort with the same error.
>
>  -- Pavan
>



More information about the mpich-discuss mailing list