[mpich-discuss] Problem running mpiexec within a queue (LSF)

[Pavan Balaji]

By cli, are you creating an interactive session with LSF and then 
running the job?

For the non-interactive version, can you send the content of your bsub file?

  -- Pavan

On 03/07/2012 04:15 AM, Dr.Peer-Joachim Koch wrote:
> Hi,
>
> we are currently replacing the OS of our HPC cluster.
> So I have updated the compiler and also mpich2 to the latest
> stable release 1.4.1p1.
>
> All test are fine. Also running mpi jobs from the cli is no problem and
> working.
>
> Using our LSF queue fails. I get the following error.
> Could somebody please explain, what's missing or wrong ?
>
>
> ####output from LSF ###
> .
> .
>
> The output (if any) follows:
>
> [mpiexec at io4] HYD_pmcd_pmi_alloc_pg_scratch
> (./pm/pmiserv/pmiserv_utils.c:595): assert (pg->pg_process_count *
> sizeof(struct HYD_pmcd_pmi_ecount)) failed
> [mpiexec at io4] HYD_pmci_launch_procs (./pm/pmiserv/pmiserv_pmci.c:103):
> error allocating pg scratch space
> [mpiexec at io4] main (./ui/mpich/mpiexec.c:401): process manager returned
> error launching processes
> Job
> /usr/local/apps/lsf/7.0/linux2.6-glibc2.3-x86_64/bin/mpich2pgi12_wrapper
> ./MPI2hello
>
>
>
> ###
>
> running the     mpiexec -np 8 ./MPI2hello  or      mpirun -np 8
> ./MPI2hello    will work.
>
>
> Any hint ?
>
>
>
>
> _______________________________________________
> mpich-discuss mailing list     mpich-discuss at mcs.anl.gov
> To manage subscription options or unsubscribe:
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss

-- 
Pavan Balaji
http://www.mcs.anl.gov/~balaji