[R-sig-hpc] Rmpi working with OpenMPI and PBSPro but snow fails

[Huw Lynes]

On Thu, 2009-03-05 at 11:24 +0900, Ei-ji Nakama wrote:
> Hi
> 
> To see whether your Open MPI installation has been configured to
> use the PBS launcher
> 
> $ ompi_info | grep tm


Yes it is.

[scmhl2 at arccacluster8 Rmpi_tests]$ ompi_info | grep tm
              MCA memory: ptmalloc2 (MCA v2.0, API v2.0, Component v1.3)
                 MCA ras: tm (MCA v2.0, API v2.0, Component v1.3)
                 MCA plm: tm (MCA v2.0, API v2.0, Component v1.3)


> 2009/3/4 Huw Lynes <lynesh at cardiff.ac.uk>:
> > WARNING: ignoring environment value of R_HOME
> 
> Please imagine the working of openmpi and PBS when MPI_spawn.

I'm not sure I understand what you mean. Could you elaborate?

> You try to define the environment variable in the shell of PBS.

In this instance the shell of PBS is just a bash script launched by
pbs_mom so setting environment here or in the login shell is
functionally equivalent except that PBS isn't entirely reliable when it
comes to sending complex environments through.

> set it before qsub.
> 
> $ man qsub       # It might be different in your environment.
>   -V    Declares that all environment variables in the qsub command's
>          environment are to be exported to the batch job.
> 
> $ module load apps/R
> $ module load libs/R-mpi
> $ qsub -V pbsshell.sh

Behaves exactly the same. Including the warnings about R_HOME.

Thanks,
Huw

> 
-- 
Huw Lynes                       | Advanced Research Computing
HEC Sysadmin                    | Cardiff University
                                | Redwood Building, 
Tel: +44 (0) 29208 70626        | King Edward VII Avenue, CF10 3NB