[torqueusers] jobs completing with processes still running - SOLVED

[Brock Palen]

For amusement, I have included how we build openMPI  with OFED (Cisco  
IB support)  and TM. With the intel compilers.

./configure --prefix=/home/software/rhel4/openmpi-1.2.6/intel-10.0 -- 
with-tm=/usr/local/torque --with-openib=/usr CC=icc CXX=icpc FC=ifort  
F77=ifort

I hope that helps.

Brock Palen
www.umich.edu/~brockp
Center for Advanced Computing
brockp at umich.edu
(734)936-1985



On May 8, 2008, at 4:31 PM, Lloyd Brown wrote:
> Jerry Smith wrote:
>> I would have to second this thought (OpenMPI, as well as OSC's  
>> mpiexec for your current setup).
>> Have you looked into the different epilogues
>> that float around on this list as a way to make sure processes  
>> that may end up
>> outside of the TM interface, get cleaned up?
>>
>> http://www.clusterresources.com/wiki/doku.php? 
>> id=torque:appendix:g_prologue_and_epilogue_scripts
>>
>> --Jerry
>>
> If you do decide to use OpenMPI, though, be sure that the version  
> you install has the TM interface enabled.  For example, I've been  
> testing NPACI Rocks 4.3 and 5.0, and, since they're based on  
> CentOS, the OpenMPI package doesn't include TM by default.  You  
> might have to recompile.  See "http://www.open-mpi.org/faq/? 
> category=building#build-rte-tm", for details, including how to tell  
> if your precompiled version already has it.  It's not too tough.
>
>
> Lloyd Brown
>
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers
>
>