[torqueusers] Orphaned processes

Garrick Staples garrick at clusterresources.com
Fri Mar 2 15:59:19 MST 2007


On Fri, Mar 02, 2007 at 01:55:59PM -0800, scoggins alleged:
> I am having a problem with mpi processes hanging around after job  
> completion leaving orphaned processes.  I have implemented the  
> mpicleanup program from http://bellatrix.pcl.ox.ac.uk/~ben/pbs/ 
> epilogue but it is still not cleaning up the nodes.  What else can I do?
> 
> torque version -  1.2.0p2-1.caos
> OS Release = WS  3
> Kernel = 2.4.21-20.EL
> Interconnect = IB using topspin ib
> MPI = topspin-ib-mpi-rhel3-3.0.0.-160

Use OSC's mpiexec to launch your MPI jobs and the processes will be
properly killed.

http://www.osc.edu/~pw/mpiexec/



More information about the torqueusers mailing list