Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: [OMPI users] Question on MPMD runs
From: Victor Vysotskiy (victor.vysotskiy_at_[hidden])
Date: 2013-05-30 06:40:50

Dear OpenMPI Developers and Users,

I have general question on signal trapping/handling within mpiexec/mpirun. Let me assume that I have 2 cores and I start two different (independent) prog1 and prog2 programs in parallel via the mpirun/mpiexec strartup command:

mpiexec -n 1 prog1 : -n 1 prog2

What happens if one of the programs just is abnormally crashed/terminated while the second one is still running normally? Is it correct observation that in such case the OpenMPI immediately starts a cleanup process and automatically terminates all spawned/running jobs? If it is like that, is there any way to force mpiexec/mpirun to don't cleanup all processes on error and wait until all spawned processes either successfully complete or abnormally terminate their execution?

Thank you in advance!