torque+mpiexec+mvapich = strange behavior
Anton Starikov
A.Starikov at utwente.nl
Sun Dec 5 20:57:15 EST 2004
Hello!
I've found really strange thing.
I'm using torque-1.1.0p4 + mvapich-0.9.4-103 mpiexec-cvs
And observe something strange.
When I submit interactive job, I can start mpi job without any problem
in interactive session.
But when I submit non-interactive MPI job, I see:
"mpiexec: Error: read_ib_startup_ports: accept iter 0: Invalid argument"
Do you have ideas what can be responsible for a such behavior?
I fill that there is something around fork (in start_tasks), but can't
proove :))
I'm not a big specialist in multithreading, but it seems that both,
parent and child call "prepare_ib_startup_port", and in interactive job
only after this call to "read_ib_startup_ports". In case of
non-interactive job call to "read_ib_startup_ports" is executed after
first call of "prepare_ib_startup_port" and before second one.
Best regards,
Anton Starikov.
More information about the mpiexec
mailing list