Sorry for the delay. Runing mpirun whith wrong OMPI_MCA_plm_rsh_agent doesn't
give any explicit message in OpenMPI-1.10.0.
How I can show the problem:
I request 2 nodes, 1cpu on each node, 4 cores on each cpu (so 8 cores availables
with cpusets). Node file is:
[begou@frog7 MPI_TESTS]$ cat $OAR_NODEFILE
frog7
frog7
frog7
frog7
frog8
frog8
frog8
frog8
I launch the application (I've added a grep here to limit the output on stdout
and juste check processes location):
[begou@frog7 MPI_TESTS]$ mpirun -np 8 --hostfile $OAR_NODEFILE --bind-to core
./location.exe |grep 'thread is now running on PU'
(process 2) thread is now running on PU logical index 2 (OS/physical index 6) on
system frog7
(process 3) thread is now running on PU logical index 3 (OS/physical index 7) on
system frog7
(process 0) thread is now running on PU logical index 0 (OS/physical index 0) on
system frog7
(process 1) thread is now running on PU logical index 1 (OS/physical index 5) on
system frog7
(process 6) thread is now running on PU logical index 2 (OS/physical index 2) on
system frog8
(process 7) thread is now running on PU logical index 3 (OS/physical index 3) on
system frog8
(process 4) thread is now running on PU logical index 0 (OS/physical index 0) on
system frog8
(process 5) thread is now running on PU logical index 1 (OS/physical index 1) on
system frog8
So one process on each core, no oversubscribing allowed with the patch applied
in OpenMPI.
Now I set OMPI_MCA_plm_rsh_agent so something wrong and launch agin the job
(without the final grep to have all informations):
[begou@frog7 MPI_TESTS]$ export OMPI_MCA_plm_rsh_agent=do-not-exist
[begou@frog7 MPI_TESTS]$ mpirun -np 8 --hostfile $OAR_NODEFILE --bind-to core
./location.exe
--------------------------------------------------------------------------
A request was made to bind to that would result in binding more
processes than cpus on a resource:
Bind to: CORE
Node: frog7
#processes: 2
#cpus: 1
You can override this protection by adding the "overload-allowed"
option to your binding directive.
--------------------------------------------------------------------------
The message only show only that OpenMPI try to allocate all processes on the
local node.
Of course:
[begou@frog7 MPI_TESTS]$ which do-not-exist
/usr/bin/which: no do-not-exist in (/home/PROJECTS/...............
Patrick
--
===================================================================
| Equipe M.O.S.T. | |
| Patrick BEGOU | mailto:patrick.be...@grenoble-inp.fr |
| LEGI | |
| BP 53 X | Tel 04 76 82 51 35 |
| 38041 GRENOBLE CEDEX | Fax 04 76 82 52 71 |
===================================================================
