[mpich-discuss] mpiexec errors when running cpi

[Kelvin Wong]

Dear all, 

 

I have installed the latest build v1.3.3rc1 of MPICH2 on a Scientific Linux
v5.4 server (node01) and have NFS-shared that installed folder to another
Scientific Linux v5.4 server (node02). When I try the provided cpi program
on node01 using the machinefile with node02, i.e.:

 

node01$ cpiexec -f machinefile-with-node02 -n 1 ./cpi

 

it will return: 

 

[mpiexec at node01] control_cb
(/setup/mpich2/mpich2-1.3.3rc1/src/pm/hydra/pm/pmiserv/pmiserv_cb.c:154):
assert (!closed) failed

[mpiexec at node01] HYDT_dmxu_poll_wait_for_event
(/setup/mpich2/mpich2-1.3.3rc1/src/pm/hydra/tools/demux/demux_poll.c:77):
callback returned error status

[mpiexec at node01] HYD_pmci_wait_for_completion
(/setup/mpich2/mpich2-1.3.3rc1/src/pm/hydra/pm/pmiserv/pmiserv_pmci.c:189):
error waiting for event

[mpiexec at node01] main
(/setup/mpich2/mpich2-1.3.3rc1/src/pm/hydra/ui/mpich/mpiexec.c:385): process
manager error waiting for completion

 

 

Anyone got the same problem here? Do I have to do anything special on node02
in order to ensure the proper execution with the shared MPICH2 installation?


 

Regards,
Kelvin

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20110301/8ca0cbb2/attachment.htm>