[Mpi-comments] hydra problem ?

[Chun Meng]

Hi all,

I have an error message while I tried to do a parallel job on a cluster.
The error is listed here:
[mpiexec at M001-********] HYDU_sock_read (./utils/sock/sock.c:222): read
errno (Input/output error)
[mpiexec at M001-********] control_cb (./pm/pmiserv/pmiserv_cb.c:249): assert
(!closed) failed
[mpiexec at M001-********] HYDT_dmxu_poll_wait_for_event
(./tools/demux/demux_poll.c:77): callback returned error status
[mpiexec at M001-********] HYD_pmci_wait_for_completion
(./pm/pmiserv/pmiserv_pmci.c:206): error waiting for event
[mpiexec at M001-********] main (./ui/mpich/mpiexec.c:404): process manager
error waiting for completion

I'm using mpich2-1.3.2 to do the job.It seems the error is pointing to
hydra?
And I'm not sure about if hydra is installed in this cluster.So Maybe I
need to install a same version hydra with mpich2 first?
Also I'm confused how hydra is linked to mpich2-1.3.2 while installing them?
Is there anyone who can give me some hint?

Thanks a lot!

Best,
Chun
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpi-forum.org/pipermail/mpi-comments/attachments/20140729/77146d15/attachment.html>