[mvapich-discuss] mpiexec keeps erroring out. Don't know what to do

[wei huang]

Hi Fred,

What is the output if you do "mpdtrace"?

By the way, it seems that you are oversubscribing your job (you are
running 6 processes on a 4-core node), which is generally not recommended.

Thanks.

Regards,
Wei Huang

774 Dreese Lab, 2015 Neil Ave,
Dept. of Computer Science and Engineering
Ohio State University
OH 43210
Tel: (614)292-8501


On Tue, 11 Mar 2008, Stecher, Fred wrote:

> Hi,
> Just acquired SGI XE1300 with Intel Xeon quad core cpus. It has the SUSE
> 10.0 OS. Have installed MVAPICH. MPD is running on all the nodes. When I
> use the following command from the head node:
>
> mpiexec -machinefile ~/bin/noden01n02 -n 12
> /home/apps/cth/cth8.1/bin/mpicth i=wh
>
> The following error message is output:
> mpiexec: unable to start all procs; may have invalid machine names
>     remaining specified hosts:
>         10.0.1.2 (n002)
>         10.0.1.1 (n001)
>
> If I execute the command on node n001, the following is output:
> mpiexec: unable to start all procs; may have invalid machine names
>     remaining specified hosts:
>         10.0.1.2 (n002)
>
> File noden01n02:
> n001
> n001
> n001
> n001
> n001
> n001
> n002
> n002
> n002
> n002
> n002
> n002
>
> What is happening?
>
>
> Thanks,
>
> Fred
>
>