[mpich-discuss] Fw: unable to start all procs

[Dave Goodell]

Hello Sima,

Please don't spam several MPICH2 mailing lists at once.  It won't get  
you help with your problem any faster and it has the potential to  
cause minor problems with some of our automated mail processing  
software that a person has to go in and clean up by hand.


As to your problem, please use mpdboot to start mpd daemons on all of  
your nodes as described in the Quick Start section of the MPICH2  
Installer's Guide:

http://www.mcs.anl.gov/research/projects/mpich2staging/balaji/mpich2/documentation/files/mpich2-1.1-installguide.pdf

Specifically, step 13 shows an example of the mpdboot command,  
although you should make sure to read and follow all of the steps in  
that section.

-Dave

On Jul 21, 2009, at 6:03 AM, sima sima wrote:

>
>
> ----- Forwarded Message ----
> From: sima sima <simasima_64 at yahoo.com>
> To: mpich2-maint at mcs.anl.gov
> Sent: Tuesday, July 21, 2009 3:16:41 PM
> Subject: unable to start all procs
>
> Hi Dear
> I installed the mpich2-1.0.8 in a cluster with 30 node,
> but when I use mpirun command for parallel run of WRF model such as  
> fallow :
> [hamzelou at rserver run]$ mpirun -np 20 ../wrf.exe
> the all of 20 prossesser go on rserver while it have only 5  
> prossesser,
> then I used machinfile option
> [hamzelou at rserver run]$ mpirun -machinefile /home/hamzelou/hamzehlou/ 
> machine20 -np 20 ./wrf.exe
> mpiexec: unable to start all procs; may have invalid machine names
>   remaining specified hosts:
>   172.17.30.1 (rnode1.research.com)
>   172.17.30.4 (rnode4.research.com)
>   172.17.30.2 (rnode2.research.com)
>   172.17.30.3 (rnode3.research.com)
>
> I can login with ssh on all of nodes,
> I use mpdtrace command output is only
> rserever
> Please help me to run it on all of nodes
> Thanks
> Sima
>
>
>