[mpich-discuss] Error in the mpdboot step

[Pavan Balaji]

On 02/02/2010 04:52 AM, Prashantha Hebbar wrote:
> mlscub1 at mlscub1-desktop:~$ mpdboot -n 2 -f mpd.hosts --ncpus=1
> unable to obtain IP for host: mlscub1 at mlscub1-desktop.local
> unable to obtain IP for host: mlscub2 at mlscub2-desktop.local
> totalnum=2  numhosts=1
> there are not enough hosts on which to start all processes

The /etc/hosts file on each machine needs to have information on all the
nodes in the system.

> So, can you please tell me what might have gone wrong? I do not think
it is
> a problem of my /etc/hosts setting.
> mlscub1 at mlscub1-desktop:~$ cat /etc/hosts
> 127.0.0.1       localhost
> 172.16.17.24    mlscub1-desktop

The file doesn't contain information on mlscub2-desktop

You can use the mpdcheck utility to look for such errors in your setup.

> I tried it in other way round with specifying hostname of mpd master in
> slave system. That works fine.
>  
> mlscub1 at mlscub1-desktop:~$ mpd &
> [1] 22750
> 
> mlscub1 at mlscub1-desktop:~$ mpdtrace -l
> mlscub1-desktop_50100 (172.16.17.24)
> 
> mlscub2 at mlscub2-desktop:~$ mpd -h mlscub1-desktop.local -p 50100 &
> [1] 11418
> 
> mlscub2 at mlscub2-desktop:~$ mpdtrace -l
> mlscub2-desktop_50514 (172.16.17.93)
> mlscub1-desktop_50100 (172.16.17.24)
> 
> mlscub1 at mlscub1-desktop:~$ mpdtrace -l
> mlscub1-desktop_50100 (172.16.17.24)
> mlscub2-desktop_50514 (172.16.17.93)

Good to know that this works. This is another option, but is more
cumbersome to use, so we don't usually prefer it.

> I have another problem with executing programs. I find something like
> permission denied error messages.
> 
> mlscub1 at mlscub1-desktop:~$ mpiexec -n 5
> /home/mlscub1/libraries/mpich2-1.2.1/examples/cpi.c
> problem with execution of
> /home/mlscub1/libraries/mpich2-1.2.1/examples/cpi.c  on  mlscub1-desktop:
> [Errno 13] Permission denied
> problem with execution of
> /home/mlscub1/libraries/mpich2-1.2.1/examples/cpi.c  on  mlscub2-desktop:
> [Errno 2] No such file or directory
> problem with execution of
> /home/mlscub1/libraries/mpich2-1.2.1/examples/cpi.c  on  mlscub2-desktop:
> [Errno 2] No such file or directory
> problem with execution of
> /home/mlscub1/libraries/mpich2-1.2.1/examples/cpi.c  on  mlscub1-desktop:
> [Errno 13] Permission denied
> problem with execution of
> /home/mlscub1/libraries/mpich2-1.2.1/examples/cpi.c  on  mlscub1-desktop:
> [Errno 13] Permission denied

Why are you trying to execute cpi.c ? Do you mean to compile it and
execute cpi?

% mpicc examples/cpi.c -o examples/cpi

% mpiexec -n 5 ./examples/cpi

 -- Pavan

-- 
Pavan Balaji
http://www.mcs.anl.gov/~balaji