[MPICH] MPICH_PORT_RANGE environment variable

[Ralph Butler]

Hi Martin:

The code I ran is available via the release candidate at:
     ftp://ftp.mcs.anl.gov/pub/mpi/mpich2-1.0.4-rc1.tar.gz

mpd uses MPIEXEC_PORT_RANGE if available, else uses
MPICH_PORT_RANGE if it is available.

--ralph

On WedJul 5, at Wed Jul 5 8:48AM, Martin Schwinzerl wrote:

> Dear all,
>
> I've still problems / open questions regarding the usage of the  
> MPICH_PORT_RANGE variable
> and would be very thankful if somebody here could again try to help  
> me :
>
> As discussed earlier (http://www-unix.mcs.anl.gov/web-mail-archive/ 
> lists/mpich-discuss/2006/06/msg00026.html),
> I've so far not succeeded to limit the port range of mpd et al by  
> storing the desired port range into an env. variable
> (cf example below).
>
> my open questions :
> ----------------------------
>
> Since Ralph Butler was able to use the mpd sucessfully with the  
> MPICH_PORT_RANGE variable,
> I would like to ask, if this was achieved using the developer  
> version from CVS, or the "public" version 1.0.3 ?
>
> Is the MPICH_PORT_RANGE variable sensitive to other enviroment  
> variables (MPIEXEC_PORTRANGE ?)
> or other enviromental influences, so that there might be an  
> interference ?
>
> example :
> ----------------------------
>
> mas at thph037214:~> export MPICH_PORT_RANGE=8100:8200
> mas at thph037214:~> echo $MPICH_PORT_RANGE
> 8100:8200
> mas at thph037214:~> mpd --echo &
> [1] 31058
> mas at thph037214:~> 1047
> 1047
>
> mas at thph037214:~> mpdallexit
> [1]+  Done                    mpd --echo
>
> System information :
> -----------------------------
>
> mpich2-1.0.3 , Compiled with Intel's icc 9.0 or gcc 4.x (aboves  
> example
> is reproducible with both of them), compiled with f77,f90 and c++  
> bindings,
> python 2.4.1,  network consists at the moment for testing purposes  
> of two
> OpenSuSE 10.1 powered 32 Bit Pentium4 class computers.
>
> Many thanks in advance !
>
> Best regards
> Martin
>
>
>