<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7036.0">
<TITLE>RE: [mpich-discuss] novice question</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<P><FONT SIZE=2> Hi,<BR>
To run your MPI program across multiple hosts use the "-machinefile" option of mpiexec (See the window's developer's guide for more information about the various options available with mpiexec. The guide is available at <A HREF="http://www.mcs.anl.gov/research/projects/mpich2/documentation/index.php?s=docs">http://www.mcs.anl.gov/research/projects/mpich2/documentation/index.php?s=docs</A>).<BR>
<BR>
(eg: mpiexec -n 2 -machinefile mf.txt a.exe)<BR>
<BR>
Please try providing the ip addresses of the machines in your machinefile so that you don't have problems with the host name resolution. Also make sure that the executable is available (or accessible) at both the machines. If you are sharing the executable using the network mapped drive, make sure that you map the drive using the "-map" option of mpiexec (eg: mpiexec -n 2 -machinefile mf.txt -map j:\\sharecompname\sharedFolderName j:\a.exe). Instead of sharing the executable using the network shared folder you can also copy the executable to both the machines.<BR>
<BR>
>>> First question - do I need Windows server (or server computer cluster) on the other machine? <BR>
No (eg: You can run your MPI program across 2 WinXP machines)<BR>
<BR>
>>> Second - Do I need any of the mpich stuff on the other machine?<BR>
Yes, You need to install MPICH2 on all the machines which are part of your MPI job.<BR>
<BR>
>>> Third - does the MPICH library have a way to verify that other hosts are reachable?<BR>
You can check the status of the process manager, smpd, on each machine using the "-status" option of smpd command (eg: "smpd -status 10.1.1.1")<BR>
<BR>
Let us know if you need further help.<BR>
<BR>
Regards,<BR>
Jayesh<BR>
<BR>
-----Original Message-----<BR>
From: owner-mpich-discuss@mcs.anl.gov [<A HREF="mailto:owner-mpich-discuss@mcs.anl.gov">mailto:owner-mpich-discuss@mcs.anl.gov</A>] On Behalf Of Mark Sale<BR>
Sent: Thursday, June 26, 2008 3:51 PM<BR>
To: mpich-discuss@mcs.anl.gov<BR>
Subject: [mpich-discuss] novice question<BR>
<BR>
I'm working on a parallel/MPI Fortran application for windows (XP and/or Vista). Runs fine on one machine with mpiexec -n X a.exe but I'm having trouble getting it to run on other hosts, e.g., mpiexe -n X -host slave11 a.exe I get this:<BR>
Unable to connect to 'slave11:8676',<BR>
sock error: generic socket failure, error stack:<BR>
MPIDU_Sock_post_connect(1228): unable to connect to slave11 on port 8676, exhausted all endpoints (errno -1)<BR>
MPIDU_Sock_post_connect(1244): gethostbyname failed, No such host is known. (errno 11001)<BR>
<BR>
<BR>
First question - do I need Windows server (or server computer cluster) on the other machine? <BR>
Second - Do I need any of the mpich stuff on the other machine?<BR>
Third - does the MPICH library have a way to verify that other hosts are reachable?<BR>
I'm using Intel Fortran (version 9), with Visual Studio.<BR>
<BR>
thanks<BR>
Mark<BR>
<BR>
<BR>
<BR>
<BR>
</FONT>
</P>
</BODY>
</HTML>