Newsgroups: comp.parallel.mpi
From: Antoine Jonas <jonas@sun.ipl.fr>
Subject: Running MPICH on a cluster of SUN4, SOLARIS, and LINUX workstations
Organization: IPL
Date: Mon, 18 Mar 1996 17:31:01 +0000
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Message-ID: <314D9DD5.41C67EA6@sun.ipl.fr>

Hi !

I have installed MPICH with the device ch_p4 on three workstations, respectively
running SUN4, SOLARIS, and LINUX.

After having tested the basic example "cpi", I have run "cpilog", to test the MPE library.
The results I have obtained are the next one :

- It works fine on any workstation when all of the processes are on the same machine.

- It works fine when some processes are on the SUN4 workstation and some on the
  SOLARIS workstation.

- When I'm trying to run the application with some processes on the SUN4 workstation,
  and some processes on the LINUX workstation, the calculations of PI are done, but I can't get
  any logfile. The message I get after having answered "0" to the number of intervals is the next one :

      p0_13764:  p4_error: interrupt SIGSGV: 11
      bm_list_13765:  p4_error: interrupt SIGINT: 2
      p1_31504: Trying to receive a message when there are no connections; Bailing out
      Segmentation Fault

- I get a similar result when I try to run the application with some processes on the SOLARIS workstation,
  and some processes on the LINUX workstation.

Do you know a method to fix this ?

Thank you for your help.

