Newsgroups: comp.parallel.mpi
From: rdaoud@magnus.acs.ohio-state.edu (Raja B Daoud)
Subject: Re: lam52's bug in IRIX 5.2
Organization: Ohio Supercomputer Center
Date: 29 Jan 1995 09:02:32 GMT
Message-ID: <3gflj8$2lk@charm.magnus.acs.ohio-state.edu>

Rysanek Martin <rysanek@rhino.cis.vutbr.cz> wrote:
>I found in the file netexec.c (??) a such a bug in definition of
>remote environment during Remote Shell execution. It is completely
>unpossible to execute .profile to create remote evironment.
>So I changed the execution of this .profile to other environment definition
>file (.envdef). Otherway I was not able to setup cluster of computers.
>(I can not execute my .profile remotely.)

Hi Martin,
The last sentence is the key.  By design, you need to be able to execute
your shell startup file (.profile or .cshrc) remotely in order to have the
search path set.  Having the search path set is an easy way to give LAM
the flexibility of locating the binaries on all nodes without forcing the
users to place their files in a fixed pre-determined directory.
If you can't execute the shell startup file remotely, usually because of
commands that require a terminal (such as stty(1)), then just change this.
Either remove the offending command(s), or place them into a segment that
only gets executed if the shell is interactive.  I think this is simpler
than modifying the code.
Regards,

--Raja

-=-
Raja Daoud				raja@osc.edu
Ohio Supercomputer Center		http://www.osc.edu/lam.html

