[391] in athena10
Re: Larvnet design discussion
daemon@ATHENA.MIT.EDU (Greg Hudson)
Wed Aug 6 12:34:54 2008
From: Greg Hudson <ghudson@MIT.EDU>
To: Timothy G Abbott <tabbott@mit.edu>
Cc: athena10@mit.edu
In-Reply-To: <alpine.DEB.1.10.0808051517420.21810@vinegar-pot.mit.edu>
Content-Type: text/plain
Date: Wed, 06 Aug 2008 12:34:09 -0400
Message-Id: <1218040449.12433.127.camel@error-messages.mit.edu>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
I went back to the drawing board and wound up with a single package
named debathena-larvnet containing:
* An Xsession.d script which wraps the session in larvnet-wrapper
* larvnet-wrapper, a /bin/sh script which uses nc to send a status
packet before and after the login session
* busyd, a perl script to answer busypoll requests
busyd determines if anyone is logged in by running "w" and looking to
see if someone is logged in on tty7. That's not especially general, but
it's general enough for cluster machines. Because no flag file is
needed to decide whether the machine is busy, there is no need to run
code as root before and after the login session.
I'll check this in after testing it.
