Received: from PACIFIC-CARRIER-ANNEX.MIT.EDU by po7.MIT.EDU (5.61/4.7) id AA16751; Mon, 27 Nov 95 16:08:02 EST
Received: from TAMAR.MIT.EDU by MIT.EDU with SMTP
	id AA23780; Mon, 27 Nov 95 16:04:57 EST
From: miki@MIT.EDU
Received: by tamar.MIT.EDU (5.57/4.7) id AA17202; Mon, 27 Nov 95 16:05:54 -0500
Message-Id: <9511272105.AA17202@tamar.MIT.EDU>
To: dcns-dev@MIT.EDU, dcns-syssupport@MIT.EDU, network@MIT.EDU
Cc: miki@MIT.EDU
Subject: Solaris2.4
Date: Mon, 27 Nov 1995 16:05:53 EST



    In the last n weeks (where n > 5) I worked on trying to establish the
feasibility of Releasing Solaris2.4 on the Athena Solaris Workstations.

    The conclusions are :

     a. The process of building an OS image in an AFS locker is much more
straight forward . We have an os_24 now in AFS.

     b. The process of rebuilding Athena on the new OS release was far from
being easy. A number of more than 70 files would have to be changed in the
"Athena source tree". Nevertheless we have now an srvd for 2.4 with 99% of the
binaries rebuilt.

    c. We can install machines with the new 2.4 release, using the installers
we have, solar and solar2, without needing a new CNBOOT program (diskette). This
is a very important point as we do not have yet the knowledge of updating the
CNBOOT and the kernel programs on which the installation depends on.  At this
time I am inclined to think that as long as we do not update solar and solar2,
we can continue to use them for installing Solaris2.4, Solaris2.5 ...

   d. Having built the new release , based on Solaris2.4, we could make it 
ready at any time, and make it available to a small "cluster" of testers.

   e. The experince of rebuilding the Solaris release allows me to build
now any new release (2.4 or 2.5) in less than a week. It shows as well that
we do not need to update our installation procedure (CNBOOT and kernel friends). 


Miki.
P.S. Thanks to Ron for taking Central out of its misery. 
Thanks to Bill for the fruitful talks with Sun - they were instrumental in
finding a more elegant solution for getting the OS image into AFS. 
Thanks to Craig for putting up with my screaming about the source tree. 

   
