[comp.mail.sendmail] more on site-wide aliasing

news@bellcore.bellcore.com (news) (11/14/89)

I'm still getting mail about site-wide aliasing.  I've suggested to the sender
of the message below that he make his stuff available via anonymous ftp.


=From dwells@fits.cx.nrao.edu Mon Nov 13 23:46:53 1989
=Received: from fits.cx.nrao.EDU by buzz.bellcore.com (5.61/1.34)
=	id AA00633; Mon, 13 Nov 89 23:46:50 -0500
=Received: by fits.cx.nrao.edu (4.0/SMI-DDN)
=	id AA15097; Mon, 13 Nov 89 23:45:46 EST
=Date: Mon, 13 Nov 89 23:45:46 EST
=From: dwells@fits.CX.NRAO.EDU (Don Wells)
=Message-Id: <8911140445.AA15097@fits.cx.nrao.edu>
=To: nrh@buzz.bellcore.com
=Subject: global aliases at NRAO
=Cc: dwells@fits.cx.nrao.edu
=Status: R
=
=I just saw your newsnote.
=I had not seen your prior request, or I would have answered.
=
=NRAO has had a global alias system for about three years now.
=It has two main features:
=(1) account names are standardized across all sites (7 sites in 4 states)
=and OSes (VMS & Unix [several flavors, both Bell and Berkeley]).
=(2) master alias table kept on one machine, distributed after midnight
=each night to all NRAO machines; requires rule-based transformation
=program to translate from DECnet notation [the master table] from 
=perspective of a VMS user to Unix/Internet notation suitable for either
=Berkeley sendmail or Bell mailer. the rule based program is coded in
=Fortran-77 (surprise!); rules are patterns that are matched and result
=patterns with matched fields filled in (not full reg exps, though).
=
=Currently more than 600 aliases for our organization.
=The solution definitely works, even though it is a brute force kluge.
=Well, some people might say it is elegant, but I who designed it always
=considered it brute force even though it has a fancy piece of Fortran
=at its heart. There is also a shell script to hack the Internet alias
=table for the local host (you must delete local references to avoid 
=loops) and a DCL script to hack the master DECnet table for VMS
=installation (likewise avoiding the mail loops).
=
=Several people here have been involved in the long evolution of this
=solution. As far as I know we are willing to give any of it away
=to anybody who can use it. Of course we are always understaffed, and
=so even gathering it all up in minimal form is a hassle.
=But if all you want is solution for Unix systems it is a lot easier.
=If you want the VMS side too then it is a little harder to get hold
=of the code (I myself am responsible only for Unix things).
=
=The effect of the solution is that anybody on any of our machines
=(VMS or Unix) at any of our sites can send mail to any of the 600+
=aliases. Whenever an account is moved from machine to machine the
=new alias can be moved within 24 hours.  We have wide area mail 
=service on four networks, with three different hosts acting as gates.
=Mail arriving on any of these nets at the respective gates can be 
=addressed to any user on any of NRAO's computers. I.e., BITnet mail
=arriving on the VMS-based BITnet gateway can forward to a user on the
=Unix side of the Observatory, and mail arriving on our main Internet
=mail host ("nrao.edu"="cv3.cv.nrao.edu") can forward to a user whose
=mailbox is on any of seven VMS hosts in three states. It works, most
=of the time, with failures mainly due to fact that we cannot afford to
=buy the best quality gateway/mailer code for the VMS machines. sigh...
=
=enough for now. you may summarize this to the net, or even broadcast
=it intact if it suits you. i will be willing to put some effort into
=packaging some of our code for any person seriously interested in the
=technology.
=
=Regards,
=
=Donald C. Wells, Associate Scientist | NSFnet: dwells@nrao.edu [192.33.115.2]
=National Radio Astronomy Observatory | SPAN:   NRAO::DWELLS    [6654::]
=Edgemont Road                        | BITnet: DWELLS@NRAO
=Charlottesville, VA 22903-2475 USA   | UUCP:   ...!uunet!nrao.edu!dwells
=+1-804-296-0277  (38:02.2N/78:31.1W) | TWX=510-587-5482, Fax=+1-804-296-0278