home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.mail.mh
- Path: sparky!uunet!uaisun4!mrl
- From: mrl@uai.com (Mark R. Ludwig)
- Subject: Re: Differences invoking sendmail or using socket (localhost,SMTP)
- In-Reply-To: jromine@buckaroo.ics.uci.edu's message of 17 Dec 92 18:45:39 GMT
- Message-ID: <MRL.92Dec18091531@sun4.uai.com>
- Sender: mrl@uai.com (Mark R. Ludwig)
- Organization: Universal Analytics, Inc., Torrance (LA), CA
- References: <MRL.92Dec14095012@sun4.uai.com> <VIXIE.92Dec16113711@cognition.pa.dec.com>
- <MRL.92Dec17092427@sun4.uai.com> <2B30CAD2.24855@ics.uci.edu>
- Date: 18 Dec 92 09:15:31
- Lines: 45
-
- In article <2B30CAD2.24855@ics.uci.edu> jromine@buckaroo.ics.uci.edu (John Romine) writes:
-
- |> You can give a list of MTAs to contact in the "servers:" line of the
- |> MH mtstailor file. For better fault-tolerance, you could list some
- |> servers which belong to another department or organization. I have
- |> three mail hubs for our 300+ workstations here, and have no problems.
-
- Can one have more than one host in the "localname:" and "pophost:"
- fields? (BTW, what's the chemistry between these two for POP? The
- MH-6.8 documentation, like 6.7.2, wants them both to point to the POP
- service host.)
-
- More generally, has anyone successfully distributed their Electronic
- Post Office for a given group of people? In Real Life, I don't know
- what happens if my local Post Office burns, explodes, implodes, or is
- otherwise converted to dust. I imagine some mail is lost (which was
- sitting in the building waiting to be delivered), and it takes a while
- for mail in the system to find me as some other Post Office takes over
- delivery for my neighborhood. Electronically, a single mail hub for a
- given group of people has identical characteristics.
-
- |> If you have a group of diskless workstations and one file server, it's
- |> pretty obvious that running an MTA only on the file server (and letting
- |> the clients post mail with SMTP) is the way to go. The workstations
- |> won't be doing much if the file server is down; they don't need an MTA.
-
- Right. That's _not_ my situation, and something I should have
- mentioned earlier. The network is a bunch of heterogeneous disk-full
- boxes, each from a different major vendor. When one of them is
- incapacitated, people shift their work to some other machine. Life
- goes on, and management tends not to get too excited about getting it
- repaired. Usually the box is fixed in a couple days. Sometimes the
- box is never repaired, but just pulled by the vendor, and usually
- replaced with a different one. It's wild!
-
- |> For larger sites, other tradeoffs come into play. You should pick
- |> whatever configuration you are comfortable with.
-
- That's what I'm trying to do. There are a lot of considerations which
- are not obvious at first glance. This discussion is helping expose
- them.$$
- --
- INET: mrl@uai.com NIC: ML255 ICBM: USA; Lower Left Coast
- "A computer is one of life's joys; it follows simple rules. Just like
- children, adults need toys, only we like to call them 'tools.'" -- Dave Ross
-