🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: comp.mail.misc
Article #2294

Re: When should one use MX records?

#2294
From: Christopher Davis
Date: Sat, 10 Oct 1992 05:35
41 lines
2865 bytes
[this is leaving mail and heading for DNS territory.  Watch followups]
Tom> == Tom Fitzgerald <fitz@wang.com> 
RD>  == dhesi@rahul.net (Rahul Dhesi) writes:

 RD> c: Several MX records are often assumed to be more reliable for
 RD> mail.  However, it's not clear to me that they buy you much.
 RD> Presumably if foo.company.uk is down, mail will remain queued for
 RD> it at one of the MX hosts.  It might as well just remain queued at
 RD> the sending host.

We define two sorts of failures: machinewide and sitewide.  Machine wide
is level 0 backups, disk reconfigurations, blown ether MAUs, that kind
of thing.  Sitewide is usually either a power outage or our Internet
link going down (the latter mostly).

Machinewide failures of our mailhost (eff.org) get mail delivered to the
first backup MX, which is on the same ether.  Then when we bring the
main machine back up, we just /usr/lib/sendmail -v -Reff.org to flush
all mail with 'eff.org' in the recipient address.  Over the local ether
between two SS2s this is quite fast.  Much faster than waiting for all
those 1 hour (usually) queue runs to come around again.

Sitewide failures fall back to our 'last ditch' MXes at the net
provider.  This is more so that if we're expecting Real Important Mail
while we're down, we can arrange to have them flush their queue at us
when we return to life than anything else.

 Tom> For ftp, it seems like there's a growing tradition for having a
 Tom> CNAME for ftp.company.uk pointing to the ftp server.

We do this too.  The "service" CNAME is a really useful thing to have;
when we added machines, we moved the ftp server to one of the new
machines to spread out the load.  Folks who used 'ftp.eff.org' were set
as soon as the old records timed out where they were; folks who used IP#
or "true" hostname lost (there is a redirect message on the old machine
still, though).
--
Christopher K. Davis      |       NET.INSIGHTS INTO ARTISTIC CRITICISM:
<ckd@eff.org>   EFF #14   | ``[Y]ou don't cure bad art by locking it all up
System Administrator, EFF | in one museum, you cure it by throwing tomatoes
+1 617 864 0665  [CKD1]   | at bad artists.'' --Barry Shein <bzs@world.std.com>
Technical Details
Message-ID: <CKD.92Oct10013531@loiosh.eff.org>
Path: arc1-ca.pugleaf.net!sparky!uunet!spool.mu.edu!sol.ctr.columbia.edu!eff!ckd
References: <BvtG9B.GBo@breeze.rsre.mod.uk> <BvtKy7.4qr@rahul.net> <bvw0j1.1ys@wang.com>

Show Full Headers
Complete Article Headers
Newsgroups: comp.mail.misc,comp.protocols.tcp-ip.domains
Path: arc1-ca.pugleaf.net!sparky!uunet!spool.mu.edu!sol.ctr.columbia.edu!eff!ckd
From: ckd@eff.org (Christopher Davis)
Subject: Re: When should one use MX records?
In-Reply-To: fitz@wang.com's message of Sat, 10 Oct 92 04:01:48 GMT
Message-ID: <CKD.92Oct10013531@loiosh.eff.org>
Followup-To: comp.protocols.tcp-ip.domains
Sender: usenet@eff.org (NNTP News Poster)
Nntp-Posting-Host: loiosh.eff.org
Organization: Electronic Frontier Foundation Tech Central
References: <BvtG9B.GBo@breeze.rsre.mod.uk> <BvtKy7.4qr@rahul.net>
 <bvw0j1.1ys@wang.com>
Date: Sat, 10 Oct 1992 05:35:35 GMT
Lines: 41