Re: Secondary MX record for mail backup
> > Does anyone have any pointers or a doc file somewhere on how to setup
a
> > secondary MX server. Basically, if the primary MX is down, then try
the
> > secondary MX (or at least compliant mail servers should try secondary
> > MX...). The secondary MX server should more-or-less spool/store the
> > messages, and as soon as the primary MX comes backup, send the
messages
> > back to it.
> >
> > Thus, no mail is lost or bounced, and everything is back to normal as
soon
> > as the primary MX comes back up.
> >
>
> All you need for qmail is rcpthosts. See the O'Reilly DNS book for
> the rest. Not much to it.
Yeap... no problem with the secondary MX setup in BIND... its simply
adding another MX line. So *ALL* that needs to be done to make qmail work
as a secondary MX for a particular domain is the simply add that domain
name to the rcpthosts file?
> The whole point of an MX host is reliability. That reliability
> is best provided off site on another system. However, if you do
> not have the ability to control spamblocks and filters on the MX
> host, then spam will route around your primary server and come into
> your system via MX host. You can't block the MX host. :-)
Well, we use Spamcop's (spamcop.net) blocks on all the mail servers we
operate, so presumably if spam is bounced at the primary it will also
bounce at the secondary.
> FWIW, we just killed our off site MX hosts for that reason. spam is
> a bigger problem than reliability. Nowadays most mail servers will
> just queue the mail anyways so your mail server doesn't have to be
> up always. .
>
Yeah... i know. But for customers and clients, and even general business
email, it just doesn't look good to have a bounced email telling them that
it wasn't deliverable the first time, and will try again (even though the
mail isn't lost).
Sincerely,
Jason
http://www.zentek-international.com
--
To UNSUBSCRIBE, email to debian-isp-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: