On Fri, 20 Sep 2002, Suresh Ramasubramanian wrote:
> It is perfectly acceptable to have only an A and no MX record. In fact, if
> you see most universities / large organizations where people have the option
> of getting mail delivered to their own workstations rather than to the
> university mailhub, all they need is an A record.
Yes, but it is a right PITA. The reason is that in many universities /
large organizations where people do *not* have the option of getting
mail delivered to their workstations, if they accidentally send out mail
with their workstation name in it, a sending MTA will try for several
days (the usual timeout) before giving up. [In this University, our
local central mailers reject mail to any .cam.ac.uk addresses that don't
have MX records, because we register all MTA-running systems, and give
them MX records.]
The "use A if no MX" rule was for backwards compatibility when MX was
introduced. Now that all kinds of equipment is getting an IP address
(think printers for a start - soon people will be trying to email your
toaster), it is IMHO silly to continue with it. But such is the lethargy
of the standards process, I suspect it will last a long time.
The big mistake was to register hosts in the DNS instead of registering
services. That was one thing the late lamented Janet X.25 network got
right.
> > After weeks of looking at why spam is marked as spam, I've pretty much
> > figured that pretty much all that spam comes from non-reversable IP
> > addresses.
>
> Well, you won't really find that a reliable metric I think - but as you
> please.
There are certainly sites that apply this rule. It's just a matter of
whether it is politically acceptable to your users/customers.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
