secondary MX (was Re: [TriLUG] co-lo or at home?)

[Mike Broome]

On Sat, Mar 23, 2002 at 11:54:32AM -0500, Tanner Lovelace wrote:
> Hmm.. you do realize there are problems with having
> MX records point to CNAME records (I believe that's actually
> against the specfication).

Can you point to a reference for that?  I just skimmed through RFC 1034
and RFC 1035 -- the two main DNS RFCs -- and I don't see anything about
not having a CNAME for an MX record.  The only warning I found regarding
CNAMEs is in section 5.2.2 ("Aliases") of RFC 1034:

  Several special conditions can occur with aliases.  Multiple levels of
  aliases should be avoided due to their lack of efficiency, but should
  not be signalled as an error.  Alias loops and aliases which point to
  non-existent names should be caught and an error condition passed back
  to the client.

This makes sense from an efficiency point of view and avoiding loops is
always a Good Thing.  We're only dealing with a single level of CNAME
in our setup and no loops so there shouldn't be any problem.

I also googled around a little and nothing jumped out at me about
avoiding having a CNAME for an MX record.  (But since there are a
gazillion hits for CNAME and MX record, I could have easily missed it.)

> I personally would be hesitant to 
> point my backup mx to a dynamically allocated IP.  I
> would rather not run the chance of it IP address changing in the
> middle of mail delivery.

This is a valid concern and does present a potential problem.  In our
case, the dynamic IP address from our ISP does not change very often
which helps to minimize the issue.  Of course, as the amount of mail
being handled goes up and/or the frequency of changing the IP address
goes up, the chance of hitting this condition increases.

AFAIK, we have had this problem.  However ... if we have hit this
condition and the server delivering the mail silently drops the mail
rather than retries, we would be none the wiser.  I would expect most
mail servers to retry a number of times on error conditions rather than
giving up after the first try.

I don't think I would recommend this setup for business critical mail
handling, especially for large quantities of mail, but for personal use,
it seems to work just fine.

Mike

-- 
Mike Broome
mbroome(at)employees.org