Re: [cobalt-users] Mail help on RAQ 4

[flash22@xxxxxxx]

On Sat, 19 Jan 2002, Jeff Lasman wrote:

> flash22@xxxxxxx wrote:
> 
> > > ... while talking to chi.maxsolns.com.:
> > 
> > ehems.org       canonical name = chi.maxsolns.com
> >                 ^^^^^^^^^^^^^^
> > You can't do that, you aliased the domain with a cname, you can not also
> > specify an MX for it, the alias is all you are allowed to do if you use
> > it, the MX data in fact comes from where it points, you have created a
> > loop ;P
> 
> I'm not sure what you mean, because mail.maxsolns.com and
> chi.maxsolns.com are NOT necessarily the same machine; I don't see why
> you see a loop.  Please explain so we can all learn <smile>.
> 
> I do see, btw, that pcnet2..pcnet.net is listed as a nameserver for the
> domain but it doesn't seem to be authoritative for it.

Perhaps i should have made that a seperate sentence, sigh, the 'loop' is
indicated by 'excessive recursion' during canonification ;P 

Following out exactly what the loop is requires also knowing what alias's
are on chi.maxsolns.com...

But note that this machine isn't supposed to be getting the mail in the
first place, he has MX records pointing to pcnet1.pcnet.com and
mail.maxsolns.com, presumably he intends these machines to handle that
mail, the fact that the reporting MTA is 'chi...' tells me right there
that the wrong machine got the mail, and it got it because the domain is a
cname to another domain, so the MX records don't belong there...They were
in fact, ignored by the client...

I'd guess 'chi' has no clue what it's supposed to do with the mail and is
doing something typicaly stupid like trying to add 'www' to the domain in
the RCPT field hoping in vain that it eventually resolves to the name of
an interface it is supposed to accept mail on ;p [I really love that
cobalt 'feature', but we won't go there']

as to the his nameserver, it's an improvment, when i looked before, there
was only 1 NS record for the domain, now he seems to have 3, 2 of which
are duplicates ;P

I think he needs to sit down and review his dns ;)

ps: I'm not gonna quote the cname/MX rule from the RFC, i already did that
, it's in the archives -

gsh