John S Russell wrote:
On Mon, 3 Sep 2007, Bojan Zdrnja wrote:
Then I said:
Please tell us that both [A] and [B] know all your users and domains (in other words: they reject e-mails for non existent users during the SMTP session). And you said:
[A] and [B] are networks, not mail servers. Sure, they are networks, but they both have MTAs. If your MTA in network [A] is your primary (and let's hope it's properly configured to reject e-mails for non existent users immediately) and the one in [B] has no idea about your users but knows that it has to accept everything for [A]'s domain then you have a problem I mentioned before.
I'm not sure why you're bringing this up. Everyone else is talking about basic low-level connectivity - they ability for a mail server on network (C) of the diagram to deliver mail to and MAX on (B) if (A) is unreachable, or vice versa.
You're the only one who's brought up higher-level aspects of mail processing and handling at this point. We're all talking about getting mail delivered to a valid MX, and you're the only one talking about what happens if the MX config is screwed up.
Sure, if Joe has his MX on network (A) configured correctly, and his MX on network (B) configured to do Stupid Shit, then (B) will surely do Stupid Shit. I don't see why you'd think this would be so.
I think the original point was that there is very rarely any usefulness in having a secondary MX record and that, in the majority of cases, it actually does more harm (to the generic community) than good. I think that everyone agrees that in *some* cases having a secondary MX is a good thing [tm] and that in *some* cases, clued people are actually capable of having a secondary MX that behaves correctly and doesn't cause as much disruption to the community. -- Steve.