Matt Sergeant wrote: > On 16-Jan-08, at 10:25 AM, Charlie Brady wrote: > >> On Tue, 15 Jan 2008, Hanno Hecker wrote: >> >>> There's no real queueing mechanism with the smtp-forward plugin, the >>> mail will be rejected with a temporary error, which means the mail will >>> stay in the client's queue and sent later again... well at least for >>> most MTAs. Some are buggy that they don't get the difference between a >>> hard failure and a temporary problem and will not retry. >> >> Really? Which? If there are any MTAs less than, say, 15 years old with >> such egregious behaviour, they should be shamed by public disclosure. > > Some older versions of Domino used to do that. I'm sure Chris Lewis will > pipe up with a more thorough list though :-)
Probably not - mysterious behaviour of random MTAs isn't terribly obvious from my perspective. Even if we do find out about an issue, we seldom find out what software was causing it. It's worth noting that RFC821 (and perhaps RFC2821) aren't very clear about behaviours in response to non-success of the "." after DATA, and some MTAs are (were) even more confused. For example, Exchange 5 treated a "550" after "." as a "retry as fast as you possibly can". Others (especially more current Exchange) don't return the associated error message. [We rely on the sender seeing the 550-after-"." message for senders hitting a FP to contact us. Then again, I believe we have vastly more people not understanding our error message than don't get it at all. What I do know is that some fraction (probably <1%) of our inbound email presented with a 4xx after "." never seem to retry. Having our mail servers throw significant numbers of these is one of the few "severity one major" conditions in our support procedures. I made a usenet posting some months ago that had a series of guesstimates for various misbehaviours.
