Wietse Venema:
> Birta Levente:
> > > plaintext retries possibily raising this to four.
> > >
> > > We had considered special-casing "4XX" replies as a reason to not
> > > retry plaintext when TLS connections are lost. If your more complete
> > > logs demonstrate a need to do that, we may add the requisite logic.
> > >
> > > Please post *complete* logging from smtp(8) showing *all* events for
> > > the session in question.
> > >
> >
> > I sent yesterday the logs, but now I just realized that my message was
> > to big.
> > Here is:
> > http://pastebin.com/ghBLBDGP
>
> That is postfix-2.12-20140209, and shows exactly the behavior that
> we implemented. It tries to deliver via TLS, the server hangs up,
> then Postfix immediately tries to deliver to the same server via
> SMTP, where the server also hangs up.
>
> Now the big question is why does this produce a different result
> from the queue manager's point of view than postfix-2.12-20140109
Ah, the difference is that 20140209 does not make the final plaintext
attempt to the "last" MX host. It prepares to do that, then discovers
that all recipients were already deferred.
Wietse
