Postfix + stunnel SMTPS = address rewriting issue
in [Postfix]
|
Prev: different user levels receiving and sending
Next: smtpd_recipient_restrictions - Is PERMIT at the end advisable ?
From: Jack Browning on 2 Jun 2010 02:48 On Tue, Jun 1, 2010 at 6:17 PM, Wietse Venema <wietse(a)porcupine.org> wrote: > Jack Browning: >> No address rewriting is occurring even though Postfix is invoking its >> SMTP client to deliver the mail to the remote host, and my generic map >> (after postmap and a reload) contains an entry like this: >> >> zzzzzz(a)jnjroos.net xxxxxx(a)att.net > > Sorry this is very incorrect. > > The from= line, logged by the queue manager, is not subject to SMTP > generic mapping. It never was, and it never will. > > SMTP generic mapping is implemented in the Postfix SMTP client. > This mapping is done only for information that is sent over the > network. This also explains why: > > 1) SMTP generic mapping has no effect on SASL password lookup. It > never did, and it never will. > > 2) SMTP generic mapping does not change with the destination host. > It never did, and it never will. > > Wietse Live and learn. Since the fix suggested earlier in this thread -- keying the remote username:password entries to the users' local addresses -- produces the desired result, the issue didn't involve generic mapping at all. No generic mapping was done until the message was actually sent to the remote server, and the message was not being sent because the mis-keyed password lookup table resulted in bad (or no) credentials being presented to the remote server, which rejected the authentication attempt and terminated the session before the message itself was transmitted. No transmission = no rewriting. Anyway, that's my story, and I'm sticking to it. JEB
From: Wietse Venema on 2 Jun 2010 08:25
Jack Browning: > Live and learn. Since the fix suggested earlier in this thread -- > keying the remote username:password entries to the users' local > addresses -- produces the desired result, the issue didn't involve > generic mapping at all. No generic mapping was done until the message > was actually sent to the remote server, and the message was not being > sent because the mis-keyed password lookup table resulted in bad (or > no) credentials being presented to the remote server, which rejected > the authentication attempt and terminated the session before the > message itself was transmitted. No transmission = no rewriting. > > Anyway, that's my story, and I'm sticking to it. Are you sticking with your subject line "Postfix + stunnel SMTPS = address rewriting issue"? All I have seen sofar is that sender-dependent SASL password lookups are working as promised. Wietse |