"Morten Trab" <[EMAIL PROTECTED]> writes: > ----- Original Message ----- > From: "Tim Legant" <[EMAIL PROTECTED]> > To: "Morten Trab" <[EMAIL PROTECTED]> > Sent: Friday, January 23, 2004 9:31 AM > Subject: Re: Sender>Postfix>TMDA>Postfix>Cyrus ?? > > > >> > The problem is that Cyrus uses virtual users, say >> > [EMAIL PROTECTED] is a non existing account, instead the >> > account is named [EMAIL PROTECTED] and the >> > other is an alias...The user BlackChart is non existent on the >> > system... >> >> Ok, so how does Postfix know how to deliver "[EMAIL PROTECTED]" >> to the Cyrus user? Is Cyrus responsible for doing the virtual user >> address to Cyrus account conversion, or is Postfix? > > Postfix does that, from a MySQL database containing the aliases and the real > addresses...
[...] >> In your proposed path (from Subject:), how is Postfix to determine >> whether the mail just came from Sender or whether the mail came >> from TMDA? > > Mail rules in Postfix...From subject, replace TMDA with Anomy/Spamassassin, > then you have the current setup :) Just need to change Anomy/Spamassassin > with TMDA as I see TMDA far more efficient in catching spam... Ok. I'm still trying to understand this. Your original Subject implies that mail *leaves* Postfix, enters TMDA, leaves TMDA and *re-enters* Postfix. You say, in the immediately preceding paragraph, that what you have today is the same situation except that Anomy/SpamAssassin re-inject the mail to Postfix. How do they do it? Or perhaps I've misunderstood -- does a message actually never leave Postfix and re-enter Postfix later? Is Postfix simply using the results of the Anomy/SpamAssassin pair to determine how to *continue* processing a message? Your Subject implies re-injection, but if Anomy/SpamAssassin are able to re-inject the message back into Postfix for final delivery, then TMDA should be able to do the same. If the message never really leaves Postfix, does Postfix have some "magical" understanding of Anomy/SpamAssassin, or does it just look at the exit code from those programs? Does it treat them like a Unix filter, continuing to process the mail if they write it to stdout? >> If it can't, you've just created a mail loop from Postfix to TMDA >> back to Postfix, etc. If it can, that may be the key to figuring >> out the correct forwarding address for the second injection to >> Postfix. > > I don't see how I'd be able to forward it to any address at all?! If I > forward all to ie. [EMAIL PROTECTED] then ALL mail would go to > [EMAIL PROTECTED], right? Or have I misunderstood something in your > explanation?? DELIVERY needs to be set differently per-recipient. If you forward mail to [EMAIL PROTECTED] to [EMAIL PROTECTED] and you forward [EMAIL PROTECTED] to [EMAIL PROTECTED] they won't both end up in the blackchart user's mailbox.... DELIVERY can be set differently each time tmda-filter runs, since both /etc/tmdarc and ~/.tmda/config are both written in executable Python. The trick is discovering what DELIVERY should be set to for a given user and then how to set DELIVERY properly, per-message, based on that information. Tim _____________________________________________ tmda-users mailing list ([EMAIL PROTECTED]) http://tmda.net/lists/listinfo/tmda-users
