MailScanner Postfix question

[Glenn Steen]

On 10/02/06, Drew Marshall <drew at themarshalls.co.uk> wrote:
>
> On 10 Feb 2006, at 13:40, Rodney Green wrote:
>
> >
> >
> > Dhawal Doshy wrote:
> >> Rodney Green wrote:
> >>> Hello,
> >>>
> >>> Does MailScanner bypass Postfix's cleanup daemon? I can't
> >>> remember at what point mail is reintroduced to Postfix from
> >>> MailScanner.
> >>
> >> The cleanup daemon takes care of header_checks, which is
> >> responsible for putting mails in the hold directory.. so no,
> >> mailscanner doesn't bypass the cleanup daemon.
> >>
> >> See: http://www.postfix.org/cleanup.8.html, for more details..
> >
> > Thanks Dhawal. I was thinking that SMTPD did the header_checks.
> > That is why I was confused as to whether or not cleanup is run.
> >
> > Rod
>
> Clean up also does the virtual alias resolving also, which is a point
> worth remembering as any addresses that are re-injected by
> MailScanner need to be deliverable by Postfix using either Transport
> (Maps if the route can not be identified by MX/ DNS) or local
> delivery agent (As these happen/work after the initial clean up and
> therefore after MailScanner does it's bit). If you use a virtual
> alias they will bounce as these are resolved and translated (Hence if
> you check your logs you will see
> real at virtual.dom,alias at virtual.domlisted as recipients. One is the
> address Postfix will use for delivery, the other the address the
> message was addressed to) before MailScanner gets the message.
>
> Drew
>

I'm sure you will correct me if I'm thinking in muddled puddles (as I
seem to have been doing the last couple of days, or so), but there is
one more thing worth noting with the singel queue/hold setup... Jules
(and others) have been asking how to make Postfix split messages to
more than one recipient, so that the messages are handled one
message/recipient.... Here is why it doesn't really play with PF. The
settings (*_recipient_limits) are enforced _after_ MailScanner does
it's bit, so any such splitting would be entirely pointless....
Unfortunate, but (I think.... I confess I've been indulging in some
nice Montepulchiano, just to honour the Olympics.... So might not have
the best-working intellect:-) the only way to get at that
functionality is to take Joshua Hirshs advice and use a dual-queue
setup... Or does the defer happen prior to ... Oh bugger... You'd need
a strange dual setup, or a *triple* setup to get this.... The first
relaying through some delivery thing (and splitting them) onto the
hold one, or the defer one... Sigh. Not good.

--
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se