Re: Individual spam restrictions for mailbox

On Wed, Oct 13, 2004 at 01:46:24PM +0300, Nick K. Kozubsky wrote:

.. snip ...
> >> I think, that antispam must move to ROUTERS CONFIGURATION of exim...
.. snip ...


I have per-user spam thresholds configured in two different ways.  One is
to just use ACLs as hinted at by the docs.  in the check_rcpt ACL I
compare the individual recipient with those that have come before - if it
matches then it's accepted.  If it has different spam parameters then it's
temporarily rejected and will (we hope) make it through on the next
attempt.

Another reply already spoke of the problem of doing per-user having
already told the sending MTA that you'll accept the email.  The way I
worked around this for a simpler (albeit kludgey) solution was to declare
a constant "reject threshold" but a dynamic "flag threshold".  End result
was to pipe the email through spamc, and in my router I had a
delivery_filter that was a fairly simple perl script that simply accepted
the email and re-munged the headers based on whether or not the mail
should have been flagged.


> May be I will try to set second spam-check into ROUTERS for each delivery...
> Is this stupid? Or not?
>

To summarize --- no need.  You've done the spam check once, and already
have the information you need (assuming you're running the same ruleset
for each address, that is).

Use the information already injected into the headers and do something
clever with a delivery_transport and you may have a solution that's
workable for you.


Dave

-- 

-    Dave Baker      :      [email protected]      :      http://dsb3.com/    -
GnuPG:  1024D/D7BCA55D / 09CD D148 57DE 711E 6708  B772 0DD4 51D5 D7BC A55D