Hi Wolfgang,
On Wed, 16 Jan 2008 15:05:33 +0100
Wolfgang Breyha <wbreyha@???> wrote:
> I use DCC for both counting and greylisting. dccifd (-Gon) returns the
> greylisting results.
[snip]
> That's why i defined dcc_result first and never implemented it finally;-)
>
Ah ok, I understand better now. Nice to hear the think is actually
working with greylisting (as I never tried it myself) :D
> > Therefore, I rewrote a bit your patch (basically removed everything
> > related to rebuilding the recipients list and I also added return
> > values for dcc_result which weren't set apparently), so that I can now
> > have an ACL like this:
>
> Wasn't that part commented already? But yes, my patch is very close to my
> needs here currently.
>
That was actually part of the problems I had. Some parts where
commented but not others so that in the end some counters weren't
updated and thus the return value wasn't right.
> One "special feature" i didn't mention in the readme is the
> dcc_direct_header_add = yes
>
> If you use spamd within DATA ACL and call dcc before, you can set this to add
> the X-DCC-Header "in deep" to the spool file. In this way spamd already sees
> the header (what's not the case if add_header was used) and simple matching
> rules can be used instead of the full SpamAssassin DCC module. eg:
> header __XXDCC_HIGH_BODY X-DCC-xx-Metrics =~ /^.*Body=\d{6,}\s+Fuz.*$/
> header __XXDCC_HIGH_FUZ1 X-DCC-xx-Metrics =~ /^.*Fuz1=\d{6,}\s+Fuz.*$/
> header __XXDCC_HIGH_FUZ2 X-DCC-xx-Metrics =~ /^.*Fuz2=\d{6,}$/
>
> This also prevents double counts etc....
>
That sounds very nice. Do you set the "dcc_direct_header_add = yes"
right into the ACL then?
BTW, I'm now wondering if it would be interesting to split the results
and to return separate values for Body, Fuz1 & Fuz2 to exim. What do
you think?
> > I haven't tested it for long yet but so far it seems to be working fine.
> >
I've now tested it some more and also deployed it on some more heavily
used servers and no problems so far.
> Fine! I'll try to get the patch and documentation reworked again on weekend
> and send it to Tom Kistner for CVS inclusion. The main reason why I released
> the patch mostly "as is" was that I wanted to wait for feedback first.
>
That would be great. Let me know if I can help in some way. :)
Cheers,
> Greetings,
> Wolfgang
Chris
--
,''`. Christopher `Physicman' Bodenstein <cb@???>
: :' : Physicman.Net : http://www.physicman.net/
`. `' Debian GNU/Hurd : http://www.debian.org/ports/hurd
`- The IPv6 Portal : http://www.ipv6tf.org/
