Matt would certainly be able to explain it better than I, but I'll give it a shot.
I can't dig up previous discussions (circa July 2004) on the mail-archive either...
 
A regular ip4r test will test all all of the IP address hops (up to the number defiend by HOPCOUNT).
Unless it has DYNA/DUL/DUHL in the name, in which case it just checks the most recent hop (also called the first hop).
 
One thinking at the time was to use a DYNA for the tests to weight the most recent hop at 80% of the test weight and to use a ALL to weight all hops at 20%. I use this to reduce false positives.
For example in October,
My SORBS-FORMMAIL-ALL all had 434 results that fell in my non-spam weights.
The SORBS-FORMMAIL-LAST has 10 that rsults that fell in my non-spam weights.
So by breaking the tests apart, I'm helping minimize false positives.
 
DUL space tests should have DUL/DYNA in their name to use the previous hop only as should Whitelist tests link Bonded Sender.
 
Now the LAST test uses a dnsbl test in place of the ip4r test.
I believe the advantage here is that in cases where your domain names (maybe it was IP address...) are being spoofed, the dnsbl test will run whereas the ip4r test would not be run.
 
Sample out of my global.cfg:
SORBS-HTTP-LAST  dnsbl %IP4R%.dnsbl.sorbs.net  127.0.0.2 0 0
SORBS-HTTP-ALL  ip4r dnsbl.sorbs.net   127.0.0.2 0 0
SORBS-SOCKS-LAST dnsbl %IP4R%.dnsbl.sorbs.net  127.0.0.3 0 0
SORBS-SOCKS-ALL  ip4r dnsbl.sorbs.net   127.0.0.3 0 0
SORBS-MISC-LAST  dnsbl %IP4R%.dnsbl.sorbs.net  127.0.0.4 0 0
SORBS-MISC-ALL  ip4r dnsbl.sorbs.net   127.0.0.4 0 0
SORBS-SMTP-LAST  dnsbl %IP4R%.dnsbl.sorbs.net  127.0.0.5 0 0
SORBS-SMTP-ALL  ip4r dnsbl.sorbs.net   127.0.0.5 0 0
SORBS-SPAM-LAST  dnsbl %IP4R%.dnsbl.sorbs.net  127.0.0.6 30 0
SORBS-SPAM-ALL  ip4r dnsbl.sorbs.net   127.0.0.6 10 0
SORBS-FORMMAIL-LAST dnsbl %IP4R%.dnsbl.sorbs.net  127.0.0.7 0 0
SORBS-FORMMAIL-ALL ip4r dnsbl.sorbs.net   127.0.0.7 0 0
----- Original Message -----
Sent: Tuesday, November 02, 2004 9:10 AM
Subject: RE: [Declude.JunkMail] Earthlink Porn Spam

Matt,
Thanks -- didn't see that.
Can you post your (ALL) and (LAST) global configs and a brief explanation of how you're using them?
 
Thanks
 
Mark


From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Matt
Sent: Tuesday, November 02, 2004 9:31 AM
To: [EMAIL PROTECTED]
Subject: Re: [Declude.JunkMail] Earthlink Porn Spam

Mark,

mail-archive.com converted the text attachment to just a part of the message if you wish to cut and paste it from there.

    http://www.mail-archive.com/declude.junkmail%40declude.com/msg21757.html

Matt



Mark E. Smith wrote:
Matt,
Can you resend that filter? I checked on the archive and the attachment isn't there.
Thanks.
 
Mark


From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf Of Matt
Sent: Tuesday, November 02, 2004 12:36 AM
To: [EMAIL PROTECTED]
Subject: Re: [Declude.JunkMail] Earthlink Porn Spam

Danny,

It's a special construct that I use to kludge a way to provide a difference in scoring of last hop DNSBL hits and prior-hop DNSBL hits.  For instance, if you score a test on the last 3 hops and it hits an open relay type of list on the first hop, that isn't anywhere nearly as indicative of spam as a last hop open relay hit.

With Declude, you can kludge it so that you can score both the last hop only or all hops.  If I get a hit for both SPAMCOP(ALL) and SPAMCOP(LAST), this means that SpamCop hit minimally on the last hop.  If I only get a hit for SPAMCOP(ALL), that means that the hit was on a prior hop.  Yes, this is most definitely very effective, and I absolutely do wish there was a better way to do this in Declude by assigning the range of hops to test per entry in your config.  An example of how to configure this with SpamCop would be as follows:

    SPAMCOP(LAST)        dnsbl    %IP4R%.bl.spamcop.net            127.0.0.2    4    0
    SPAMCOP(ALL)        ip4r    bl.spamcop.net                127.0.0.2    2    0

This is primarily effective with DNSBL's that track primarily open relays and not necessary with most static spammer lists although SBL has been acting like idiots as of late and including random blocks all the way up to whole class B's on residential class networks which severely weakens the value of SBL when scored the same on every hop.

As far as my filter goes, you can remove all of the lines beginning with the one targeting SNIFFER hits.  It will work just fine without these, but I included them just for good measure as I expect the spam patterns to change eventually.  I do of course expect to see spammers cracking AUTH with much more frequency, and Earthlink at least appears to be inept at stopping it since this has been happening for over 3 months now and growing in scope.

Matt






Danny K wrote:
Matt,
 
What does the (ALL) do as in "SPAMCOP(ALL)"? 
 
 
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Matt
Sent: Monday, November 01, 2004 1:47 PM
To: [EMAIL PROTECTED]
Subject: Re: [Declude.JunkMail] Earthlink Porn Spam

i360 Support wrote:
I am still getting a ton of porn spam from Earthlink.
I report it but it does not help much.
 
Any suggestions on how to stop this crap?
 

Attached is the filter that I use to kill this stuff.  Last I checked, there were two different spammers that were cracking AUTH to get this stuff through, and their patterns don't seem to have changed, although they probably will and/or more will come.

Matt
-- 
=====================================================
MailPure custom filters for Declude JunkMail Pro.
http://www.mailpure.com/software/
=====================================================

-- 
=====================================================
MailPure custom filters for Declude JunkMail Pro.
http://www.mailpure.com/software/
=====================================================

-- 
=====================================================
MailPure custom filters for Declude JunkMail Pro.
http://www.mailpure.com/software/
=====================================================

Reply via email to