On Wed, 30 May 2007, [EMAIL PROTECTED] wrote:
My postfix log summaries keep getting tagged as spam. Here's the header
info:
X-Virus-Scanned: amavisd-new at someserver.com
X-Spam-Flag: YES
X-Spam-Score: 11.415
X-Spam-Level:
***********
X-Spam-Status: Yes, score=11.415 required=5
tests=[AWL=0.651, BIZ_TLD=1.169,
INFO_TLD=0.813,
MAILTO_TO_SPAM_ADDR=0.276, NO_RELAYS=-0.001,
???
URIBL_JP_SURBL=3.36, URIBL_OB_SURBL=2.617, URIBL_WS_SURBL=1.533,
??? URI_NOVOWEL=0.997]
No matter how many
times I train the bayes filter with this message as ham, it makes no
difference.
If these tests:
URIBL_JP_SURBL=3.36, URIBL_OB_SURBL=2.617, URIBL_WS_SURBL=1.533
were out of the picture, it would seem the message would be fine. Those
tests are triggering on some domain in a URL within the body of the
message.
You should be able to use the debug output from 'spamassassin -D' on the
message to figure out what domain URL is triggering. Then again, you may
only have one domain URL in the body or know what the domain is. You can
also look up domains in the URIBL's from here:
http://www.rulesemporium.com/cgi-bin/uribl.cgi
Or, if you have a means of bypassing the spam filtering for these log
summaries, that would be the best way.
OT: I run two instances of Postfix. One that filters and one that doesn't.
Each use their own IP. MX records get pointed to the proper IP.
