Score:0

Spamassassin not scoring correctly

ao flag

I have Spamassassin running on a Ubuntu/Plesk server. It was working relatively well until recently but now seems to be letting through random spam messages.

These messages seem to trigger plenty of rules but the score is still coming in really low. I have lots of custom scores defined and they are very much on the high side because I don't mind a few false positives.

Here is and example of an email header run through Spamassassin that seems to clearly be spam:

X-Spam-Status No, score=0.2 required=6.0 tests=BAYES_50,BODY_SPAM, HTML_MESSAGE,MIME_HTML_ONLY,RCVD_IN_DNSWL_HI,RCVD_IN_MSPIKE_H2, RCVD_IN_SBL,SPF_HELO_NONE,T_SPF_PERMERROR,URIBL_BLOCKED autolearn=no autolearn_force=no version=3.4.4

BODY_SPAM alone has a score of 2.0 as does BAYES_50

When I run spamassassin -D --lint it all seems fine. So I'm assuming my cf file is OK

Would be good to find out what causes this and if not at least how do I test or log Spamassassin.

Score:1
ar flag

RCVD_IN_DNSWL_HI and RCVD_IN_MSPIKE_H2 are normally considered negative indicators for spam, and carry a negative score. RCVD_IN_DNSWL_HI has a default score of -5.

So ending with a score of 0.2 sounds reasonable.

Chris Leather avatar
ao flag
This would explain it. The list of default scores I have been using says RCVD_IN_DNSWL_HI gets a 0.0... Is there any way of showing all the scores implemented by Spamassassin.
vidarlo avatar
ar flag
https://spamassassin.apache.org/old/tests_3_3_x.html
I sit in a Tesla and translated this thread with Ai:

mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.