|
FastMail Forum All posts relating to FastMail.FM should go here: suggestions, comments, requests for help, complaints, technical issues etc. |
|
Thread Tools |
27 Feb 2014, 11:32 AM | #1 |
Member
Join Date: Aug 2003
Location: Oxford, UK
Posts: 78
|
Epic amounts of spam
Is the SpamAssassin problem fixed? Because I don't think it is.
Ever since the "problem" I have been been getting about 10ish spams a day landing in my inbox and not being caught by the filter. I toss them all into SpamCop to see where they're from and they're ALL from burst.net. (Though they all have faked From: addresses.) Here are the X-Spam headers from just the most recent one: Code:
X-Spam-score: 3.6 X-Spam-hits: BAYES_99 3.5, BAYES_999 0.2, RP_MATCHES_RCVD -0.001, LANGUAGES en, BAYES_USED user, SA_VERSION 3.3.2 X-Spam-source: IP='184.22.135.203', Host='carmelpartners.tawfire.com', Country='US', FromHeader='com', MailFrom='com' X-Spam-charsets: plain='us-ascii' Please save me from the never-ending deluge of crappy spam! UPDATE: To pre-emptively answer the question, yes, I am using the "User" spam database, and not global, and I'm marking all these spams as spam in Fastmail as well as sending them to SpamCop. Last edited by Mwongozi : 27 Feb 2014 at 11:36 AM. Reason: Update |
27 Feb 2014, 05:59 PM | #2 |
Master of the @
Join Date: Sep 2004
Posts: 1,583
|
The problem appears to have been with the BAYES_999 rule.
http://emaildiscussions.com/showpost...6&postcount=10 http://www.gossamer-threads.com/list.../183590#183590 FM have changed their settings so I would assume the issue is fixed. If all of your spam is coming from one place, could you not set up a custom rule to discard or at least filter them? |
27 Feb 2014, 09:27 PM | #3 |
The "e" in e-mail
Join Date: May 2002
Posts: 2,804
|
I doubt it's anything to do with BAYES_999, Fastmail rescored that pretty quickly and subsequently Spamassassin has redefined it so it's additional to BAYES_99 - see the above example.
If your database is well trained and BAYES isn't misclassifying ham, I'd set-up a rule to filter the higher BAYES results into the spam folder. |