Luke Shannon wrote: > Good Morning; > > I have SpamAssassin 2.55 running on my server with a required hits > setting of 2.5. Already my spam is down about 60 messages a day. > > I am still getting about 40 a day. This is still a 4-1 spam ratio for > me. > > How can I improve this? > Your upgrade plain is a good starter. SA 2.55's mime parser is actually vulnerable to a remotely exploitable DoS attack.
http://cve.mitre.org/cgi-bin/cvename.cgi?name=CAN-2004-0796 > I'm going to upgrade to 3.1.7 for starters. Should I look at > implementing a Bayesian system? Yes, I would look at using SA's built-in bayes. Try to grab a handful of spam and nonspam emails to use as starter training for sa-learn. Also, if you choose to use a single site-wide bayes DB, read the docs on bayes_path VERY closely. It's not a path, but a path plus partial filename. Also be sure to use 7's not 6's in your bayes_file_mode, as it can be used to create temp directories. > Do I focus on adding patterns to the > *.cf files? Are there patterns out there I can get? > Other than using sa-update to update the official rules, I would not focus on adding on rules until you've had a chance to see how the system behaves. from there, you can information on add on rulesets at www.rulesemporium.com, and http://wiki.apache.org/spamassassin/CustomRulesets
