SpamAssassin Bayes Frequently Asked Questions

Bayes isn't working for me! Help!: BayesNotWorking

How can I set up a site-wide Bayesian filter?: SiteWideBayesSetup

How can my (site-wide) users feed back mail for the Bayesian learner?: SiteWideBayesFeedback

If a message has been learned incorrectly, what do I need to do to fix it?: FixingBadLearning

Will the learner break if I learn SpamAssassin-modified spam messages?: LearningMarkedUpMessages

How can users forward mail for learning without losing the headers?: ProcmailToForwardMail

How can users train Bayes using an IMAP folder?: RemoteImapFolder

How long is token data retained?: BayesExpiration

Why isn't autolearning working for me? (aka: "autolearn=no"): AutolearningNotWorking

I can't get Bayes running with postfix!: SitewidePostfixBayes

How good is SpamAssassin's Bayes implementation?: BayesAccuracy

Bayes seems to take a long time on certain messages, or on accessing the database: DbFileSleepBug

Why don't my old bayes_* files work on my new server (tie failed)?: DbDumpAndLoad

Will I overtrain Spamassassin if I feed it all my hams and all my spams?

Answer: No. As long as you ensure that everything fed to Bayes is accurately classified, manually verified, then you can only improve Bayes' accuracy by training it on everything.

Running parallel fault-tolerant SpamAssassin servers?: BayesBitMe

How do I migrate from one storage backend to another?: BayesMigration

My SQL Bayes database contains data for users who no longer exist. How do I clear this? BayesSqlClearUsers

How can I tell how many ham/spam messages have been added to the Bayes database?: You can get this information with Sa-learn by using

 sa-learn --dump magic