I'm not sure my bayes database is working correctly. How can I fix it?

There are two common problems that may arise with the bayes database: Too large a database, and bayes database poisoning.

1.) Too large a database can lead to SpamAssassin not being able to load the database and use it. Sometimes bayes expiry will fail and this will cause the bayes database to grow very large. This will affect the accuracy of spam identification. You can check the status of your bayes database in WHM > ConfigServer MailScanner Front-End > MailControl > Menu > SpamAssassin Bayes. The number of tokens represents the size of the database.

If your bayes database is very large (this will vary from server to server so it is not possible to give any guidelines) and/or the last expiry date is not in the last 24-48 hours, then you should WHM > ConfigServer MailScanner Front-End > MailControl > Menu > MailControl Settings > MailControl Database Tools > and click the button "Repair SpamAssassin Bayes database" (Fix Bayes).

If the database is quite large this may take several minutes.

To delete the bayes database and start over, go into WHM > ConfigServer MailScanner Front-End > MailControl > Menu > MailControl Settings > MailControl Database Tools > and click the button "Create a new SpamAssassin Bayes database" (New Bayes)".

2.) With Bayes database poisoning, the bayes database starts identifying obvious spam as low-probability and therefore gives it a negative score which may cancel out any positive SpamAssassin test scores and allow the spam to be delivered. In this case, the only thing to do is to remove the database and start over. Go into WHM > ConfigServer MailScanner Front-End > MailControl > Menu > MailControl Settings > MailControl Database Tools > and click the button "Create a new SpamAssassin Bayes database" (New Bayes)".

After deleting and recreating the bayes database as above, SpamAssassin will automatically start building up the bayes database again based on emails identified as spam and not-spam (over a score of 12 for spam and under the score of 0.1 for non-spam). It will start using the bayes database in scoring once the number of spam and ham emails inspected has reached 200 for each category. How long this will take depends on how much email traffic goes through your server.