Hi, > I am using Spamassassin on our SMTP servers with almost 20000 mails > an hour. The problem is the machine is almost always heavily loaded. > Spamassassin takes a lot of time and I think the Bayes checking / > learning is the real cpu hog ? Depending on how much and how often you're learning, yes, it can be a hog. Lookups aren't expensive but depending on your backend learning can hit you. At 20000 mails an hour, I'd check if you're using auto learning of spam/ham and would seriously consider using a real SQL RDBMS as a backend (MySQL/Postgres).
> I would like to completely disable bayes , can someone provide some > inputs on this. Add the following to your local.cf (usually /etc/spamassassin/local.cf): use_bayes 0 This will turn everything Bayes related. Regs, Sven
