On Sat, 21 Sep 2019, Kenneth Porter wrote: > I've been doing sendmail -> MIMEDefang -> SpamAssassin/clamd and > then sendmail -> procmail -> SpamAssassin. Yeah, SA gets run twice, > once to reject scores > 10 by the milter and then again by each user > to incorporate their Bayes scores. I'd love to run it only once but > haven't invested time in figuring out how to do that. But I only > have a few users so it hasn't been a big enough load to worry about > it. Have you considered running the SpamAssassin Milter? https://savannah.nongnu.org/projects/spamass-milt/ It's available via EPEL. You can reject high-scoring spam during the SMTP transactions. It also allows per-user preferences/Bayes rules to run, with the caveat that a message addressed to multiple users can't take advantage of the per-user run. (Unlike a setup using mimedefang, spamass-milter can't resubmit a message for each addressee.) -- Paul Heinlein heinlein at madboa.com 45°38' N, 122°6' W