Steve Jordan wrote: > We are using MIMEDefang 2.47 and are interested in using SpamAssassin from > MIMEDefang if it doesn't increase the workload too much. It appears to me > that to use SpamAssassin individual preferences would require the use of > stream_by_recipient() and that would probably increase the workload too > much.
Not in our experience. (That's how our CanIt-PRO product works). Most incoming mail is for only one recipient; streaming the remainder doesn't cause too many load problems that we can see. You need to be a little careful about how you handle the streamed mail, but it boils down to book-keeping. (For example, you probably want to action_discard() rather than action_bounce() streamed mail that you want to reject.) > And we are using Sendmail 8.12.10. We are keeping the individual > preferences in a SQL database. Is there a different way other than to use > stream_by_recipient() to do this or a later version that would help? The problem is inherent in the design of SMTP; it's not a symptom of a particular MTA or particular version of Sendmail. Unless/until an extension to SMTP is developed that allows one response per RCPT after the end of DATA, we're kind of stuck. > Am I correct in thinking that filter_end() is called once per email and > not once per recipient? That's correct. -- David. _______________________________________________ Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list [email protected] http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
