Actually this server is hosting lots of different domains that I do not control, I do not choose whether there is a backup MX or not. However I noted that some (most) of those backups MX already had a working spamassassin configuration and that it would be a waste of time to re-scan those mails. I was wondering if there would be some rule like
header MAIL_FROM_BACKUP_MX Received =~ /some known backup mx/ do_not_scan Since I run qmail-scanner I guess the best way for me to achieve this would be to scan the mail before sending it to spamassassin, and if it's spam then feed the bayes system with it myself. On Jeu 26 octobre 2006 18:28, Chris St. Pierre a écrit : > On Thu, 26 Oct 2006, Renaud Drousies wrote: > >>Hello, >> >>Is there a way to prevent Spamassasssin to double-scan the mails that >> have >>already been scanned by the backup MX but still having him to autolearn >>what it receives through it (based on the score given by the backup mx, >>but according to its own auto learn threshold rules)? The goal being to >>save some scan time on the primary MX, since most spams come from the >>backup ones. > > You could have the two MXs share a SQL database with the Bayesian > database in it. That way, the backup could scan and learn, and you > wouldn't have to worry about setting up what you describe, which I'm > pretty sure is impossible anyway. > > Chris St. Pierre > Unix Systems Administrator > Nebraska Wesleyan University > >
