On Thu, 2010-01-14 at 23:58 +1100, Jean-Yves Avenard wrote: > 2010/1/14 Matus UHLAR - fantomas <uh...@fantomas.sk>: > > well, you either trust SA on secondary MX - then don't run the mail through > > SA again. > > But not all mails go through the 2nd MX ; so this is exactly what I > want to do: don't run SA if it went through the 2nd MX and was tagged > as spam.
Ahhh... so you want to skip whole scanning process if the mail is already marked as spam? You should then set SA on the first machine so it adds a unique header (and pray that noone starts forging it) and filter on this header _before_ feeding mail to SA on the second machine. How to do this filtering depends on how you feed SA there. In my case it'd just be a simple rule in system-wide procmailrc. And you could also want to rewrite the header, so you can filter on one spam status header instead of two. -- \.\.\.\.\.\.\.\.\.\.\.\.\.\ .\.k...@epsilon.eu.org.\.\. \.http://epsilon.eu.org/\.\ .\.\.\.\.\.\.\.\.\.\.\.\.\.
