Hello Dennis, thank for your comments, they are much appreciated. I hope I understand enough to formulate a valid reply.
On 20/10/11 12:08, Dennis Guhl wrote: > On Thu, Oct 20, 2011 at 10:31:50AM +0200, Daniele Nicolodi wrote: >> On 19/10/11 21:00, Tom Hendrikx wrote: >>> http://wiki.apache.org/spamassassin/IntegratedSpamdInPostfix, whose >>> first lines clearly mention the flaws you're system will run into >>> (generate backscatter, for instance) >> >> I read this document, and I question the validity of this statement. As > > The second paragraph from the above mentioned link is: > > The easiest way to integrate postfix and spamassassin is to > use spamd in an after-queue inspection. This configuration > does not allow rejecting messages within the SMTP transaction, > so it unfortunately contributes to backscatter email. On the > other hand, it has important performance advantages over > before-queue inspection. > > Where do you see room to question Tom's statement? My point is that no backscatter is generated from the use of spamassassin as content filter. The back scatter is generated from the rejection of an email in the content filter. If I explicitly instruct spamc to tell postfix to reject the email based on the spamassassin rules, then, I would have backscatter. My English is maybe not very good, but I find the above statement confusing, because at a first read it may look like that the filtering itself may cause backscatter. >> I understood it, postfix bounces a message only if the content filter >> program return an error. This happens for spamc only if there has been > > What led you to this conclusion? If spamassassin tells postfix to > reject the email, postfix will reject the email. If this email was > accepted by postfix beforehand the rejection will most likely be > backscatter. That's exactly the point. My spamassassin does not tell postfix to reject the email, I simply discard it, without notice to the sender, as I would do using spamassassin with procmail (as other suggested). > To stoer every spam email just to maybe not lose a ham mail is > nonsense. Hardly anybody really search through all the spam to find a > potential ham email. But if I reject at SMTP-level I can be sure that > every legit sender receives a bounce, stating the sent email is > considered as spam and can take appropriate actions. So tagging and > storing spam destroys ham, rejecting spam at SMTP-level does not. Agreed. > The second part of the statement is just frightening. To expect > innocent third parties to filter out my backscatter is just > ridiculous. I agree. That the reason why I do not reject emails in the content filter, but simply discard them. It is no much different than using spamassassin with procmail or similar solutions. What is the recommended way to use spamassassin to reject emails at the SMTP level? A few have been suggested in this thread, but honestly I do not have time to try them all. Simplicity, assuming a low volume smtp server, is highly appreciated. Thank you. Cheers, -- Daniele
