On 12/11/2009 12:53 PM, Thomas Bolioli wrote:
This is an interesting. You are saying run amavisd as an MTA in between
postfix and the sending MTA to reject the spammy/virus messages inbound
and then send off to postfix everything else for delivery. What happens
to non deliverables? They will make it through the first transaction. Or
does amavisd reference postfix's configuration at all?
Thanks,
Tom
Please don't top-post.
The suggestions are to use an smtpd_proxy_filter (amavisd-new
works well for this), or a milter that supports spamd (google
for "spamd milter" for a hundred choices). Either of these
approaches will allow postfix to reject unknown users and the
filter to reject unwanted mail during the SMTP transaction.
Reading the documentation should answer most of your questions.
http://www.postfix.org/SMTPD_PROXY_README.html
http://www.postfix.org/MILTER_README.html
From an external viewpoint, there is no difference between
using a milter or a postfix smtpd_proxy_filter -- they offer
identical end results. Which one you choose depends more on
what filter software you want to use and which configuration
you're more comfortable with rather than technical merits.
You can also integrate clamav into the filter to reject
viruses. If you do use clamav, I highly recommend the
Sanesecurity add-on signatures to detect phish and spam.
http://www.sanesecurity.com/usage.htm
-- Noel Jones
