On Tue, 21 Oct 2008, Jim Balo wrote:
From: Jim Garrison <[EMAIL PROTECTED]>
I can highly recommend gray-listing. It's all I use on
two Postfix servers, and SPAM is reduced by 98%. A few
get through, but it's quite tolerable, and I
haven't seen
a false-positive in at least two years.
Hi,
Do you have any recommendation on how-tos on doing this with
Postfix (I know policy-weight is no longer developed, so I
rather not use it)?
I don't have a how-to. However, if you know anything about Perl and MySQL
(or any other SQL backend), it shouldn't be too difficult to conjure
something up.
I have a policy I wrote in Perl for greylisting with a MySQL backend.
MySQL is shared between two servers. There is two cron jobs that run
deleting records where:
o) client IP's have not returned since the initial connection within 24
hours
o) client IP's have returned and are older than four weeks from the
initial connection
The IP of the connecting server is stored in a user selectable format
depending upon how many octets of the IP are recorded.
I have had a few customers and servers that I have whitelisted from
greylisting because of the respective connecting server's inability to
comprehend the SMTP 450 response code in thinking it is a 5xx response
code.
I also have built into the policy a mandatory greylist if any of the MX
records for the sending domain are found to contain a CNAME or an IP
address. MX records are suppose to contain a fully qualified domain name
(FQDN).
If you would like to know more, please see me off-list.
P.s. Even though policyd-weight may be old, I've heard good things about
it. We have a customer that uses it and swears by it.
-d
