On Tue, Aug 09, 2011 at 03:54:55PM -0500, Noel Jones wrote:
> On 8/9/2011 2:41 PM, Wietse Venema wrote:
> [snip discussion of minimizing deep protocol test delays]
> 
> 
> What do you think about an option to skip the "after 220"
> tests based on dnswl results?
> 
> If an IP is listed on a dnswl, it's quite likely a real mail
> server and would pass all the disruptive tests eventually.

I've seen one particular ESP which is on DNSWL failing after-220 
tests. Funny thing, it looks like they fixed it today.

Aug  7 05:22:27 harrier postfix/postscreen[8311]: CONNECT from 
[207.45.161.31]:52622 to [207.223.116.211]:25
Aug  7 05:22:27 harrier postfix/dnsblog[8313]: addr 207.45.161.31 
listed by domain list.dnswl.org as 127.0.15.0
Aug  7 05:22:28 harrier postfix/postscreen[8311]: PREGREET 26 after 
1.1 from [207.45.161.31]:52622: HELO mailer2.ixsmail.com\r\n
Aug  7 05:22:28 harrier postfix/postscreen[8311]: NOQUEUE: reject: 
RCPT from [207.45.161.31]:52622: 550 5.5.1 Protocol error; 
from=<[email protected]>, 
to=<rcpt@harvested>, proto=SMTP, helo=<mailer2.ixsmail.com>
Aug  7 05:22:28 harrier postfix/postscreen[8311]: HANGUP after 0.09 
from [207.45.161.31]:52622 in tests after SMTP handshake
Aug  7 05:22:28 harrier postfix/postscreen[8311]: DISCONNECT 
[207.45.161.31]:52622

I'm quite sure this one was spam, because that recipient is in my 
spamtrap list. And the performance is quite spammy, because I have 
given them 550 5.5.1 rejections since May!

# cat >> /etc/postfix/postscreen_access.cidr
# 2011-08-09 eWayDirect whitelisted, but hitting spamtraps
# was having PREGREET protocol errors before today
207.45.161.0/24         reject
^D

> I suppose we could overload the postscreen_dnsbl_threshold 
> parameter for this, something like
> postscreen_dnsbl_threshold = reject-boundary;pass-boundary
> where reject-boundary is required (default 1), pass-boundary
> is optional/no default/unset.
> example    = 1;-1

So with a negative score from dnsblog, it would bypass after-220 
tests? What if, like eWayDirect above, it is a pregreeter? If a 
certain test is already failed, I see no benefit in allowing a client 
to proceed (whitelist it if you want it.)

Otherwise I think it's a good idea. I'd set my pass-boundary at -2. 
I'm only giving those dnswl.org .0's a -1 score.
-- 
    Offlist mail to this address is discarded unless
    "/dev/rob0" or "not-spam" is in Subject: header

Reply via email to