Wed to Fri are cancelled, with pay.

Look at this new reject message:

Sep  2 18:46:34 xxx postfix/smtpd[9094]: A478753520: reject: RCPT from 
adsl-157-99-134.clt.bellsouth.net[66.157.99.134]: 554 
<adsl-157-99-134.clt.bellsouth.net[66.157.99.134]>: Client host rejected: 
ACL The PTR hostname "adsl-157-99-134.clt.bellsouth.net" does not match the 
HELO hostname.; from=<[EMAIL PROTECTED]> to=<[EMAIL PROTECTED]> proto=SMTP 
helo=<compuserve.com>

This loser is really in deep sh!t with me:

1. it's a subscriber net

2. the bellsouth ADSL line says helo as compuserve.com

3. and says MAIL FROM [EMAIL PROTECTED]

blocking by 1. is easy enough to understand, but blocking 2 and 3 aren't. I 
do them with restriction classes.

a. if helo hostname is <somebigISP>, then the PTR hostname must be from the 
same.

b. if the @sender.domain is <somebigISP>, then the PTR hostname must be 
from the same


b. is what did the above reject.  It catches 1000's / day.

======================

btw, I tried another restriction class for this policy:

if (PTR hostname doesn't exist)  _AND_ (helo hostname is unfindable in 
DNS), then reject, else accept.

The reject on missing PTR hostname alone causes too many false positives, 
so I figured if the jerk at least got the HELO hostname right, I'd accept 
his mail.  both wrong = plonk.

But, no, there are still a lot of false positives where legitmate, 
card-carding jerks have no PTR hostname and HELO name is not findable in 
DNS.  Anyway, with warn_if_reject, this filter should be good for 
harvesting the true positives manually.

=================

another restriction being evaluated says:

if PTR hostname is <bigISP>, then the mail from: sender.domain must be from 
same domain.

eg, I refuse mail from: @hotmail.com when send from AOL PTRs.


Len


Reply via email to