|
Hmmm, actually, now that I think about it, it can
be forged to be whatever the address space owner wants it to be. For
example, if 111.222.111.0 was delegated to me by ARIN or my ISP, I could setup
the RDNS for the IP address to resolve however I want them to. I could
have:
111.222.111.1 PTR yahoo.com.
111.222.111.2 PTR hotmail.com.
111.222.111.3 PTR aol.com.
and messages that were sent from these IP addresses
would meet the following whitelist requirements and get delivered:
WHITELIST REVDNS ENDSWITH
yahoo.com
WHITELIST REVDNS
ENDSWITH hotmail.com
WHITELIST REVDNS ENDSWITH aol.com
even though these are
not legitimate Yahoo, Hotmail, or AOL IP addresses. The RDNS test
simply looks at the IP address of the connecting server (following the HOP and
IPBYPASS setting, of course) and does a reverse lookup on that IP address and
the response will come back however the IP address space owner has configured it
in their in-addr.arpa zone file for the IP
address.
So, even my last post
about using "ENDSWITH" instead of "CONTAINS" with RDNS filters is NOT
infallible since "ENDSWITH" is NOT secure against being forged. However,
if the RDNS test instead queried for the network instead of the IP address,
that would be a much more reliable test since the network owner is registered
with ARIN. For example:
dig mx yahoo.com returns
in part:
yahoo.com.
5853 IN
MX 1
mx1.mail.yahoo.com.
dig mx1.mail.yahoo.com
returns:
mx1.mail.yahoo.com.
940 IN
A
64.157.4.78
dig -x 64.157.4.78 (full IP address query) returns: 78.4.157.64.in-addr.arpa. 983
IN PTR
mta-v22.level3.mail.yahoo.com.
dig -x 64.157.4. (network query)
returns:
4.157.64.in-addr.arpa.
302 IN
SOA hidden-master.yahoo.com.
hostmaster.yahoo-inc.com.
----------
dig mx hotmail.com returns in
part:
hotmail.com.
2123 IN
MX 5
mx2.hotmail.com.
dig mx2.hotmail.com returns in
part:
mx2.hotmail.com.
2007 IN
A
65.54.252.230
dig -x 65.54.252.230 (full IP address query)
returns:
230.252.54.65.in-addr.arpa. 3600
IN PTR
mc6.law1.hotmail.com.
dig -x 65.54.252. (network query)
returns:
252.54.65.in-addr.arpa. 3600
IN SOA ns1.hotmail.com.
dns.hotmail.com.
----------
dig mx aol.com returns in
part:
aol.com. 2585
IN MX 15
mailin-03.mx.aol.com.
dig mailin-03.mx.aol.com returns in
part:
mailin-03.mx.aol.com.
156 IN
A
64.12.137.121
dig -x 64.12.137.121 (full IP address query)
returns:
121.137.12.64.in-addr.arpa. 3600
IN PTR
xh.mx.aol.com.
dig -x 64.12.137. (network query)
returns:
137.12.64.in-addr.arpa. 3600
IN SOA dns-01.ns.aol.com.
hostmaster.aol.net.
Although again not infallible, if a spammer wanted to
forge the RDNS for single IP addresses, the network query could only really
respond with one domain in the SOA. So the "ENDSWITH" filter attribute in
RDNS could only be subject to forgery for one domain, not many like the
individual IP address queries
could.
Thoughts?
Bill
|
Title: Message
- [Declude.JunkMail] WHITELIST REVDNS Kami Razvan
- Re: [Declude.JunkMail] WHITELIST REVDNS R. Scott Perry
- Bill Landry
