Penetration Testing mailing list archives

Re: [PEN-TEST] IP Tunneling over DNS


From: "Dunker, Noah" <NDunker () FISHNETSECURITY COM>
Date: Mon, 11 Sep 2000 14:03:48 -0500

I caught an employee at a customer site using his RedHat workstation to get
back in through the firewall.  He was using "rtelnet", which is a cheesy tcl
(I kid you not) script that connects to a pre-determined IP address's
listening port (you listen on that machine with netcat.  It tries every <nn>
seconds to connect to that port, and when it does, it asks for a passwd.
upon password matching, you can fee dthe firewalled internal machine
commands almost as if you were in a telnet session (anyone ever seen the
port-shell'd /bin/sh in inetd.conf?  It works a lot like that.)...

It's very true, if the firewall allows all sorts of outgoing connections,
and it's statefully inspected, then it's possible for an internal host to
connect out, and ASK for information. letting you "reply", and it accepts
the "reply" as a command.

Even applications such as yahoo messenger do this (when put in
firewall/proxy-less mode).  It requests information via port 80, and queries
the yahoo server every <nn> seconds, asking if there are any new messages or
if any of my friends have gotten online.

The hard part is finding machines that are running programs such as this
one, because of the simple fact that they don't open a listening port.
Programs such as these must be found through passive means (I found the
suspicious machine while sniffing, messed with the router, and assumed the
IP of the machine it was trying to connect to, and discovered it that way.)

After I reported this activity, I'm sure the fellow had a rather bad day...
:)

--Noah Dunker

-----Original Message-----
From: Christopher M. Bergeron [mailto:ChrisB () HGSS COM]
Sent: Monday, September 11, 2000 12:06 PM
To: PEN-TEST () SECURITYFOCUS COM
Subject: IP Tunneling over DNS


I just read an interesting post at slashdot:
http://slashdot.org/article.pl?sid=00/09/10/2230242&mode=thread

theoretically, someone from inside a secure network could tunnel out (ala
Trojan) to punch a major hole through a firewall.  Am I understanding this
correctly?


Current thread: