Bugtraq mailing list archives

Re: UPDATE: [wcolburn () nmt edu: SMTP relay through checkpoint firewall]


From: Mike Benham <moxie () thoughtcrime org>
Date: Tue, 19 Feb 2002 14:50:13 -0800 (PST)


People use the CONNECT method from inside a LAN to make SSL/HTTPS
connections through a proxy.  I think it makes sense for proxies to
support the method by default, since browsing secure pages is very
common, but it shouldn't be accessable from outside the LAN.

- Mike

--
http://www.thoughtcrime.org

On Tue, 19 Feb 2002, Steve VanDevender wrote:

It's not just Checkpoint Firewall that has a problem with HTTP CONNECT.
From what I can tell default installations of the CacheFlow web proxy
software, some Squid installations, some Apache installations with
proxying enabled, and some other web proxy installations I haven't
identified allow anyone to use the HTTP CONNECT method.  This is being
used more and more often to relay spam.  This is a boon for spammers
because unlike open SMTP relays which usually record some kind of useful
Received: header, open web proxies don't put any information in the mail
headers about the real origin of the spam.

For those of you unfamiliar with the details of this problem, unsecured
web proxies allow a remote user to use the HTTP connect method to make
arbitrary TCP connections to a specified host and port, like this:

$ telnet open.web.proxy.org 80 # or 8080, or maybe other ports
Trying 192.168.1.1...
Connected to 192.168.1.1.
Escape character is '^]'.
CONNECT victim.host.org:25 HTTP/1.0

HTTP/1.0 200 Connection established

220 victim.host.org ESMTP Sendmail 8.11.6/8.11.6; Tue, 19 Feb 2002 14:16:51 -0800 (PST)

I went around with someone at CacheFlow about this after unsecured
proxies in the cacheflow.com domain were used to relay spam, and after
seeing spam come from various unsecured CacheFlow proxies around the
Internet.  Their position is that this is supposed to be prevented by
putting the CacheFlow server behind a firewall, or using configuration
options in the CacheFlow software to prevent connections to unwanted
destination ports.  They seemed unreceptive to the idea of shipping a
CacheFlow configuration that did not allow CONNECT by default.



Current thread: