I'll take a stab at some of your questions/clarifications on application
proxy firewalls - specifically the Raptor firewall which is a hybrid
application proxying firewall. As a raptor administrator for the past
decade I think I can address some of the more important points
surrounding this firewall. First, regardless of the network segment
you're scanning and depending on the firewall's configuration, you may
or may not see a large collection of ports "open" which could then
trigger nessus plugins. This firewall allows filters to be placed on the
various interfaces much like a common packet filtering firewall which is
often used to reduce firewall load and exposure (packets dropped before
reaching the proxies) but if no filters are in place then you will see,
at the very least, open ports associated with all 12 of the native
proxies handling the common protocols like http, ftp, telnet, smtp,
nntp, rtsp, dns, ntp, cifs, nbdgram, ping, h323, and other ports handled
by the gsp proxy. Just because the ports appear "open" though doesn't
mean you're talking to anything other than the firewall's proxies. If
rules allow traffic, the native proxies will do their thing which is to
inspect packets for rfc compliance and other payload specific tests and
either accept or reject the traffic. If the traffic is accepted, the
packets are repackaged and passed on to the destination. The difference
between the native proxies and the gsp proxy is no rfc payload testing
occurs with the gsp. Some basic sanity checks are still performed such
as tcp flag settings, src/dst addressing, etc. but that's pretty much it
as I recall.

As Ron suggested, quite a few of the CVEs you listed are associated with
the standard MS networking ports 139, 445 and perhaps 135. Chances are
pretty slim that these ports would truely be opened to the outside
segment without some serious restrictions. But if they were, the cifs
proxy (139,445) is configurable at the rule level to restrict certain
types of cifs connections which could potentially affect test results as
well.

A more typical scenario is to have a web, ftp, or smtp service running
on one of the fw's aux interfaces opened up to the outside in which case
you most likely can make connections to these services either with
nessus or other means and conduct tests, exploits, etc. But
interpretation of the results is not always simple and straightforward
depending on the version of the fw software and other components that
may be licensed on the fw such as AV, IDS/IPS, and Content Filtering.
Even on the earlier versions of the fw software, an administrator could
assign url filtering, for instance, to the http proxy which helped
blocked common exploits - granted it did not handle encoded data but if
you didn't enable nessus' ids evasion features many of the nessus tests
would trigger these filters IF they were enabled. This would show up in
the fw logs as "packet denied matched url filter at line ..." or
something like that and the nessus report would, of course, report no
associated vulnerabilities (packets dropped by fw). Newer versions with
IDS/IPS may block the traffic depending on the configuration. The
IDS/IPS component installs with some default settings blocking the most
aggregious traffic and others are just logged so again you would have to
compare logs to the report.

As Ron mentioned, you'll need to compare results with fw logs and
application logs on the targets to clarify findings. I reported a
vulnerability to the Symantec group years ago involving the http proxy
cause it crashed during some nessus testing I was conducting with a
client. Nessus results didn't reflect the crash and the fw did what it
was designed to do in this situation, respawn the daemon (proxy) so
without reviewing the fw logs we wouldn't have known about this.

Greg Smith, <GSEC Gold>
Cyber Crews, Inc.
_______________________________________________
Nessus mailing list
[email protected]
http://mail.nessus.org/mailman/listinfo/nessus

Reply via email to