Comments on: Xprobe2 – Active OS Fingerprinting Tool

By: Sifmole

Sifmole — Fri, 09 Jan 2009 17:14:29 +0000

One of my problems with xprobe2 is that the only fingerprinting database I can find is from 11 July 2005. Do any of you know where you can find updated databases?

By: Xnih

Xnih — Fri, 28 Nov 2008 02:47:16 +0000

Darknet: I’ll jump on the bandwagon with Erik, but that is just because his programs uses the DHCP fingerprinting from my own program Satori [ http://myweb.cableone.net/xnih ]. p0f does a decent job of passive os fingerprinting in the area of syn and syn/ack packets, but does nothing with DHCP, ICMP, CDP, IPX/SAP, etc, nor has it been updated in years.

By: Darknet

Darknet — Wed, 21 May 2008 17:41:36 +0000

Erik: I agree, due to the nature of UDP packets they have very little overhead and therefore very little useful info. DHCP packets and anything broadcast by Windows machines is useful. NetworkMiner is a neat tool we have covered it here before, still the best for passive OS fingerprinting is p0f – I haven’t yet covered that here but will shortly. The active OS fingerprinting in nmap is pretty decent too, I haven’t tried any of those features in Unicornscan yet.

By: Erik

Erik — Wed, 21 May 2008 17:09:45 +0000

The only things about active OS fingerprinting is that it can be done from any network as long as the fingerprinted host is reachable. Apart from that I really prefer doing passive OS fingerprinting.

Is there by the way any point in doing OS fingerprinting based on UDP packets? My feeling is that they don’t hold enough information in order to fingerprint a host properly. A much better fingerprinting method is to look at DHCP packets since they reveal a lot about the host. The good thing with DHCP is also that it is broadcast, so it’s perfect for doing passively. In fact the new version of NetworkMiner [ http://networkminer.wiki.sourceforge.net/NetworkMiner ] supports passive OS fingerprinting using both TCP and DHCP fingerprinting.

I haven’t tried unicornscan though, it also seems to have some passive OS fingerprinting functionality.

By: Darknet

Darknet — Tue, 20 May 2008 16:00:34 +0000

Chris: Well it depends how you look at it, it’s very subjective. For me nmap started out as a decent TCP connect scanner and evolved from there, the main featureset is on the TCP side – it’s definitely not a competent UDP like Unicornscan is. Yes Unicorn does TCP as well and it does it better for large networks but I still find nmap better for TCP investigation and trickery (idle scans, decoys, banner grabbing etc). For me saying it should be better at UDP scanning is the opposite of one tool for one purpose.

By: Chris

Chris — Tue, 20 May 2008 10:59:07 +0000

I did? I personally find that nmap just cannot do the job on large networks, i.e /18-. It takes too much time, and there is no way to control how it interacts with its environment.

If we’re talking about portscanning, I include UDP/TCP for that. unicornscan does just that, it does portscanning, nothing more. For that reason I prefer unicornscan over nmap. It’s not however to say I think nmap is bad, for certain things it performs very well. It all comes down to the job, purpose and the tester. The tester should be comfortable with all the tools he decide to use, they are never a replacement for good understanding of x.y.z.

If I contradicted myself, that was not my intention. (And I still fail to see how I did!)

- Chris

By: Darknet

Darknet — Tue, 20 May 2008 09:59:28 +0000

Chris: You kind of went against what you were saying there, for UDP scanning Unicornscan is better…for larger networks either Unicornscan, Scanrand or Advanced LAN Scanner. Or a paired down version of nmap with the right options. Different tools for different jobs.

By: Chris

Chris — Tue, 20 May 2008 09:32:29 +0000

I must politely disagree with Matt here. Personally I don’t use nmap at all, but I very much agree with the unix philosophy of one tool for one type of purpose. nmap is already doing too much, it should in my opinion do less.

Let xprobe2 stay the way it is, and try and slim nmap down a bit. What they should do with nmap is improve their UDP-scanning. Integrate a payload system, like the one found in unicornscan. Also they should stop doing automatic flow-control. It just messes up badly on larger networks. You need to be able to calculate efficiency and time.

My two cents.

By: Jinesh

Jinesh — Mon, 19 May 2008 13:39:08 +0000

Words are not enough to describe this tool.

By: matt

matt — Fri, 16 May 2008 21:25:32 +0000

Sounds like a very useful tool. Would be nice to see all of these features combined with nmap. …maybe a new project for myself. Thanks for the article!