On Tuesday 23 December 2008 18:14, Shironeko wrote: > Dear Freenet Support Team, > > I send you this message because I've stumbled upon a "curiosity" which I'd > like to get explained since I'm not able to find any other documentation > regarding this issue. > > I was browsing through my hard drive's Freenet Directory, looking at the > latest logs when I suddenly realized that there were IP adresses written in. > > This is an example: > > dic 23, 2008 17:06:14:078 (freenet.node.NodeDispatcher, UdpSocketHandler for > port 266XX(2), NORMAL): Rejecting CHK request from 213.238.213.XX:387XX > preemptively because Insufficient output bandwidth > > I may not fully understand the protocol Freenet uses for data transmission > but these IP's are uplookable and can represent a problem for anyone who > connects from a country like China.
I don't see why. For it to be a problem the bad guys would have to already have seized (or electronically compromised) your node, in which case they probably have your browser history, your datastore, your Friends list ... > > Also, I wonder if it would be possible to collect valuable information by > gathering the LOGs of many different nodes and following a specific IP's > requests. Yes, but you'd need to compromise all the nodes on the path of that request. > > Finally I'd like to ask you about this message I found in the logs too: > > "Note that this version of Freenet is still a very early alpha, and may well > have numerous bugs and design flaws. > In particular: YOU ARE WIDE OPEN TO YOUR IMMEDIATE PEERS! They can eavesdrop > on your requests with relatively little difficulty at present (correlation > attacks etc)." > > I suppose that this must be an old message since the Freenet project is not > in a very early alpha version anymore and I'm using 0.7, the latest. This is partly true. There are a number of known attacks on Freenet, which cannot be completely eliminated short of new features which we have not yet implemented. On the other hand, for some situations, Freenet may be the best currently available. For example, Freenet's scalable darknet functionality is fairly unusual, allowing you to only connect to people you trust, and also it is easier to safely publish a website on Freenet than on a Tor hidden service afaik (due to e.g. issues with configuring apache to not give away incriminating details, and much harder intersection attacks). The bottom line is if you are going to stake your freedom and/or life on the security of an anonymous network, you need to seriously consider the pro's and con's of each possible option, including doing nothing; Freenet has had severe bugs in the past, and is pre-1.0, but apart from that, we have fairly serious known attacks... There are 4 basic powerful attacks on Freenet that we are concerned about: 1. Harvesting. Finding lots of Freenet nodes quickly, in order to e.g. block them on a national firewall. Most anonymous networks do not address this problem at all. On opennet, harvesting is relatively easy (slightly harder than on Tor or I2P); on darknet, harvesting should be fairly hard. 2. Datastore seizure. What happens when/if the bad guys either electronically compromise or physically seize your computer? At the moment everything you download through Freenet is cached in your datastore. Temporary files are encrypted with ephemeral keys, but for long-term downloads we have to store the keys to disk. 3. Snooping on your peers. It is probably possible, under some assumptions (e.g. being able to identify the content, it being sufficiently large), to do statistical attacks to figure out what those nodes you are connected to are downloading/uploading. This is yet another reason to use darknet. 4. Mobile attacker tracing the source of a stream of content. If an anonymous identity publishes data that can be identified (e.g. reinserting known content, posting to FMS boards, posting to a known freesite), it may be possible to gradually approach his location. Reinsertion of known content makes this much easier, because of CHKs; because we always insert the top block (the freesite USK e.g.) last, if the content isn't guessable in advance it is very difficult to pull this off against large inserts, because the attacker can only identify the stream after the top block (or the FMS post referring to the new file) was inserted; if the content *is* guessable, the attacker can move towards the target continually over the course of the insert. All of these attacks we have some mitigation against, but all of them are feasible to some extent under some mostly-reasonable assumptions. Later versions of Freenet will make them much harder with new features e.g. rendezvous tunnels. > > Thank you very much. > > Shiro. > > PD. I also wonder where the cached and encrypted files on my HD are > gathering. In the freenet directory, generally speaking. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 827 bytes Desc: not available URL: <https://emu.freenetproject.org/pipermail/support/attachments/20090105/920715d8/attachment.pgp>
