I agree... ARK servers would be a single point of failure. However, we do still have the various problems with NATs and dynamic IP addresses. With ARKs, if we get one connection we can then find out what the new addresses of our peers are, and tell them our new IP. With STUN, we can determine our IP address by relying on a centralized (but standard, and widely used) service; with UP&P, on the rare occasions where it works, we can determine our IP address from that, *and* forward the port.
The other proposal is rendezvous plugins. We could for example have a Jabber address for a node. This would only be used to send a rendezvous message to in order to exchange IP addresses. Obviously this is centralized, though other rendezvous plugins might not be. (e.g. the DNS technique published in Phrack last year). On Wed, May 03, 2006 at 09:09:21PM +0200, freenetwork at web.de wrote: > >> Dave Baker wrote: > >> > Also, wouldn't connecting to a known ARK server give away the fact that > >> > you > >> > were running Freenet? > >> > >> Agreed - like I said, the ARK server (or someone eavesdropping on it) > >> would be able to harvest addresses. But I'm not suggesting a single > >> central server - there could be any number of servers, each trusted by a > >> few individuals to the extent that they don't mind the server knowing > >> that they're running Freenet, but either they don't trust the server > >> quite enough to make it a darknet peer, or the server doesn't have > >> enough bandwidth to make them all peers. > >> > >> The suggestion was motivated by the fact that static IP addresses are > >> increasingly rare - if you're lucky enough to have one, you could serve > >> the network by helping dynamically addressed peers to find one another. > >> > >> Cheers, > >> Michael > > > >I don't like the idea, it's way too centralised, it's better to do > >everything in a non-centralised way and if needed rely on services > >widely used by joe average so they can't be shut down just like that, > >they'd still have the monitoring problem but it's better than creating > >our own central servers. > > Huh? ARK-Servers?! > > AFAI can remember, ARKs have never been anything else than date-DBR-SSKs, or > in 0.7, probably USKs. What's this "server" talk coming from? > > *totally confused* > > What _servers_?! > > As soon as any "non generic" node is being invented, the freenetproject has > died for me. The strength of the whole system comes from the fact that all > nodes are the same; no single point of failure: every single node has to be > taken down to have the net ceased. -- Matthew J Toseland - toad at amphibian.dyndns.org Freenet Project Official Codemonkey - http://freenetproject.org/ ICTHUS - Nothing is impossible. Our Boss says so. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/tech/attachments/20060503/a0c61baf/attachment.pgp>
