On Tue, Jun 20, 2006 at 03:12:20AM -0400, Colin Davis wrote: > As a lowly user, I've not found UP&P to work well for my, on any > network I've worked with it on. > Again, I know my opinion doesn't count for much, but it seems like it > wouldn't be as helpful as other tasks. > > I can see two additional benefits, however. > When combined with a delay in Keepalive/Handshaking for properly > port-forwarded nodes, this could allow more bandwidth to be used for > actual traffic, rather than infrastructure.
As I have stated on numerous occasions, the handshake traffic is not a serious problem. What is a serious problem is people adding hundreds of nodes to their refs via scripts that trawl #freenet-refs. Only if you have hundreds of refs will the bandwidth used by handshake traffic be a problem. But we warn the user with a relatively small number of refs as a large number of disconnected peers strongly implies you've been adding refs from people who haven't added you. > > This would increase the number of users who are accessible to those > who do not know their IP address. (These users can only access > someone who is properly port forwarded) 90%+ of nodes don't know their IP address when first installed, because they are behind a router/NAT/PAT. UP&P and STUN let us find it. The problem with STUN is that it can probably be fingerprinted. Additionally UP&P lets a NATed node which doesn't know its IP address connect to another node. Right now a new NATed node which doesn't know its IP address can only connect to a port forwarded node - this is a serious limitation to connectivity! > > -Colin > > On Jun 19, 2006, at 10:19 PM, Matthew Toseland wrote: > > >Do we need to implement UP&P support? It would help in many areas: > >- It would allow us to forward ports and detect our real IP address! > >- It would make connection work more reliably on dynamic IPs, > >especially > > with nodes with poor uptime. > >- It would expand the range of nodes which can be seednodes on > >opennet. > > (To be a seednode you need to be directly connected or port > > forwarded). > >- It would allow us to implement something like the distribution > > servlet. > >- It would allow us to usefully implement support for > >"invitations", one > > use darknet references which come with authorization to add the > >other > > side. > > > >Unfortunately: > >- It is grossly insecure if run on a LAN with untrusted users. We > >would > > have to ask the user during setup. > >- It is blocked by default on Windows XP SP2. > >- Stats on another p2p app which supported UP&P showed it only working > > successfully about 50% of the time even when detected... That may > >have > > been bugs in their implementation of course... > > > >So is it a panacea or a nightmare? Do we want UP&P support? > >-- > >Matthew J Toseland - toad at amphibian.dyndns.org > >Freenet Project Official Codemonkey - http://freenetproject.org/ > >ICTHUS - Nothing is impossible. Our Boss says so. > >_______________________________________________ > >Tech mailing list > >Tech at freenetproject.org > >http://emu.freenetproject.org/cgi-bin/mailman/listinfo/tech > > _______________________________________________ > Tech mailing list > Tech at freenetproject.org > http://emu.freenetproject.org/cgi-bin/mailman/listinfo/tech > -- 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/20060620/0719e14d/attachment.pgp>
