I2P dev meeting, February 24, 2004 @ 22:00 CET

Quick recap

  • Present:

baffled, DrWoo, duck, human, Janonymous, jrand0m, kaji_, madman2003, mihi, Trix, wilde,

Full IRC Log

[22:00] <jrand0m> 0) hi
[22:00] <jrand0m> 1)
[22:00] <jrand0m> 2) docs
[22:00] <jrand0m> 3) ???
[22:00] <jrand0m> 0) hi
[22:00] <human> hi
[22:00] * jrand0m waves to the newly-restarted iip-ircd :)
[22:01] <jrand0m> (and, uh, to you :)
[22:01] <jrand0m> weekly status notes (that we're running off) located at http://i2p.dnsalias.net/pipermail/i2p/2004-February/000148.html
[22:01] <jrand0m> (and posted to the mailing list, obviously)
[22:01] <jrand0m> 1)
[22:02] <jrand0m> Dev has been making some good headway on the 0.2.5 release, which will allow and exploit both 2+ hop tunnels and clients with multiple inbound tunnels
[22:03] <jrand0m> the key functionality that will provide will be increased reliability and functional anonymity
[22:04] <jrand0m> the 1-hop tunnels we have now exposes you to statistical attack by an active opponent, but with 0.2.5 you'll be able to determine the length of your own hops (and increasing the default to 2) making the statistical attack much more complex
[22:06] <jrand0m> i also found a pair of bugs in the client send process and the network db which could account for some of the latest instability
[22:06] <jrand0m> (bugfixes underway)
[22:07] <jrand0m> as an aside, I think the roadmap [http://wiki.invisiblenet.net/iip-wiki?I2PRoadmap] is still an accurate reflection of the dev schedule
[22:07] * mihi has joined #i2p
[22:07] <jrand0m> heya mihi
[22:07] * protocol has quit IRC (Ping timeout)
[22:07] <jrand0m> ok, thats it for the router dev status, moving on to 2) docs
[22:07] * human would like to say that he finds I2P unusable since 0.2.4 (it seems to behave *far* worse than the 0.2.3 age, at least on my PC)
[22:07] <jrand0m> hm
[22:08] <human> maybe we could talk about it after the meeting...
[22:08] <jrand0m> in reliability terms, latency, CPU, bandwidth?
[22:08] * protocol has joined #i2p
[22:08] <mihi> hi jrand0m, hi all
[22:08] <human> jrand0m: i can't reach any eepsite or I2P service (with few temporary exceptions)
[22:08] * mihi seconds human
[22:09] <jrand0m> most eepsites are down - duck, baffled, madman2003 are the most consistently up lately
[22:09] <human> jrand0m: i can't reach them, nor irc.*.i2p
[22:09] <jrand0m> squid I use constantly for all my web browsing - are you unable to use that?
[22:09] <human> jrand0m: nope
[22:09] <jrand0m> hm
[22:10] <madman-away> well an uptime of about 8 hours daily for my i2p site
[22:10] <human> jrand0m: we could talk about it after the meeting, i don't want to monopolize the discussion :-)
[22:10] * madman-away is now known as madman2003
[22:10] <mihi> it might be my provider's problem as well, http://babelfish.altavista.com/babelfish/urltrurl?tt=url&amp;url=http%3A%2F%2Fwww.expressnet.de%2Fnews%2Fnews.php&amp;lp=de_en :(
[22:10] <Janonymous> me too
[22:10] <human> jrand0m: just to point out that some problems seems to exist (that weren't shown before)
[22:10] <Janonymous> mine should be up now
[22:11] <madman2003> what destination?
[22:11] <jrand0m> well, 'k, if things are going backwards in reliability we need to address that before moving on to 0.2.5
[22:11] * mihi hates babelfish's english :(
[22:11] <jrand0m> heh
[22:12] <jrand0m> (well, it got the Thank you for your understanding. sentence correct at least...)
[22:12] <human> jrand0m: it should be investigated... i thought about an ISP issue, too, but the problem seems to be constant since 0.2.4 (and doesn't seem to happen with other network services)
[22:12] <jrand0m> was shit, as was
[22:12] * wilde has joined #i2p
[22:13] <human> jrand0m: i know, and it worries me...
[22:13] <wilde> hey i2p
[22:13] <jrand0m> heya wilde
[22:13] <madman2003> one thing i noticed is that tunnels tend to get unstable more often
[22:13] <human> jrand0m: i didn't change too much with (at least for me)
[22:13] <human> jrand0m: s/ i / it /
[22:14] <jrand0m> madman2003> thats easily due to routers going on and offline (which will be a big problem until 0.3)
[22:14] <jrand0m> hmm ok
[22:14] <wilde> jrand0m: does that mean we should avoid running transients for now?
[22:15] * mihi has quit IRC (Ping timeout)
[22:15] <jrand0m> hm, I think there are going to be significant fixes in 0.2.5, but we can hold off on moving from 0.2.5 to 0.3 until after the reliability issues are cleared.
[22:16] <jrand0m> wilde> i don't like the term transients, it makes me think of another project that treats unreliable routers differently than reliable ones.  we treat all routers equally (and need to, for anonymity)
[22:16] <jrand0m> but as long as routers generally stay up or generally down, they're fine
[22:17] <jrand0m> (just not up 10 minutes, down 10 minutes, up 30, down 30, etc)
[22:17] <madman2003> i do have one request: an option for the router(and tunnels) to be reastablished
[22:17] <baffled> i2p is an equal router opertunity organization?
[22:17] <jrand0m> heh baffled
[22:18] <jrand0m> madman2003> router to be reestablished?  your router is shutting down still?
[22:18] <madman2003> i mean reconnecting everything
[22:18] <madman2003> sort of a warm restart of the router
[22:18] <madman2003> without pissing of the other routers
[22:18] <madman2003> (i have to restart router and tunnels a lot)
[22:18] <jrand0m> you can safely restart your client apps (e.g. i2ptunnel eepproxy, etc) without touching the rotuer
[22:19] <jrand0m> you should /never/ need to restart your router.
[22:19] <jrand0m> (almost all config settings are updated dynamically)
[22:19] * Trix has joined #i2p
[22:19] <jrand0m> hi Trix
[22:19] <Trix> hi
[22:19] <madman2003> usually restarting the tunnels does the trick
[22:21] <jrand0m> there's only one situation where that's technically necessary (old lease expirations in a client's leaseSet, which occurred on startup randomly), and thats been fixed in CVS, so you shouldn't need to do that.
[22:22] <jrand0m> (in fact, restarting tunnels can cause temporary problems, depending on the type of tunnel)
[22:22] <madman2003> sometimes i just don't know if i'm causing problems or if someone else is
[22:22] <jrand0m> if your router console doesn't have any of the red warnings, its the network (or someone else)
[22:22] * jnk has joined #i2p
[22:23] <jrand0m> patience fixes more of the current i2p bugs than restarts do ;)
[22:24] <jrand0m> but we'll have another series of bugfix releases after 0.2.5
[22:24] <jrand0m> (like testnet, except without the restrictions on the userbase)
[22:25] <jrand0m> (and, as always, whenever things break, logs are appreciated :)
[22:25] <jrand0m> anyway, moving on to 2) docs
[22:26] <jrand0m> as posted in http://i2p.net/pipermail/i2p/2004-February/000147.html there've been some new overview docs
[22:27] <jrand0m> I'd appreciate some critiques to improve them, as they referred to pages are essentially going to turn into the main starting point for learning about I2P
[22:28] <madman2003> i read them and i hope you were right about the possibilty to safely reduce the amount of crypto
[22:29] <wilde> layers of crypto or # of bits?
[22:29] <jrand0m> I'm not convinced that the crypto is the bottleneck, but its a possibility
[22:30] <jrand0m> we couldnt safely reduce the layers, but we could use different levels of crypto at different layers, rather than reusing the same code for everything
[22:30] <madman2003> the problem is finding your way
[22:30] <jrand0m> hmm?
[22:31] <madman2003> a static path is usually well found
[22:31] <madman2003> a more dynamic one is more difficult to establish
[22:32] <madman2003> (i'm talking about the inability to properly handle routers going offline)
[22:32] <jrand0m> ah, thought you were talking about crypto
[22:32] <jrand0m> its going to be fairly easy to handle unreliable routers, its just the 0.3 code
[22:33] * jrand0m has ~30 pages of notes on different techniques, its all workable, just lots to do
[22:33] * protocol has quit IRC
[22:34] <madman2003> maybe an idea to have backup routes ready
[22:34] <madman2003> tunnel redundancy
[22:34] <jrand0m> right, thats 0.2.5 - multiple leases
[22:35] <jrand0m> (lease == declaration that a destination can be reached through a specific tunnel)
[22:35] <madman2003> i'll be awaiting that :)
[22:36] <jrand0m> w3rd
[22:37] <jrand0m> well, if anyone has any suggestions for improving the docs, feel free to hit the wiki, post to the list, or send me an email
[22:38] <jrand0m> ok, moving on at a rapid pace to 3) ???
[22:38] <jrand0m> anything people want to bring up and discuss?
[22:39] <DrWoo> potatoes are fucking cheap yet potatoe chips are expensive, what's up with that?
[22:39] <DrWoo> :)
[22:39] <jrand0m> its a conspiracy!
[22:40] * DrWoo thinks jrand0m has the answer for most anything :)
[22:40] <jrand0m> of course, you can blame anything on conspiracies.
[22:40] <jrand0m> ok
[22:40] <wilde> Stego ?
[22:40] * human accepts suggestions about how to expose I2CP-like message-oriented functionality to non-java apps
[22:41] <wilde> how I2P will implement Stego so an ordinary portscan will reveal nothing 
[22:41] <wilde> not even random bytes
[22:41] * human may (does?) sound repetitive... he's thinking about VPNs over I2P with http://openvpn.sf.net/
[22:41] <jrand0m> well, for one, PHTTP.
[22:42] <jrand0m> openvpn does look very interesting - I hadn't realized tun/tap had windows ports
[22:42] <jrand0m> a simple message oriented socket bridge for I2CP should be very easy
[22:43] <wilde> Isn't freenet calling it Silent Bob, when the node shuts up it you don't give the secret knock (know the router ID)
[22:43] * madman2003 has quit IRC (12( www.nnscript.de 12:: NoNameScript 3.8 12:: www.XLhost.de 12))
[22:43] <baffled> okay as usual I have another appointment in 15m so I'll catch up later.
[22:43] <jrand0m> right, if we wanted to integrate with a webserver/etc to silent bob, we could
[22:43] <jrand0m> cool, later baffled
[22:44] <jrand0m> (but silent bob doesnt prevent portscan detection, it just makes it look like another service)
[22:44] <wilde> I rather not have random people or ISP:s portscan me and find ports open
[22:44] <wilde> ok
[22:44] <human> jrand0m: ok, i'll work on it when I2P will work again on my PC :-)
[22:44] <jrand0m> UDP would work as well
[22:44] <jrand0m> :) human
[22:45] * kaji has joined #i2p
[22:46] <jrand0m> I hadn't realized reliability had gone down that bad, we'll go through sufficient iterations after 0.2.5 to get it back for you human
[22:46] <wilde> is there a way of hiding an open port from java program, without messing with the OS or firewall
[22:46] <human> w00t!
[22:47] <jrand0m> you mean to have a listening TCP socket that can't be portscanned?  no, not directly from Java.
[22:47] <wilde> ok
[22:48] <jrand0m> (i dont even know how to do that in other langs)
[22:48] <jrand0m> udp would probably be the best way to go for that
[22:48] * human invites people to try to telnet human.i2p (tunneled TCP echo server) and type something in
[22:48] <wilde> that would be a little C program filtering and forwarding to another port maybe
[22:49] * kaji_ has joined #i2p
[22:49] <jrand0m> if it accepts TCP connections, its already too late, if I understand your concern correctly.
[22:49] <Janonymous> how do you telnet?
[22:49] <kaji_> finaly
[22:50] <kaji_> that took forever, iip usability sucks dick atm
[22:50] * kaji has quit IRC (Ping timeout)
[22:51] <duck> if you are concerned about open ports, you could use rTCP / PHTTP / whatever couldnt you?
[22:51] <Janonymous> damn...  hey, jr, are those new docs accessable from the main i2p page?
[22:51] <human> Janonymous: java -jar lib/i2ptunnel.jar -nogui -e "config localhost 7654" -e "client 12221 human.i2p"
[22:51] <jrand0m> no Janonymous, just from the links on that email
[22:51] <human> Janonymous: then telnet localhost 12221
[22:51] <jrand0m> duck> right
[22:52] <Janonymous> k
[22:52] <duck> (ofcourse whatever can be a silentbob/stealth/stego transport)
[22:52] <jrand0m> human> Message send failed after 61226ms with 391 bytes
[22:53] <human> jrand0m: and it means that...?
[22:53] <jrand0m> that means I cant reach your echo
[22:53] <duck> -nogui is depricated :)
[22:53] <jrand0m> can you reach duck.i2p?
[22:54] <wilde> scary, i googled for rtcp: http://dret.net/glossary/rtcp
[22:54] <jrand0m> right, rtcp is taken :/
[22:54] <human> jrand0m: i'm trying, but i can't reach duck.i2p since a looong time ago...
[22:54] <jrand0m> wilde> http://wiki.invisiblenet.net/iip-wiki?RelayingTCP
[22:54] <wilde> "Wilde's WWW Online Glossary"
[22:55] <jrand0m> hehe oh yeah :)
[22:55] <jrand0m> human> thats a definite Bad Thing, as its up almost always - could you bounce me your log-*.txt?
[22:56] <human> Started on:  Tue Feb 24 10:21:22 GMT 2004
[22:56] <human> Version: Router: / SDK:
[22:56] <human> Bandwidth used: 56096295 bytes sent, 34308394 bytes received (avg 1.44KBps sent 0.88KBps received)
[22:56] <human> jrand0m: ok, logs coming on meshmx
[22:56] <jrand0m> gracias
[22:56] <jrand0m> ok, anyone have anything else to bring up?
[22:58] * jrand0m winds up
[22:58] * jrand0m *baf*s the meeting closed