I2P dev meeting, December 16, 2003 @ 22:00 CET

Quick recap

  • Present:

duck, FireRabbit, jrand0m, lonelynerd, mids, mihi, MrEcho, protocol, TC, wiht,

完全な IRC ログ

[22:04] <jrand0m> 0) hi 
[22:04] <jrand0m> 1) iip 
[22:04] <jrand0m> 2) 0.2.3 &amp; 
[22:04] <jrand0m> 3) hi 
[22:04] <jrand0m> 0) hi 
[22:04] <jrand0m> welcome to the ... something'th meeting 
[22:05] <jrand0m> (68?  69?) 
[22:05] <MrEcho> damm its 1pm here
[22:05] <jrand0m> GMT-8? 
[22:05] <duck> 69
[22:05] <jrand0m> h0t. 
[22:06] <jrand0m> ok, 1) iip 
[22:06] *** Signoff: tusko (EOF From client)
[22:06] * MrEcho compiles a kernel for the meeting
[22:06] <jrand0m> iip is acting crazy.  all i know is nop is "moving servers", whatever that means.  i don't know when it'll be done, etc. 
[22:06] <jrand0m> anyone have any more info they want to share with the class? 
[22:06] *** mids (mids@anon.iip) has joined channel #iip-dev
[22:06] <MrEcho> no info from nop
[22:07] <mids> this morning I was told that I could start Trent again
[22:07] <mids> (I did do so already last night)
[22:07] <jrand0m> wikked 
[22:07] <jrand0m> gracias 
[22:07] <mids> so that indicates that nop believes that IIP is more stable again
[22:07] <mids> if that is worth anything...
[22:07] <mids> *cough*
[22:07] <jrand0m> ok cool 
[22:08] <jrand0m> [woot roommate just handed me a glass of wine for the meeting] 
[22:08] <MrEcho> lol
[22:08] <jrand0m> ok, since nop is online and won't come to hte meeting, we'll have to save the lynch mob for later 
[22:09] <jrand0m> 2) 0.2.3 &amp; 
[22:09] <mids> what specific question do you want to ask him?
[22:09] <protocol> when is the meeting
[22:09] <jrand0m> specific question> when will he make an official announcement describing the past problems and how the future ones will be addressed? 
[22:09] <jrand0m> the meeting is now 
[22:10] <jrand0m> (aka, at what point should we explore non-iip means of communication) 
[22:10] <mids> if I get an answer I'll let you know.
[22:10] <jrand0m> thanks 
[22:11] <jrand0m> ok, i2p stuff.  0.2.3 came out yesterday, and while most of hte kademlia code is working fine, there are some 0.2.2 bugs showing up as well as some other bugs being explored. 
[22:11] <jrand0m> i've committed a change to use tunneled messages for dbStore instead of garlics, which should reduce the load tc (et al) have been seeing on servers 
[22:12] <jrand0m> there is also a new persistent sessionKeyManager that will make it so restarts won't totally b0rk a router for 15 minutes 
[22:12] <MrEcho> what about client connect times to routers?
[22:12] <duck> so far it feels as good/bad as 0.2.2; unless my router/tunnels go down again this night, in which case it is worse as 0.2.2
[22:13] <jrand0m> MrEcho> that seems to be in the interaction of two bugs from 0.2.2 thats acting up more than before.  those two are my top priority. 
[22:13] <MrEcho> ok cool
[22:13] <jrand0m> duck> my feeling is that its worse than 0.2.2, from an end user perspective.  i'm working on fixing that without sacrificing anonymity or security. 
[22:13] <MrEcho> its hard to work on the dns with that damm bug .. i have to restrt the dns server alot
[22:14] <jrand0m> MrEcho> with local only routers i have not been able to reproduce the bugs - does it work for you w/ local only? 
[22:15] <MrEcho> no
[22:15] <jrand0m> could you send me debug logs for that? 
[22:15] <MrEcho> already deleted
[22:16] <jrand0m> ok, if you try again and it doesn't work, if you could send me debug logs from both the router and client I'd appreciate it. 
[22:16] <MrEcho> its doing the samething as before .. client gets msg that its sent .. but never makes it to the client
[22:16] <MrEcho> to the other client
[22:17] <MrEcho> ya .. ill see what i can do
[22:17] <jrand0m> ok, sounds like the i2psessionImpl2 bug.  i haven't been able to reproduce that locally, but once its fixed for remote hopefully it will work for your situation 
[22:17] <jrand0m> gracias 
[22:17] <jrand0m> in any case, thanks for y'all's patience with the update.  we're making progress, even if it doesn't feel like it on the surface 
[22:18] <protocol> shine on you crazy diamond
[22:18] <duck> in the future, say once i2p is actually used, how will the development / release process change to prevent broken releases from mess up the net?
[22:19] <jrand0m> once 1.0 is out, i'll do dev &amp; roll out to an insane group of volunteers to play with for a week, then if things wokr great, it'll get rolled out to general release. 
[22:20] * FireRabbit will be an insane vollunteer
[22:20] <jrand0m> right now i've got to battle with kaffe &amp; jetty for updates on i2p.dnsalias.net 
[22:20] <duck> what species?
[22:20] * MrEcho already is
[22:20] *** tusko (~tusko@anon.iip) has joined channel #iip-dev
[22:20] <jrand0m> y'all already are insane (and very helpful) volunteers :) 
[22:20] <FireRabbit> thank you!
[22:20] <FireRabbit> :)
[22:21] *** TC (~TC@anon.iip) has joined channel #iip-dev
[22:21] <jrand0m> hey if it aint tc 
[22:21] * MrEcho wips TC .. your late
[22:21] <TC> hey
[22:21] <TC> we back up and running?
[22:21] <MrEcho> ya i can type todya...
[22:22] <jrand0m> iip seems up... 
[22:22] <TC> yay
[22:22] <jrand0m> in any case, i'm hoping to have out in the next few days, once the two critical bugs get fixed (the cpu overload tc has seen has already been updated) 
[22:23] *** wiht (anon@anon.iip) has joined channel #iip-dev
[22:23] <TC> what was the cause?
[22:23] <FireRabbit> i seem to have noticed increased disk activtiy since updating to 0.2.3 but i havent spent any time to see if thats actually i2p or just the comp being stupid
[22:23] *** Signoff: wiht ((null))
[22:23] <TC> FireRabbit, how much memmory do you have?
[22:24] <FireRabbit> that computer has 128 i believe
[22:24] <FireRabbit> you think it could be the paging file?
[22:24] <jrand0m> the cause was that 0.2.3 sends all dbStore messages via garlic routed messages instead of directly, which uses either ElGamal or AES+SessionTag (depending on whether tags are known).  the persistentSessionKeyMAnager will make tags last longer, and will send dbStore messages through tunnels instead 
[22:24] <TC> because i have 512 and i2p gave me an 'out of memmory' error last night
[22:24] <jrand0m> really?  shite 
[22:24] <FireRabbit> oh, intresting
[22:25] <MrEcho> wow
[22:25] <jrand0m> yeah, thats #3 on the list of bugs left to crack (though thats not a showstopper) 
[22:25] <jrand0m> OOMs don't use all 512 
[22:25] <TC> but it seems to be running fine now
[22:25] <jrand0m> they only use what java's given (e.g. 64M) 
[22:26] <TC> yes
[22:26] <duck> Memory: In use: 8187KB
[22:26] <jrand0m> word 
[22:26] <duck> that is not much!
[22:26] <duck> yet
[22:26] <MrEcho> Memory: In use: 8908KB Free: 4088KB 
[22:27] <jrand0m> right, there is something growing in there, i hope to have it tracked down by 0.3  
[22:27] <jrand0m> cool, free means it used to use 12.9M, now it only uses 8.9 
[22:27] <TC> its running at 30megs of memory at the moment but last night it jumped up to (what windows told me) '70' about then is where it crashed
[22:27] <jrand0m> yeah, kaffe does that for me tc 
[22:28] <jrand0m> ok, in any case, people should subscribe to the i2p mailing list 
[22:28] * FireRabbit is thinking when he gets home today hes going to rewrite the meshwork lib since it has some problems
[22:28] <FireRabbit> sigh
[22:28] <jrand0m> ((Link: http://i2p.dnsalias.net/pipermail/i2p/)http://i2p.dnsalias.net/pipermail/i2p/) 
[22:28] <jrand0m> d'oh FireRabbit 
[22:28] <FireRabbit> this thing is never going to gte done
[22:28] <TC> yah, and memory is no biggy for the most part
[22:28] <jrand0m> heh, no project goes as easily as one hopes 
[22:28] <FireRabbit> nope
[22:28] <protocol> jrand0m: the maillist triggers Yahoo! spam protection
[22:28] <protocol> just a heads up
[22:28] <jrand0m> really protocol? 
[22:29] <protocol> yeah
[22:29] <jrand0m> perhaps thats what triggered the spam guard when i cc'ed iip-dev 
[22:29] * jrand0m will write my isp
[22:29] <jrand0m> (or perhaps its the .dnsalias.net thing) 
[22:30] <protocol> i didn't get any mailings so far, and i emptied my bulk mail b4 i could check
[22:30] <duck> or the jrandom nicknam
[22:30] <jrand0m> lol duck 
[22:30] <FireRabbit> :)
[22:30] <jrand0m> that'd be awesome if my nick was filtered :) 
[22:30] <FireRabbit> hehe
[22:30] *** wiht (anon@anon.iip) has joined channel #iip-dev
[22:30] <jrand0m> wb wiht 
[22:30] <jrand0m> speaking of which, I suppose I should inject 3.1) apps :) 
[22:31] <jrand0m> hey MrEcho, how goes the battle? 
[22:31] <wiht> jrand0m: Hello.
[22:31] <MrEcho> the day somone writes a autodetect program for the linux compile config
[22:31] <MrEcho> well its on its way
[22:31] <duck> knoppix uses some autodetect thing, isnt it?
[22:31] <jrand0m> ./configure ; make ; make check ; make install ; reboot 
[22:31] <duck> </offtopic>
[22:31] <MrEcho> ive pritty much maped out how i want to do everything
[22:31] <jrand0m> word 
[22:32] <jrand0m> do you have a clear view on how i2ptunnel could be updated to make use of what you're doing MrEcho? 
[22:32] <FireRabbit> i think knoppix uses hotplug
[22:32] <MrEcho> 0.1 wont be/might be locked down .. dont know yet
[22:32] <jrand0m> coo' 
[22:33] <TC> oh jrand0m, i have a question about the cvs
[22:33] <jrand0m> que tal? 
[22:33] <MrEcho> for dns querys im going to have a server port on the Client and RS side for Names querys
[22:33] <FireRabbit> ok jrand0m so enlighten me on this, if you have two arrays, one thats storing data just recieved and one thats acting as a buffer what would you name them
[22:33] <MrEcho> and im going to build a lib for any app to use
[22:33] <jrand0m> FireRabbit> src, dest 
[22:34] <FireRabbit> humm
[22:34] <TC> i thought it would be a good idea if i updated the host file directly to the i2p based cvs so it could be included with future versions
[22:34] <jrand0m> definitely tc 
[22:34] <FireRabbit> this is a pretty big class, i think id want to go a little more specific than that
[22:34] * jrand0m should get you a cvs account
[22:34] <TC> im just wondering how to connect to it
[22:34] <duck> TC: you want (Link: http://www.tortoisecvs.org/)http://www.tortoisecvs.org/
[22:34] <duck> easiest CVS client for windows that I know
[22:35] * MrEcho uses the dos ver :)
[22:35] <mihi> duck: for windows != win9x ;)
[22:35] * FireRabbit uses the cvs command line port
[22:35] <duck> mihi: I did test it with win9x
[22:35] <jrand0m> tc> have you used cvs before?  or are you concerned w/ anonymity?  (you should be able to cvs through i2p at the moment) 
[22:35] * mihi uses either WinCVS or the cygwin cvs
[22:35] * jrand0m uses cvs.exe
[22:35] <TC> ok, so i use that client and set up the proxy?
[22:35] <TC> no, ive never used cvs before
[22:35] <jrand0m> ok, i'll walk you through the setup after the meeting 
[22:36] <TC> sure, thanks
[22:36] <duck> about cvs-ing through the tunnel:
[22:36] <duck> wouldnt the double messages be a big problem?
[22:36] *** Signoff: wiht (Ping timeout)
[22:37] <duck> especially for commits
[22:37] <jrand0m> yes duck, but I haven't run into that problem (cvs messages are typically small) 
[22:37] <jrand0m> >64k messages (e.g. the specs .pdf or .sxw) should for now be done through the normal internet 
[22:38] <duck> jabber msges get also duplicated quite often
[22:38] <jrand0m> you're right though, in that its not a rock solid solution for cvs yet 
[22:38] <duck> even though they are XML, they are not that big
[22:40] <jrand0m> right, lost acks are one of the bitches of the current lost i2psessionimpl2 bugs :/ 
[22:40] <duck> k
[22:41] <duck> (that was a partly lost ack)
[22:41] <jrand0m> (with the network this size, there should be no resends ever, unless that the peer is offline) 
[22:42] <jrand0m> hmm ok, any other i2p stuff? 
[22:42] <mihi> jrand0m: how about adding some kinda sequence number into the i2p packets?
[22:43] <jrand0m> i2ptunnel packets? 
[22:43] <mihi> this would help with the doubling things.
[22:43] <mihi> no, i2pnp packets
[22:43] <mihi> okay, one could do it on i2ptunnel level as well.
[22:43] <TC> so jrand0m did you get your conncetion back or are you still at a cafe?
[22:43] <mihi> just if you get twice the same number, disregard the second one.
[22:44] <jrand0m> those already handle dup ids for most things, though you're right in that there's going to be an update on 0.3 for the remaining messages 
[22:44] <jrand0m> right, currently we keep a history of the last 1000 msgIds to drop dups 
[22:44] <mihi> okay, if anyone volunteers to write a good tcp impl for i2p, that would be better ;)
[22:44] <jrand0m> yes! :) 
[22:44] *** Nostradumbass (nostradum@anon.iip) has joined channel #iip-dev
[22:45] * jrand0m thinks there's going to be a bounty for some [yet to be determined killer app/feature] once 1.0 gets near
[22:45] <duck> win a 1 hour private chat session with UserX!
[22:45] <jrand0m> lol 
[22:45] <MrEcho> lol
[22:46] <jrand0m> ok, any other i2p things, or iip things, or anything else for this, the 69th iip-dev meeting? 
[22:46] <jrand0m> (other than userx pinup girl comments) 
[22:47] <duck> any other apps that duck inc. should run?
[22:47] <jrand0m> bluebeep! 
[22:47] <TC> 1. jrand0m did you fix your connection issues? 2. what do you think of my new eepsite?
[22:47] <TC> bluebeep?
[22:47] <jrand0m> oh sorry tc.  yes, i finally have net access :)  haven't seen your new eepsite beyond the board (which kicks ass), but i'll check later :) 
[22:48] <duck> TC: I like the new design
[22:48] <TC> hmm, i should change the board as well to cut down on the load time
[22:48] <duck> only think you should try to disable the email function in the phpboard, now you get an error each time
[22:48] <TC> thanks duck
[22:48] <jrand0m> dropping images would be a plus 
[22:49] <TC> good idea
[22:49] <jrand0m> (bluebeep is an old wardialer) 
[22:49] <MrEcho> ya
[22:49] <jrand0m> (and all around fun toy) 
[22:49] <duck> please keep in mind that the average age is 16 here
[22:50] * MrEcho is 24
[22:50] * duck ducks
[22:50] * jrand0m doubts there are too many 3 year olds to balance out the geriatrics among us ;)
[22:50] *** wiht (anon@anon.iip) has joined channel #iip-dev
[22:50] <MrEcho> lol
[22:50] * TC built a blackbox once
[22:50] <jrand0m> w3wt 
[22:50] <lonelynerd> is the meeting already over?
[22:50] <duck> last Q:
[22:50] *** protocol is now known as proto_afk
[22:51] <duck> how can we read the kademlia stats?
[22:51] * jrand0m hasn't !baf'ed yet lonelynerd, so ask away :)
[22:51] * MrEcho kills pcmcia support in the kernel
[22:51] <duck> just so that we understand what routerConsole.html dumps
[22:51] <MrEcho> im getting pissed
[22:51] <jrand0m> ok, the JobQueue stats I assume you mean? 
[22:52] * duck guesses that it is all obvious probably
[22:52] <jrand0m> basically when I look at JobQueue stats, I check to see that the avg execution time for the Build garlic message, buld tunnel, and handle * message jobs are small 
[22:52] <jrand0m> (those are the jobs that usually take the longest, and when the pending side of things gets large, everything suffers) 
[22:53] <lonelynerd> (actually, i better read the logs first)
[22:53] <duck> gotcha
[22:53] <jrand0m> the .1-.6s avg pending time i've been seeing is shit poor and one of the big things i'm going to aim for once its time to tune 'em 
[22:54] <jrand0m> the netDb contents liveliness and reliability are largely random numbers, as long as they're > 100.  last sent successfully means when was the last time it was sent to 2 or more peers  
[22:54] <jrand0m> (we resend randomly if it isn't local) 
[22:54] <jrand0m> (no more than once every 5 minutes though) 
[22:55] <jrand0m> is there a stat that would be helpful for people, or some other visualization that might help?  (if its nontrivial i might not throw it in, but if its easy, i probably would) 
[22:56] <duck> thanks
[22:57] <jrand0m> any other comments / questions / concerns / frisbees? 
[22:59] <jrand0m> in that case 
[22:59] * jrand0m winds up
[22:59] * jrand0m *baf*s the meeting closed