I2P dev meeting, March 28, 2006

Quick recap

  • Present:

ashter, bar, cervantes, Complication, jrandom, nextgens, tethra,

Full IRC Log

15:08 < jrandom> 0) hi
15:08 < jrandom> 1) Net status and
15:08 < jrandom> 2) Use case survey
15:09 < jrandom> 3) ???
15:09 < jrandom> 0) hi
15:09  * jrandom waves
15:09 < Complication> Finally loaded, reading :)
15:10 < jrandom> weekly status notes posted up at dev.i2p.net/pipermail/i2p/2006-March/001274.html
15:10 <@cervantes> *** connection reset
15:10 < jrandom> heh
15:11 < jrandom> ok, while y'all dig into that, lets jump on in to 1) Net status
15:12 < jrandom> about 2/3rds of the net has upgraded to (thanks!), and results have been mixed
15:12 < jrandom> anyone out there on low bandwidth links have any experiences they want to share?  better / worse / no difference?
15:13 < jrandom> or, any results from folks on dsl-class links?
15:13  * jrandom has heard (and felt) some results on faster links (largely negative, unfortunately)
15:14 <+Complication> Well, I wanted to say that net status is a bit flaky. :) But the net said it first. :D
15:15 <+Complication> On the scale of recent disconnects, this was a very rapid recovery, though.
15:16 <+Complication> Haven't had any more massive message jams, but it still loses a lease now and then
15:17 <+Complication> Also, I think the last router run ended when a lease couldn't be renewed, so it concluded "Router hung!"
15:18 < jrandom> ah col
15:18 <+Complication> Had been ticking for 15 hours or so
15:18 < jrandom> perhaps we should adjust the watchdog to stop restarting the router under those situations
15:19 <+Complication> Retransmission is also the same as before (uncomfortably high, but apparently possible to live with - which in itself is good news)
15:19 < jrandom> the restart used to be necessary, but recurrant tunnel failure should be recoverable
15:19 < jrandom> hmm, <10%, <20%, >20%?
15:20 <+Complication> > 20%
15:20 <+Complication> I don't know many protocols which work tolerably when every third message goes missing
15:21 <+Complication> This one works :) But it used to be around 7%
15:21 < jrandom> well, thats averaged across all of the peers, so its probably quite low for most peers, but quite high for highly congested peers
15:21 < jrandom> (as shown on peers.jsp)
15:22 <+Complication> True, and I haven't taken a look at that side of the distribution yet
15:23 <+Complication> Might need to check, if for nothing else, then to verify how it's distributed
15:24 < jrandom> cool, thanks Complication 
15:24 < jrandom> ok, anyone have anything else on 1) Net status?
15:25 < bar> Complication: may i ask what burst limit you are using? mine are set to 60% of my theoretical upload max, and i currently have a retransmission ratio of 11% 
15:26 <+Complication> bar: it's around 80% of line speed
15:26 < bar> ok
15:26 <+Complication> On the same level as it was, when retransmission was around 7%
15:26 <+Complication> Had it higher meanwhile, but brought back down
15:28 < bar> i'll try to use 80% for a day or so to see if anything happens
15:28 <+Complication> And sustained transfer limit is around 65%
15:28 <+Complication> Actual transfer, if the total indicator is correct, averages near 60% of line speed
15:29 <+Complication> (peaks are higher)
15:30 < ashter_> for my part lot of 'no lease' thing for local destination (as i said today)
15:30 < ashter_> and a node a bit more congested
15:30 <+fox> <nextgens> hi
15:30 < jrandom> heya nextgens
15:30 < jrandom> ashter_: hmm, are you on dialup, dsl/cable, or faster?  or, better said (more anonymously), are you congested?
15:31 <+fox> <nextgens> cool, jrandom is around :) you might help me :)
15:31 < jrandom> (as in, network congestion, not the numbers i2p displays)
15:31 < ashter_> dsl/cable
15:32 < jrandom> ok thanks
15:33 < jrandom> ok, if there's nothing else on 1) Net status, lets jump on over to 2) Use case survey
15:34 < jrandom> I don't expect anwers immediately, but if y'all could put some thought into the questions from the mail and post up replies (either to the forum, syndie, the list, etc), it'd be much appreciated
15:37 <@cervantes> *cough*
15:38 <+tethra> oh dear :/
15:39 < jrandom> (|grep -v -- -\!- ;)
15:39 < jrandom> ok, as I said though, bounce word through whatever fashion you care to use at your convenience.  gracias
15:39 < jrandom> movin' on to 3) ???
15:39 < jrandom> anyone have anything they want to bring up for the meeting?
15:40 <@cervantes> http://forum.i2p.net/viewtopic.php?p=7442  <-- sticky thread for use case discussion
15:40 < jrandom> ah cool, thanks cerv
15:42 < ashter> (erf that happened once again, and when this occurs participating tunnels grows insanly :( )
15:43 < jrandom> hmm, to the thousands, or hundreds?
15:43 < jrandom> (there are a few fixes for bursts of new tunnels pending, should be out later this week)
15:43 < ashter> thousands
15:44 < ashter> (ok thank you)
15:44 < jrandom> ok cool.  you may want to conider lowering your bandwidth limit or share percentage in the meantime
15:44 < jrandom> ok, anyone have anything else for the meeting?
15:45 < jrandom> if not...
15:45  * jrandom winds up
15:46  * jrandom *baf*s the meeting closed