15:58:31 #startmeeting tor anti-censorship meeting 15:58:31 Meeting started Thu Oct 15 15:58:31 2020 UTC. The chair is cohosh. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:58:31 Useful Commands: #action #agreed #help #info #idea #link #topic. 15:58:47 here is our meeting pad: https://pad.riseup.net/p/tor-anti-censorship-keep 15:58:57 feel free to add to it, our agenda is pretty sparse today 16:00:07 o/ 16:01:13 i have a quick remark: i'm considering stopping the monthly report blog post. the idea was to get feedback from our users but that hasn't worked out well 16:01:38 most comments are unrelated and moderation is a big nuisance 16:01:59 ah :/ it was a worthwhile experiment 16:02:01 i think these ~2 hours are better spent elsewhere 16:03:14 yeah 16:03:15 anyway, end of rant 16:03:29 we'll still have it on th emailing list right? 16:03:51 yes, it's probably a good idea to keep the list. at the very least it's helpful to projects like racecar 16:04:09 thanks for undertaking that endeavor phw 16:04:18 but i don't know if it's read by one person or 10,000 16:04:41 yeah i think it's good to have a public report of our progress accessible for the future 16:04:52 even if not many users are reading it *right now* 16:05:00 but the mailing list seems good enough for that for now 16:05:12 i suppose we could also make a blog post and turn the comment section off 16:05:31 making the blog post is as simple as doing "markdown report.md > report.html" and copy&pasting the result into the blog post textbox 16:05:56 maybe, i find the blog more difficult to navigate and search in than the mailing list tbh 16:06:04 yes, that for sure 16:06:29 if we move to a different blog platform though we could revisit this 16:06:43 yes 16:07:16 okay, in other news i have some new snowflake measurements from china that might be interesting: https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/issues/32657#note_2712025 16:07:38 i am curious what are your thoughts on the results 16:08:00 snowflake seems to be generally uncensored and reachable but the throughput seems so bad that it's not really usable 16:08:21 * phw looks 16:12:07 hmm, my first impression is that this is due to packet loss or more generally: bad international connectivity 16:12:20 CN is in the top 5 of users according to dirreq-stats on the bridge 16:12:22 dirreq-v3-ips ru=56,us=48,by=32,cn=32,tm=16,??=8,ae=8,al=8,at=8,au=8,bd=8,be=8,br=8,ca=8,ch=8,co=8,de=8,ec=8,eg=8,es=8,fi=8,fr=8,gb=8,id=8,il=8,in=8,ir=8,jp=8,kw=8,lb=8,ma=8,my=8,nz=8,pl=8,ro=8,sa=8,se=8,tr=8,ua=8,za=8 16:12:44 i wonder how things would look like if somebody in china were to connect to snowflakes in adjacent places like hong kong, singapore, taiwan, etc. 16:12:50 hmm, this is by ip right, so it's probably not just my tests skewing that? 16:13:26 maybe it is just that the VPS has some connectivity issues 16:14:03 i'd be surprised if there were 30 users in china putting up with more than half of their connections failing due to timeouts 16:14:14 By IP is a different file, but looks similar 16:14:18 bridge-ips us=104,ru=96,by=48,cn=40,de=24,gb=24,br=16,eg=16,in=16,jp=16,tm=16,tr=16,??=8,ae=8,al=8,ar=8,at=8,au=8,bd=8,be=8,ca=8,ch=8,co=8,cz=8,dz=8,ec=8,es=8,fi=8,fr=8,id=8,ie=8,il=8,ir=8,kw=8,lb=8,lr=8,lu=8,ma=8,my=8,nl=8,no=8,nz=8,ph=8,pl=8,pt=8,qa=8,ro=8,sa=8,se=8,tw=8,ua=8,ug=8,za=8 16:14:34 you ran several tests per day, right? and then averaged the success rate? 16:14:42 100 tests per day 16:14:50 this is the % of tests that succeeded 16:14:55 so not even an average 16:14:56 uniformly distributed throughout the day? 16:15:12 no, they were run sequentially at UTC 0000 16:15:48 so at utc 00, we started running 100 tests, one after another? 16:15:53 yeah 16:16:21 it could be that the vps is being throttle for suspicious activity i guess 16:17:00 that's 8 am in china, just around the time everyone starts using the internet 16:17:18 hmm 16:17:38 do you think that would have this big of an impact? 16:17:38 would it be easy to isolate the measurements that happened on the weekend? 16:18:16 that i don't know 16:18:35 yeah, well each of the date labels on the graph are a tuesdsay 16:19:06 at least we have a solid idea of what using snowflake on a tuesday morning looks like ;) 16:19:06 and in any case, the days with the highest succss rate were still around 50% 16:19:22 you can kind of tell where the weekends would be in there 16:20:16 each point is a different day, so the weekend would be 2-3 points behind the date markers 16:20:17 i think it's worthwhile to spend a few hours browsing internet measurement literature to get a better understanding of link quality between china and the rest of the world 16:20:22 yeah 16:20:29 that's a good idea 16:20:49 because it's probably not uniform. i'm willing to bet that connectivity to, say, singapore is much better than to germany 16:21:06 i could probably also alter the tests to get geo data on where each of the snowflakes is 16:21:07 and if that's the case, we may want to look into prioritising certain snowflakes over others 16:21:49 agix was looking for a thesis topic. i think one could turn this into an interesting measurement project 16:22:05 yeah 16:23:02 alright, thanks for taking a look! i'd be interested to hear from those users in china to see how their experience is 16:23:17 yes, that would be useful 16:23:31 okay we have another discussion item 16:23:35 by gaba i think? 16:23:38 anyway, the % of successful bootstraps is > 0, so that's already a success! 16:23:48 on the status of https://pad.riseup.net/p/tor-anti-censorship-tickets-keep 16:24:02 phw: true XD 16:24:48 yes 16:24:50 i added 16:24:53 im going through the tickets 16:24:58 and see if all of them have the label 16:25:03 can we close that pad if we do? 16:25:14 gaba: works for me. i just made a local copy, just in case 16:25:28 ok 16:25:29 yup same 16:26:24 any other discussion items for today? 16:27:09 if not let's move on to our needs help with sections 16:27:14 sounds good 16:27:34 it looks like there's a bug in https://pad.riseup.net/p/tor-anti-censorship-bugs-keep 16:27:48 I'll copy it to gitlab 16:28:14 i think this is just from overzealous logging 16:28:20 me too 16:28:24 but we should probably fix that 16:28:39 because i get people asking this quite a bit on #tor 16:28:54 well like 4 people have asked 16:29:30 should be a simple fix, no? even i can fix that ;) 16:29:54 (i'm in favour of fixing it) 16:30:05 we don't have to litigate it now 16:31:17 okay, i also have an implementation for snowflake#40013 that could use a review 16:31:43 I don't have any time for reviews this week 16:32:05 i'll review it, no problem 16:32:13 phw: thanks! 16:32:26 and agix would like a review of #34218 16:33:15 err 16:33:29 hmm, must be a typo 16:33:40 agix: you around? 16:34:13 https://gitlab.torproject.org/legacy/trac/-/issues/34128 ? 16:34:17 i think it's #34318 16:34:39 nvm 16:34:49 (you can click on somebody's profile in gitlab and it shows you what they did most recently) 16:35:02 i'll try to review it today agix, sorry for the delay :/ 16:35:49 cool, anything else? 16:36:00 not from me 16:36:00 should we pick a reading? 16:36:56 feel free to do so but i cannot promise that i'll be able to read anything over the next few weeks. i'm swamped with financial crypto reviews :/ 16:37:15 alright let's postpone for a bit longer 16:37:25 i am also swamped with some deliverables 16:37:44 *thumbs up* 16:38:09 cool, let's wrap things up for today 16:38:13 #endmeeting