16:59:36 #startmeeting weekly network team meeting 16:59:36 Meeting started Mon May 8 16:59:36 2017 UTC. The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:59:36 Useful Commands: #action #agreed #help #info #idea #link #topic. 16:59:50 Hi all! 16:59:57 hi 16:59:58 updates over here: https://pad.riseup.net/p/SqfWDfC75Vqk 16:59:59 howdy 17:00:04 hello hello 17:00:22 ehlo 17:00:50 dgoulet, isis, isabela, mikeperry, armadev, Yawning: hello there! 17:01:03 asn, komlo, ahf, pastly, catalyst: hi! 17:01:16 Let's start with throwing some updates on the pad, and asking each other questions about them 17:01:38 general notes: it's the last week for merging features into 0.3.1; let's use it wisely! 17:01:55 hello 17:03:05 i really like the pad updates/sending them to @tor_dev after 17:03:40 ahf: are you on-track to have the remaining 278 stuff done in the next day or 2? 17:03:56 ahf: or 3 17:04:04 I want to make sure there's time to review 17:04:14 nickm: i believe so, yes. i think my schedule is that you should start reviewing tomorrow when you check in and then i do some catch up (if there is some) 17:04:33 ahf: also, you asked about additional consdiff tickets for review; that'd be #22148. No need to review today. 17:04:43 ! 17:05:10 nickm: aye! i will get to that tonight. i found it right before the meeting because i had written the wrong ticket in my notes of things i reviewed 17:05:48 dgoulet, asn: I'm not seeing anything hs-related in needs_review for 0.3.1 other than #22006 . Should I expect more to arrive this week, or are we more or less set? 17:06:06 dgoulet: are you going to have a chance to revise #21871 and/or #22060, or should we defer those to 0.3.2 ? 17:06:20 nickm: yes there is one last that I need to revise minor things: #21871 17:06:41 nickm: and yes I'll do that asap for #22060 I completely forgot about that one :( 17:06:56 will look at my 031 stuff anyway this week 17:07:26 ok 17:07:32 nickm: as for groundwork for 031, nothing new will arrive 17:07:42 great 17:07:46 (unless asn's has some up his sleeve ehhe) 17:07:50 nope 17:10:09 * ahf is done writing 17:11:17 i'm out of questions for folks, I think. I wrote a set of post-meeting ticket triage questions at the head of the pad, too: 17:12:03 is our goal with the team rotation tasks that we fill out *all* of them, since right now there's a lot of holes in the cheese 17:12:04 did people get my mail on prop247? especially wondering about armadev 17:12:04 i dont have a discussion point or question /but i do want to give a heads up that sponsor4 April report will need review help 17:12:05 I'd like everybody who is doing anything on 0.3.1, or who owns any 0.3.1 tickets, to take a couple of minutes for one more round of "defer/assign/etc" 17:12:18 prop#247 17:12:42 (i am also sharing teams roadmaps with tor-project email list this week 17:12:49 i'm planning to defer #21693 to 0.3.2 . too many fires and didnt have time to think at all about that one. 17:12:51 I pushed an update to guard-discovery-dev2 with some notes from old mailinglist discussion (in XXX comments) 17:12:53 Pleaes let me know if the "post meeting" section on the pad doesn't make sense 17:12:57 asn: no worries 17:13:01 mikeperry: yes did see. useful comments. 17:14:05 If it's possible, I'd like people to take on reviewing for more of the stuff in 0.3.1. I'll review all I can, but there are a lot of things in needs_review right now. 17:14:32 * isabela lied i do have a discussion point 17:14:38 isabela: add to pad? 17:14:50 done! 17:14:58 discussion topic: There is a hackfest being planned right now for June on the US east coast. It's original purpose was guard discovery attacks, but it might extend to more network team matters. 17:15:11 ^^ that was it :) 17:15:22 so we should let ahf, catalyst, and other members of the network team now about this :] 17:15:25 attendence optional :) 17:15:29 but invited! 17:15:29 yes attendance optional! 17:15:57 hopefully an official mail will be sent to everyone RSN, so people will know more details 17:16:01 yes 17:16:12 i'm up for it. would be nice to see you lot again and work together from there. 17:16:20 i will sync with jon today and we will organize such email 17:16:21 I think we should turn over some amount of the planning here to a single planner who can make it all happen, or else we'll be bikeshedding the plans well into July :) 17:16:24 isabela: ty! 17:17:38 - met with nickm to outline an improved model of bootstrapping so we can hopefully have a better UX in the future 17:17:43 catalyst: would you like to own #20534 ? 17:17:48 catalyst: nickm: are the results of that meeting somewhere? 17:17:58 not yet; we've got a pad we're working on 17:18:00 ack 17:18:11 catalyst: when would you expect we'll be done with a draft? 17:18:30 nickm: maybe in the next 2 days? 17:18:36 sounds good to me 17:20:43 asn US east coast sounds great :) 17:21:06 Oh, this should be kinda obvious, but: please make sure I know if you will be sending me any feature patches for 0.3.1.x that I am not currently expecting. :) 17:21:27 Any more questions for anyone? Any more discussion on hackfest? 17:21:38 komlo: yep. current aim is philly 17:21:43 :) 17:21:48 "maybe" 17:22:47 Okay, in that case let's talk about #21969 ? 17:23:09 asn and catalyst are interested in this one; alecmuffett has hit it hard, and we should probably fix it and maybe backport the fix to 0.3.0. 17:23:21 IIUC, we don't actually have a diagnosis on it yet, though. 17:23:25 yep/... 17:23:26 Any thoughts? 17:23:37 i posted some more thoughts about it today 17:23:53 but no clear diagnosis; i don't hit it in my tor browser 17:23:56 i commented on the ticket that our guard selection during bootstrap might be too strict 17:25:06 hmm. 17:25:26 I wonder if catalyst is running into this for the same reasons as asn, or different ones? 17:25:42 also a relay shouldn't appear multiple times in the guard list right? 17:25:44 When this is reproducible, solving it would be much easier. 17:25:51 catalyst: A bridge can, if it has different PTs 17:26:21 hm it would still have the same fingerprint then? 17:26:23 yes 17:26:46 yeah the situation i saw was an obfs4 bridge listed 3 or 4 times in the TBB config with diferent ports 17:27:21 that's possible for a bridge. But I wonder if that's causing this situation somehow. 17:27:50 well if a failing guard is 2 out of 3 of the selected guards, that does make the problem more likely 17:28:09 ah, hm. Maybe that is the trouble. 17:28:17 er, does make it harder to connect through a guard to a dirauth 17:28:21 asn: but you've seen this without bridges, right? 17:28:33 i havent 17:28:33 actually in my case it was all 3 guards being down, 2 of which were the same 17:28:40 but i think alec's case is not bridges 17:28:44 hm 17:28:45 hi 17:28:49 or something 17:28:54 his case is SoS HS 17:29:03 Yawning: meeting pad over here https://pad.riseup.net/p/SqfWDfC75Vqk 17:29:11 meeting starts 1700 UTC 17:29:11 and i think s7r's case is not bridges either 17:29:16 https://trac.torproject.org/projects/tor/ticket/21969#comment:1 17:29:24 Hello 17:29:26 I wonder if it's the same problem or different 17:29:35 I just woke up 17:30:00 Maybe we should see if we can add log messages to test any of our theories about what's going on here? 17:30:03 has anyone seen this class of problems with 0.2.9 or earlier? 17:30:13 the notes that I added to #21969 include the relevant torrc 17:30:15 no; this is definitely new in 0.3.0 with the prop#271 code 17:30:22 there are no bridges involved 17:31:26 I am guessing that the bridge case and the non-bridge case could be different. 17:31:40 We're not going to figure this all out at this meeting though. 17:31:42 yep i also consider the possibility that these are two different scenarios that make our logic burp 17:31:53 Suggestion: let's use this pad here https://pad.riseup.net/p/vt5NPpeoIZy3 17:31:59 if it helps i can try to make a diagnostic branch about this 17:32:05 to collect open questions and theories, and then figure out how to procede? 17:32:08 ack 17:33:19 can we log which guards are missing descriptors and possibly the contents of CONFIRMED_GUARDS in where the BUG is getting hit? 17:33:34 might work! 17:33:37 i'll add it on the pad 17:34:29 any more discussions / topics / questions / things for this week? 17:34:40 do we log (or keep state) how many times we've failed to connect to a guard? 17:35:32 catalyst: no; but we do keep state for a bunch of other stuff; see the entry_guard_t structure in entrynodes.h 17:35:54 If no more discussions / topics / questions / things, any objections to calling this meeting done for the week? 17:36:21 i had 'is our goal with the team rotation tasks that we fill out *all* of them, since right now there's a lot of holes in the cheese' ? 17:36:27 minor thing 17:36:28 I think it isn't 17:36:31 ack 17:36:34 we're going to have some things not filled for now 17:36:46 and we're holding off on trying to fill observer/liaison roles every week 17:37:11 * nickm waits another 60 sec.... 17:38:11 okay; another fast one. Thanks, everybody! Please remember to do the post-meeting things in the head of the pad. I'll send the pad contents to tor-project. 17:38:17 Have a great week! 17:38:19 #endmeeting