23:04:12 <teor> #startmeeting Network Team 4 December 2018
23:04:12 <MeetBot> Meeting started Tue Dec  4 23:04:12 2018 UTC.  The chair is teor. Information about MeetBot at http://wiki.debian.org/MeetBot.
23:04:12 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
23:04:13 <ahf> ok, so tired that i got /lastlog vs. /me wrong
23:04:39 <teor> if you need to sleep, please do
23:05:00 <teor> So what's on the roadmap for the next few weeks?
23:05:00 <ahf> no, it's fine. it's just late so i do more typos than normally :-)
23:05:40 <gaba> Just a note about the roadmap (in case is not understandable) that I divided it into 3 blocks: doing, done and todo.
23:06:19 <gaba> The doing is what we are working on right now.
23:06:32 <teor> So we have mikeperry and asn on WTF-PAD, a bunch of us on s8, and a few other little tasks
23:07:21 <teor> For team rotations, I am on bug triage, and asn is on CI/Coverity
23:07:48 <gaba> ok
23:08:38 <gaba> Do anybody need any help/work on s8? Next monday we can chat about snowflake, based on what came up from the meeting that ahf coordinated last week.
23:09:06 <ahf> i think i'm good there
23:09:53 <teor> I am ok. I will be working on fallback directory mirrors this week, because enough have failed to affect bootstrap times. (Which makes s8 harder to do!)
23:10:22 <catalyst> do we do automated monitoring of the fallback directories?
23:10:26 <nickm> (at the hotel restaurant)
23:10:56 <gaba> nickm: the paper on quic in tor is done, right?
23:11:25 <nickm> yes; I think mike would like minor changes, but it's already in the form that mozillians are consuming it in
23:11:31 <gaba> ok
23:12:16 <mikeperry> yeah I sent in some changes on top of komlo's last week
23:12:24 <dgoulet> catalyst: they are monitored by DocTor (might not be the best thing we can do but reachability is)
23:12:25 <nickm> sorry, i'd already sent it to tor-dev
23:12:33 <nickm> mikeperry: do you think we need an 0.7?
23:13:06 <catalyst> dgoulet: ok some of the behavior i was seeing seemed to be a connection succeeding but the consensus download took like 30sec
23:13:09 <mikeperry> ah I don't know. I guess I need to see exactly what rev was sent to tor-dev
23:13:12 <mikeperry> was it v4 or v5?
23:13:44 <nickm> v0.6
23:13:55 <nickm> anyways, let me know if you think we should send another version
23:13:56 <dgoulet> catalyst: right... yeah not ideal :S. We could hack DocTor one day to do a better job there at knowing if it is just unnacceptable fallback
23:14:08 <teor> I think there is a timeout on DocTor
23:14:15 <teor> Let me check my emails
23:14:22 <dgoulet> https://gitweb.torproject.org/doctor.git/tree/fallback_directories.py
23:14:29 * gaba needs to go get kids from school. hopefully will be back before meeting ends but otherwise i will read backlog later.
23:14:32 <dgoulet> oh it does check, 15 sec and more
23:14:46 <dgoulet> line 65/66
23:15:13 <teor> Yes, 15 seconds is reasonable, if we want to bootstrap in 30
23:15:24 <mikeperry> nickm: yeah I guess it depends on if we want to post it elsewhere or just give people the tor-dev post url
23:15:42 <dgoulet> teor, catalyst: oh with a notification threshold also ... 50% of fallback dir need to fail for a notify to be sent...
23:15:44 <teor> One of the reasons that fallbacks are slow is that some (custom?) clients are using them for excessive query volumes
23:16:08 <teor> dgoulet: atagar changed the threshold to 50% because he didn't want to keep getting the 25% alarms
23:16:16 <dgoulet> ok
23:16:24 <teor> It will go back to 25% once the list is rebuilt
23:16:26 <catalyst> what URL is DocTor currently hosted at?
23:16:37 <dgoulet> something Metrics could graph tbh... but another discussion
23:16:51 <teor> dgoulet: the graph is in consensus-health
23:16:52 <dgoulet> catalyst: current consensus URL I believe, using Stem
23:17:11 <nickm> mikeperry: I think it should turn into a whitepaper after its first round of comment; let's incorporate your changes then?
23:17:19 <dgoulet> teor: oh neat
23:17:29 <catalyst> dgoulet: oh i was wondering if it had a web page output or something. it looks like not?
23:17:46 <teor> https://consensus-health.torproject.org/graphs.html#fallbackdirstatus
23:17:47 <mikeperry> nickm: ok
23:17:49 <dgoulet> catalyst: it doesn't no, only a python script and notification goes by email
23:17:57 <dgoulet> teor: cool
23:17:59 <teor> (and to #tor-bots)
23:18:25 <teor> I think the consensus-health check only looks at the Running flag
23:18:31 <teor> But doctor checks speed
23:19:50 <teor> Shall we move on to the discussion?
23:21:25 <teor> gaba wanted to discuss snowflake at the next meeting
23:21:42 <teor> Does anyone need help with reviews?
23:22:09 * ahf is good
23:22:41 <dgoulet> I think I went over all the ones I have today so I'm good
23:22:58 <teor> ahf, catalyst, and mikeperry have reviews that are more than 2 weeks old. But I think catalyst has been working on some? of their reviews, and mike has been on leave.
23:23:12 <teor> https://trac.torproject.org/projects/tor/query?status=needs_review&component=Core+Tor/Tor&col=id&col=summary&col=status&col=type&col=priority&col=milestone&col=changetime&col=reviewer&col=keywords&order=changetime
23:23:38 <mikeperry> yeah I may need a little help, not sure yet
23:23:58 <teor> Ok, let us know. I have a bit of spare time this week and next week
23:24:06 <nickm> i'm happy to take on a review so long as it isn't something I wrote and I don't have to get to it before Saturday :)
23:24:11 <ahf> yeah, the 28179 finishing last week meant i didn't get all my reviews done
23:24:46 <mikeperry> yeah the next two weeks are a little tight for me. I agree with asn that we should have wtf-pad clean and ready in a week or two
23:24:48 <ahf> nickm: getting your eyes on 28179 would be much appricated, it's pretty big
23:24:49 <dgoulet> ahf: the spec ticket you have to review should go quite fast for you ;)
23:24:59 <ahf> dgoulet: yes!
23:25:00 <mikeperry> but it feels like a lot of work. mostly a lot of small things to keep track of
23:25:20 <nickm> ahf: that's in merge-ready, so it's on my queue.
23:25:24 <teor> Are there reviews you'd like to give away?
23:25:34 <ahf> nickm: awesome
23:26:23 <mikeperry> teor: I mean maybe the consensus ones?
23:27:23 <teor> The spec and the code ones? I can take those on.
23:28:09 <teor> Done!
23:28:39 <mikeperry> cool thanks
23:29:55 <mikeperry> I also have a question for nickm about the branch.. the branches are a little chaotic right now. I feel like the best way forward is to squash them down and resync with asn.. and then do fixups on top of that, linking to them in review comments but maybe squashing periodically.
23:30:19 <mikeperry> (the wtf-pad branches, which is now plural because asn and I are both working on our one ones)
23:30:20 <nickm> I am fine with doing a fresh review on a fresh WTF-pad branch when I look at it if that's what you decide
23:30:40 * gaba is back
23:30:43 <mikeperry> ok
23:32:06 <teor> gaba, you wanted to talk snowflake next monday?
23:32:13 <gaba> yes
23:32:26 <gaba> unless there is a special reason why we should do it today :)
23:33:12 <teor> I think you'll have more snowflake people on Monday
23:33:23 <gaba> yep
23:34:06 <teor> Let's move on to the next question?
23:34:20 <teor> Can we add Core Tor/sbws to the bug triage and CI roles?
23:34:35 <teor> (asn already assigned sbws reviews for this week)
23:34:49 <gaba> +1
23:36:18 <teor> OK, I will make those edits to the roles?
23:36:54 <gaba> It seems good to me.
23:37:15 <teor> Next question: Is our process for "proposed" working? <-- we still didn't go through it yet
23:37:38 <teor> I have moved features from 035-roadmap-proposed to 040-roadmap proposed
23:37:50 <nickm> my sense was "no, we need a better process" but I don't know.
23:37:57 <nickm> do we have a proposal for a better process?
23:37:59 <teor> Yeah, I think we need a better process
23:38:16 <gaba> Sorry I still didnt have time to look at it.
23:38:38 <teor> I feel like our triage process needs work in general
23:38:57 <teor> Here's one possible proposal:
23:39:12 <teor> * each release is roadmapped at our team meetings
23:39:19 <teor> * we put tickets in the release if they're on the roadmap
23:39:45 <teor> * changes to the roadmap get discussed at weekly meetings, and then we update the tickets in the release
23:40:15 <teor> Tagging tickets doesn't work, because we don't talk about specific tickets at the meeting
23:40:55 <gaba> can we have this proposal in a pad for people to look at it and then we discuss it in one of the weekly meetings?
23:42:13 <teor> (working)
23:42:55 <teor> Ok, I will write the pad
23:42:59 <gaba> thanks
23:43:13 <teor> Next question is about sponsor 8
23:43:18 <teor> 1 week and 3 days to go
23:44:26 <teor> Sponsor 8 is wrapping up in December. Right now working on this: catalyst, ahf, dgoulet, teor, nick. Any of you need help with it?
23:44:35 <teor> Oh, maybe 2 weeks
23:44:52 <nickm> I'm fine with my part.  But I am working on contingency plans...
23:45:04 <nickm> if all the branches are needs_review this week I'll be a happy merger..
23:45:07 <catalyst> i count the week of 12/10 and the week of 12/17
23:46:06 <nickm> If Ive got to 12/17 without a code to review, it will be less comfortable
23:46:11 <gaba> Let's consider until 12/21. So the week of 12/17 would be the last one.
23:46:13 <nickm> *without code
23:46:16 <gaba> ^ yes
23:46:34 <gaba> Do we know when we are going to have code to review?
23:47:33 <nickm> [food is here, I must sign off. peace all!]
23:48:52 <teor> I think my little tickets are in review, or need quick revisions
23:48:55 <ahf> i was hoping to have something working for my tor.git s8 tickets left that requires code next week
23:49:22 * catalyst has one refactoring ticket in review, and another refactoring patch almost ready for review
23:49:50 <gaba> ok. So we are on track.
23:50:59 <teor> I can do the fallbacks as part of s8, but if they don't happen by the deadline, that's ok
23:51:21 <gaba> ok
23:52:11 <gaba> That is for s8 for today then.
23:53:01 <teor> Next question: Please see tickets with tag "035-rc-blocker?" -- are any of them really rc blockers?  Are there any other true rc blockers?
23:53:20 <teor> There are two tickets tagged 035-rc-blocker
23:53:23 <teor> https://trac.torproject.org/projects/tor/query?status=accepted&status=assigned&status=merge_ready&status=needs_information&status=needs_review&status=needs_revision&status=new&status=reopened&keywords=~035-rc-blocker&milestone=^Tor:+0.3.5.x-final&group=status&col=id&col=summary&col=keywords&col=status&col=owner&col=type&col=priority&order=priority
23:54:18 <teor> I think those tickets need answers before the rc
23:54:22 <teor> Are there any others?
23:55:28 <gaba> that is before dec 15th
23:55:45 <teor> Yeah, I think we are pretty good for 0.3.5 rc
23:55:52 <teor> 3 minutes left
23:56:09 <gaba> ok
23:56:15 <teor> Two more questions: Let's suspend discussion on role "reviewer" until January. ok?
23:56:20 <teor> Release schedule: any changes?
23:56:27 <teor> Any any other questions from anyone?
23:56:51 <gaba> I added the one about role reviewer, it seems we need more digging into that. We can resume in January and otherwise bring it to the hackweek.
23:57:06 <gaba> s/and/or
23:58:12 <gaba> No more questions from me.
23:58:21 <teor> Ok, I think we're done. Any last words before I close the meeting tracker?
23:58:26 <teor> Thanks everyone for coming!
23:58:30 <gaba> Thanks!
23:59:25 <teor> #endmeeting