16:59:13 <nickm> #startmeeting weekly network team meeting, 16 April 2018
16:59:13 <MeetBot> Meeting started Mon Apr 16 16:59:13 2018 UTC.  The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:59:13 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:59:16 <nickm> hi folks!
16:59:20 <asn> hello :)
16:59:24 <nickm> Our weekly meeting pad is at https://pad.riseup.net/p/MWK6xElJ0Wnm
16:59:25 * dmr lurks, and waves hi
16:59:37 <pastly> hello
16:59:41 <dgoulet> yello!
17:00:14 <nickm> My kid is off school this week, so please expect me to be a little more confused than usual!
17:00:23 <isis> o/
17:00:25 <nickm> Also, fun fact: Today is Patriots' Day in Massachusetts!
17:01:31 <isabela> o/
17:01:42 <nickm> So, it seems we got a release candidate out on the schedule where we were planning to have stable out.
17:01:54 <nickm> it will be interesting to see whether 034 is more on schedule or not
17:01:55 <isabela> \o/
17:02:23 <nickm> (I think we shouldn't adjust our release schedule for 034 based on 033, since 033 was less-than-completely planned)
17:03:00 <dgoulet> o/
17:03:12 * arma4 is now done with phone call and mostly around
17:03:38 <nickm> So, let's do our roadmap check-in!  In theory, 034 should feature-free in ... 29 days?
17:03:45 <nickm> this is not a lot of time
17:04:35 <dgoulet> I *think* so far for stuff I have, I'm still optimistic for 034 eheh
17:04:47 <dgoulet> as my 033 stuff is gone now
17:04:59 <isabela> so
17:05:04 <mikeperry> prop#291 decisions are impacting how we do remaining vanguard patches
17:05:34 <mikeperry> so meeting about that this week would help reduce/assess remaming risk there
17:05:44 <isabela> for roadmap check in i keep on saying the same thing :) we have 2 weeks till may
17:05:55 <isabela> more or less
17:06:04 <isabela> so review what is there and see what is realistic to stay
17:06:20 <asn> mikeperry: i think we migh be overestimating how much prop#291 impacts the other tickets. I mentioned this here: https://trac.torproject.org/projects/tor/ticket/25754#comment:2
17:06:29 <isabela> you can carry the rows around on gdocs
17:06:47 <asn> mikeperry: or said otherwise, i think ideally the number of guards should not impact how the guard algorithm works.
17:06:48 <isabela> so you can just move things to may if you feel it should be moved
17:06:54 <nickm> Also, please remember: if anybody is going to spend more than 1 day this week on something _not_ on the roadmap spreadsheet, let's talk about it first.
17:07:34 <mikeperry> asn: it impacts #24487. that doesn't need to be done if we drop path restrictions.
17:07:54 <mikeperry> asn: it also changes performance test results
17:08:28 <mikeperry> #25705 also doesn't need to be done in a world without any restrictions..
17:08:42 <nickm> let's finish up any roadmap changes first
17:08:48 <nickm> are any needed this week?
17:08:59 <asn> im good with roadmap for this week
17:09:03 * dgoulet is good
17:09:09 <nickm> Let's all spend this week making progress on 034-roadmapped stuff
17:09:27 <haxxpop> hello!
17:09:28 <nickm> if you have any 034-roadmapped stuff and you haven't reached out to other people working on it, please do so asap :)
17:09:32 <nickm> hi haxxpop !
17:09:35 <mikeperry> (yeah just saying this is hard to say for my part of the roadmap.)
17:09:47 <isabela> looks like nickm dgoulet and mikeperry are primary in a lot of things
17:10:00 <dgoulet> isabela: no no no #1 doesn't mean primary
17:10:01 <isabela> you feeling good with your stuff?
17:10:04 <isabela> ahhh
17:10:05 <isabela> ok ok
17:10:06 <dgoulet> well at least how we put it in
17:10:07 <dgoulet> sorry :S
17:10:28 <isabela> nickm: that is a lot of you around anw :)
17:10:28 <dgoulet> we could use it that way but it ain't like that for now
17:10:34 <isabela> dgoulet: no, is cool
17:10:36 <isabela> no worries
17:10:37 * haxxpop status: having fun implementing onion service v3 client auth
17:10:39 <nickm> isabela: yup, but I'm not scared :)
17:10:43 <isabela> ok ok
17:10:44 <isabela> :)
17:10:46 <dgoulet> hehe
17:10:50 <dgoulet> haxxpop: o/
17:10:51 <isabela> then we good with roadmap
17:10:55 <isabela> everyone is cool w/ it
17:11:33 <nickm> everybody cool with reviews?
17:11:46 <nickm> dgoulet/asn: could #25691 get a reviewer?
17:12:03 <nickm> (I did it this morning, but it's an 0.3.3 regression)
17:12:05 <dgoulet> nickm: sure
17:12:05 <asn> catalyst: hey. i moved your rust trait review bug to nickm, after what you said last week!
17:12:10 <asn> nickm: oh ack.
17:12:17 <nickm> also, rotations!
17:12:25 <nickm> dgoulet: bug triage.
17:12:31 <nickm> isis: community legend
17:12:38 <nickm> mikeperry: coverity
17:12:40 <nickm> nickm: CI
17:12:49 <catalyst> asn: thanks! the PRNG one might be somewhat easier for me, but i haven't looked at it closely yet
17:12:54 <nickm> [nice jobs last week, everybody!]
17:13:33 <nickm> On CI, I'm going to try to end the week with Jenkins happy, Travis happy, and at least one new thing under CI
17:14:14 <catalyst> nickm: sounds ambitious and awesome
17:14:43 <nickm> any handoff from last week's rotations?
17:15:00 <nickm> catalyst: Thanks! but note that I said "try" :)
17:15:01 <mikeperry> nickm: is there info about coverity? I don't even know where that lives
17:15:07 <isabela> i owe the team the retrospective on rotations
17:15:09 <isabela> sorry folks
17:15:18 <mikeperry> nickm: also I have discussion items about jenkins. I guess they can be later?
17:15:20 <isabela> will not drop that ball just had other priorities
17:15:37 <nickm> mikeperry: +1
17:16:36 <nickm> mikeperry: It's at scan.coverity.com
17:18:08 <nickm> mikeperry: ask for help if you get stuck; it is NOT an intuitive system
17:18:22 <nickm> are we ready to move on to discussion stuff?
17:18:42 <dgoulet> good here
17:18:46 <asn> (does mike need an account? and does he have it?)
17:19:02 <nickm> he has one; I just reminded him of what email address he used
17:19:05 <asn> ack
17:19:17 <nickm> asn: so, we need a prop#291 meeting?
17:19:26 <asn> i think that might be good
17:19:34 <asn> because mailing list discussio does not seem to be leading anywhere
17:19:43 <nickm> ok. who should attend?
17:19:52 <asn> me, mike and roger
17:20:01 <asn> that's for sure IMO
17:20:13 <asn> and then people like you would also be super useful
17:20:36 <asn> (and then depending on how much explaining we need to do, we might want people like Aaron Johnson too)
17:20:43 <nickm> Okay. 1700 UTC on Wednesday is okay for me.  How about for the rest of you?
17:20:46 <asn> but in general, I feel like it's gonna be hard to decide on this topic over the internet
17:21:01 <asn> the tradeoffs are too intricate
17:21:36 <mikeperry> I think a high level choice can still be made
17:21:46 <asn> mikeperry: ack
17:21:50 <asn> who do you think should be at the meeting?
17:22:11 <arma4> 1700 utc wed good
17:22:12 <asn> (and yes this wednesday at 17:00UTC works for me)
17:22:29 <mikeperry> a decision on abandoning restrictions or not does in fact impact/eliminate the need for every child of #35546 that is not already in merge_ready
17:22:41 <mikeperry> err #25546
17:22:53 <asn> good point
17:23:15 <nickm> okay, let's plan for that time.
17:23:16 <asn> mikeperry: wed at 17:00 works for you?
17:23:20 <mikeperry> yah
17:23:24 <asn> excellent
17:23:28 <asn> so we got a wed meeting then.
17:23:30 <asn> that's good.
17:23:31 <mikeperry> I can also do tomorrow
17:23:31 <nickm> could somebody tell tor-dev@ ?
17:23:34 <mikeperry> or after this meeting..
17:23:40 <nickm> wednesday is better for me
17:23:48 <mikeperry> but two days give aaron johnson a chance to see it I guess
17:23:56 <asn> ok
17:24:02 <asn> mikeperry: wanna write mail to tor-dev? or you want me to do it?
17:24:06 <asn> i will be avle to do it tomorrow tho
17:24:10 <asn> i have to disappear shortly after this meeting
17:24:54 <mikeperry> I can send the mail
17:25:10 <asn> great thx
17:25:10 <arma4> catalyst: i think being a bit more flexible on the consensus clock skew warning is a good idea. it should be 5 minutes + whatever skew we think is acceptable for a relay.
17:25:27 <arma4> dgoulet: i am still hoping for that dir fetch diagram from your magic book
17:25:37 <dgoulet> arma4: oh yessss! will do after the meeting!
17:25:51 <nickm> catalyst: does that answer your question?
17:25:53 <arma4> catalyst: (5 minutes is because that's when dir auths produce their sigs, so in theory there *can't* be a consensus with enough sigs on it before then)
17:25:55 <nickm> mikeperry: thanks mike!
17:26:20 <catalyst> arma4: thanks! is that timing a consensus parameter or can we hardcode something?
17:26:28 <arma4> dgoulet: i have secret hopes that your diagram will make it clear that i can propose my "how about we make relays not fetch a consensus between :55 and :00" idea
17:26:31 <nickm> mikeperry: what's your jenkins question?
17:27:00 <mikeperry> I have a couple
17:27:17 <arma4> catalyst: hum. i see a "voting-delay" line in the consensus. wonder if that's relevant.
17:27:30 <mikeperry> first: who owns jenkins? weasel? can we add a commit hook to mail committers that triggered the build, if it fails?
17:28:16 <arma4> catalyst: from dir-spec.txt it looks very relevant
17:28:24 <catalyst> arma4: i think we should open a ticket on this so we can figure out the details of how we want to do this (i can do that if nobody else volunteers)
17:28:36 <arma4> catalyst: we shouldn't hard-code it, at the least because the test network makes consensuses at a different frequency
17:29:35 <isis> mikeperry: it doesn't build every commit afaik, so e.g. if my commit braeks something, and then yours is on top of it, it would email you telling you that you broke something
17:29:40 <arma4> catalyst: though, if we wanted to do a short-term hack, 5 mins is a fine number i think. the only bad effect is that maybe sometimes we wouldn't warn about slight clock skew on some test network.
17:29:43 <arma4> catalyst: a ticket sounds great
17:29:44 <nickm> mikeperry: weasel owns jenkins. I wouldn't mind getting emailed, but the false-positive rate right now is too high.
17:29:46 <mikeperry> second, which is I guess just a thing to do, is document what is involved in jenkins babysitting
17:29:49 <nickm> +1 on a ticket
17:30:15 <arma4> catalyst: i think dizum still hasn't fixed its clock :/
17:30:39 <isis> mikeperry: also, it only tests master and maint-*, also what nickm said about the false-positive rate
17:30:58 <arma4> catalyst: oh, i take it back! dizum looks right now.
17:31:05 <mikeperry> and the false positive rate is due to what? builds randomly fail due to jenkins issues?
17:31:30 <isis> yes
17:32:35 <mikeperry> maybe we can separate configure vs build vs test failures in the decision to mail
17:33:05 <mikeperry> it seems like if it makes it all the way to running the tests then fails, someone should probably get emailed?
17:33:48 <mikeperry> I guess this can go in the ticket
17:34:52 <mikeperry> like a gcc/clang error is also probably more likely a developer issue than a failure during ./configure (which is probably platform stuff missing due to janky system setup scripts)?
17:35:18 <nickm> +1 on that , for the ticket!
17:35:41 <nickm> isabela, ahf, catalyst, nickm: isabela has reminders for you/us!
17:35:57 <nickm> isabela: what is the timeframe when you need links for the sponsor8 report?
17:36:32 <isabela> end of this week
17:36:34 <isabela> if its ok
17:36:47 <nickm> okay with me
17:37:07 <isabela> also please ping me if you have any questions
17:37:13 <isabela> or need help with anything related to that
17:37:22 <catalyst> isabela: i'm ok with that timing
17:37:27 <isabela> cool
17:37:33 <nickm> wow.  We're pretty far along.  Is there any more we should talk about this week?  Is everybody confident and happy?  Is anybody stuck and in need of help from anyone else?
17:38:30 * isabela is good
17:39:16 <catalyst> ok i guess i'll make the ticket on the dirauth clock skew thing
17:41:08 <arma4> (or we could repurpose the current ticket. it's nearly the right ticket.)
17:41:11 <arma4> (whichever you prefer)
17:41:36 <catalyst> arma4: hm, that could work, but i'd have to look at it again to be sure
17:41:41 <nickm> okay, hearing nothing else -- this could be our first meeting under the new system _not_ to go right up till the end of the hour!
17:42:02 <nickm> everyone, please take time to read everyone else's updates, and ask them questions / offer help, if you have any :)
17:42:09 <nickm> last chance to prevent the meeting ending? :)
17:42:50 <nickm> okay. Thanks, everybody! is feeling really good so far
17:42:52 <nickm> #endmeeting