16:59:11 <nickm> #startmeeting network team meeting, 8 march 2021
16:59:11 <MeetBot> Meeting started Mon Mar  8 16:59:11 2021 UTC.  The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:59:11 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:59:20 <dgoulet> o/
16:59:21 <nickm> ahf is out today so I'll drive the meeting?
16:59:23 <mikeperry> o/
16:59:54 <nickm> https://pad.riseup.net/p/tor-netteam-2021.1-keep is the pad
17:00:09 <nickm> first up -- everybody happy with their kanbans?
17:00:27 <dgoulet> so far so good
17:00:30 <asn> yep
17:01:08 <nickm> great. same here.
17:01:21 <nickm> Next is 0.4.5 and 0.4.6
17:02:09 <nickm> Hm.  "We have an 0.4.5.7-stable" milestone too.  Should I move everything from it into 0.4.5.x-post-stable?  I think I should.
17:03:15 <asn> ok
17:03:48 <nickm> I'm going to the close 0.4.5.7-stable milestone if nobody minds ?
17:04:03 <dgoulet> wait why?
17:04:13 <dgoulet> isn't that milestone ... for the upcoming release?
17:04:35 <nickm> if that's true then we should have it on our pad for 'things to look at'
17:04:50 <nickm> but right now it has no open tickets, and all the tickets are in 0.4.5.x-post-stable
17:05:24 <dgoulet> the original point of that milestone was to track must need for 045.7 and not loose anything important in the -post-stable one
17:05:32 <nickm> ah
17:05:34 <dgoulet> if it is empty it is good, we did our job
17:05:43 <dgoulet> but should remain open until we actually release if anyting comes back...
17:05:59 <nickm> hm.  ok, but if so then next time we make one of these we should put it in our pad
17:06:04 <nickm> sound ok?
17:06:07 <dgoulet> np
17:06:26 <nickm> any must-fix-for-0.4.5.7 stuff left in 0.4.5.x-post-stable?
17:06:53 <nickm> I'm not seeing any blockers that aren't being dealt with...
17:07:06 <dgoulet> not to my knowledge but I'm getting a whole lot of email from "cypherpunks" at the moment so could be some gems in there for 045.7
17:07:22 <nickm> yeah, cohosh is going through our ticket approval queue
17:08:14 <nickm> so we should all look at those post-meeting, I guess :)
17:08:24 <nickm> 0.4.6 also freezes in a week
17:08:37 <nickm> any must-do things before 0.4.6.1-alpha comes out?
17:08:55 <asn> i have a prop#328 branch that i plan to post tomorrow
17:08:56 <dgoulet> asn: is prop328 a possibility for 046 or not really your goal?
17:09:00 <dgoulet> ah
17:09:02 <asn> if not the day after
17:09:07 <asn> it's not the full prop#328
17:09:14 <asn> but it's a decent subset
17:09:22 <dgoulet> ok so 046 or not? :)
17:09:22 <asn> i think it's worth having parts of it early
17:09:25 <asn> yes 046
17:09:28 <GeKo> what is missing?
17:09:38 <asn> what's missing. ok
17:09:50 <asn> 1) 90% CPU utilization detection
17:09:56 <mikeperry> I have some reservations of two different implementations of prop328 being on the network at once.. I think I prefer trying to give branches to relays that saw overload during rob's experiment, so they can run any prototypes while he re-runs
17:09:57 <asn> 2) Control port overload detection
17:10:11 <nickm> request -- if there's something assigned to you that _will not happen_ for 0.4.6, please remove it from the milestone but make sure it's on your kanban?
17:11:00 <asn> 3) Detecting a single DNS timeout. I detect when libevent reports a DNS timeout error, but  I wouldn't know it if libevent had a single timeout and got through with the second try out of three
17:11:18 <asn> I think these 3 things are missing
17:11:25 <dgoulet> MetricsPort?
17:11:28 <asn> totally missing
17:11:35 <mikeperry> if we can get the remaining issues done before 046 deadline, then that's fine for 046.. but we probably should not merge something we expect to change in this case, because it is much harder to manage relays with different versions of this thing
17:11:35 <asn> i meant from the extra-info side
17:11:40 <GeKo> i see
17:11:53 <asn> my plan was to edit the proposal with which parts got to which version
17:12:29 <asn> the only thing i know how to do from the above is the DNS timeout. i can do that, but it's gonna be annoying.
17:12:38 <asn> i have no idea how to measure CPU utilization in a multi-platform way
17:12:47 <asn> and control port overload i still need to study
17:12:51 * nickm has created an 0.4.6.x-stable milestone and an 0.4.7.x-freeze milestone
17:13:32 <GeKo> mikeperry: so you think we should get that faster out by giving folks some branches instead of getting a subset into 0.4.6 now?
17:13:46 <GeKo> how would that work?
17:13:56 <GeKo> we would contact folks pointing to some code?
17:14:18 <mikeperry> we have a full list of relays that had the perf issue. most of them are quintex guy
17:14:45 <mikeperry> and if we see more in the next experiment, we can ask them, too
17:15:41 <mikeperry> we could also version the overload line. that could work, too I guess
17:15:50 <asn> i can do that too
17:16:11 <asn> otherwise we are assuming that the overload metrics will *always* stay the same
17:16:14 <asn> i dont think that's true
17:16:19 <mikeperry> yeah, good point
17:16:22 <GeKo> +1
17:16:41 <asn> ok sounds good. noted.
17:17:26 <nickm> anybody have anything else for 0.4.6?  Remember to move out tickets that won't happen before monday.
17:17:26 <GeKo> so if we'd get the versioning going i'd slightly prefer getting the stuff into 0.4.6 which we currently have
17:17:33 <nickm> (if you can)
17:17:50 <asn> GeKo: yes will do
17:18:02 <GeKo> but i don't have a too strong opinion here
17:19:27 <nickm> if there isn't something else on this, should we move on to announcements and discussions?
17:20:11 <asn> yep
17:20:25 <nickm> the announcements this week are mostly about the freeze (next monday) and upcoming releases (next tuesday)
17:20:35 <dgoulet> great
17:20:55 <nickm> There's another announcement that we're about to get hit by DST, starting next Sunday in the US and two weeks after that in Europe.  Beware!
17:21:33 <nickm> I don't see any new discussion items.  Anybody have something there?
17:21:47 <asn> i'm good!
17:21:53 <nickm> If not it's time for s61 updates!
17:22:32 <nickm> mikeperry / asn: any updates?
17:22:47 <mikeperry> ok they are on the pad. we already discussed the overload proposal a bit. That also has a DNS timeout piece, and a shadow piece
17:23:18 <mikeperry> https://gitlab.torproject.org/tpo/core/tor/-/issues/40312 for tor piece
17:23:30 <mikeperry> https://github.com/shadow/shadow/issues/1137 for shadow, for the curious
17:24:13 <mikeperry> we're doing the prop329 conflux draft discussion this week, yeah? thursday?
17:24:22 <nickm> yup!
17:24:27 <nickm> I'll add a reminder to the announcements
17:24:40 <nickm> want to invite anybody else? if so please let 'em know the time
17:25:02 <mikeperry> idk we could ask ian and other researchers I suppose. what do you think?
17:25:26 <mikeperry> we could also discuss ourselves, and then let them know once we have something more firmed up
17:25:41 <nickm> maybe that 2nd option, since the others haven't had the proposal yet
17:25:46 <mikeperry> yeah
17:25:48 <mikeperry> ok
17:26:09 <nickm> oh, one last reminder: peer feedback is due _this_ friday
17:27:38 <nickm> anything else for this meeting?  If not we can end early and go look at those cpunks tickets :)
17:27:44 <mikeperry> oh yeah I got that wrong last monday
17:27:47 <mikeperry> yeah I am good
17:27:59 <asn> sameh ere
17:29:24 <nickm> ok.  thanks everybody!
17:29:27 <nickm> #endmeeting