16:58:21 <ahf> #startmeeting network team meeting november 9 2020
16:58:21 <MeetBot> Meeting started Mon Nov  9 16:58:21 2020 UTC.  The chair is ahf. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:58:21 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:58:28 <ahf> hello all network team hat wearers
16:58:34 <ahf> and s61 hat wearers
16:58:36 <dgoulet> o/
16:58:36 <jnewsome> o/
16:58:37 <ahf> and other people joining in
16:58:42 <ahf> our pad is at: https://pad.riseup.net/p/tor-netteam-2020.1-keep
16:58:56 <gaba> \o
16:59:38 <GeKo> o/
16:59:52 <ahf> are folks doing alright with the tpo/core part of the roadmap? https://gitlab.torproject.org/groups/tpo/core/-/boards
17:00:18 <nickm> I think so.
17:00:46 <ahf> are we doing alright with ticket review distribution?
17:00:53 <ahf> i see some not delegated
17:01:23 <dgoulet> not done yet indeed...
17:01:39 <ahf> ack ack
17:01:58 <ahf> how are folks doing with 0.4.5 tickets?
17:02:07 * dgoulet working on them
17:02:17 <nickm> Are all the 045 tickets assigned?
17:02:26 <ahf> tor#31020 is not
17:02:34 <ahf> i was wondering about that one
17:02:57 <ahf> it doesn't seem urgent for 0.4.5 though?
17:03:03 <nickm> agreed
17:03:10 <dgoulet> yeah nope
17:03:25 <nickm> The only one I see in core tor that's unassigned is tor#30187
17:03:38 <ahf> that one was merged today?
17:03:42 <nickm> but that's likely to be fixed by the patch you merged, yeah
17:03:51 <dgoulet> it is in 045
17:03:51 <ahf> and it is assigned?
17:03:57 <nickm> it does not seem assigned
17:03:58 <ahf> it's just in the wrong milestone
17:04:04 <ahf> huh
17:04:14 <ahf> oh!
17:04:15 <ahf> you are right
17:04:19 <ahf> i am looking at the zwiebelbot one
17:04:26 <ahf> i have taken that
17:04:30 <nickm> thanks
17:04:38 <ahf> and it's marked backport. sorry for confusion
17:04:49 <nickm> I am stuck on tpo/core/tor#40172; I can't repro it and I am waiting to hear about my suggested AR fix
17:06:11 <ahf> and that is setting AR for their builds?
17:06:16 <nickm> yes
17:06:20 <ahf> ok
17:06:43 * ahf removed milestone from tor#31020
17:06:45 <GeKo> nickm: sysrqb will work on that ticket
17:06:53 <nickm> great
17:06:56 <GeKo> right now we believe it's a tor-browser-build issue
17:07:01 <ahf> fallback-scripts#31020
17:07:57 <ahf> okay, next item is s61
17:08:04 <ahf> mikeperry: you wanna run this?
17:08:57 <mikeperry> ok. so on O1, asn and I are still poking at CBT experiments. trying to fix some timeout bugs
17:09:20 <mikeperry> that's the main network-team relevant piece
17:10:00 <mikeperry> I thought we were ready for review of the fixes last week but I thihnk more experiments are needed
17:10:10 <ahf> is there something that is network-health related here as well, now that geko is also joining if we use this as a gathering point for all of s61?
17:10:12 <ahf> cool
17:10:46 <mikeperry> I can go thru the other objectives but I am not sure if I should do that sync.. maybe we should have that be async on the pad in a diff section?
17:11:06 <ahf> what you prefer here i think for best overview
17:11:30 <mikeperry> ok well for O2 we are still looking at flashflow and sbws work for load balancing
17:11:41 <mikeperry> pastly said he would file tickets for the tor patchset soon
17:11:49 <ahf> nice
17:11:53 <gaba> exitmap is what geko seems to be involved next?
17:12:04 <mikeperry> rob is working on reproducing the long tail bug from https://gitlab.torproject.org/tpo/metrics/analysis/-/issues/33076 in shadow
17:12:54 <mikeperry> for O3, I still need to fix up the congestion control proposal and make a conflux one, but not yet as urgent as other work. and for O4, geko can take that.
17:13:44 <GeKo> mikeperry: if there is more important work i should do than the stuff i wrote on the pad let me know
17:14:14 <GeKo> i have no good overview if someone is blocked by network-health folks improving scanners and such
17:14:22 <dgoulet> quick note: the reachability test would probably be a good health/net overlap piece of work also
17:14:28 <dgoulet> ahf: ^
17:14:33 <ahf> GeKo: maybe the two of us can try to help with the scanner reachability issue?
17:14:47 <ahf> i am not sure what conclusions to draw from my run other than sometimes some relays are unreachable. i think it has to get a bit more systematic
17:14:47 <GeKo> i am fine with that
17:14:59 <ahf> dgoulet: ya, very much, they come from talks in the bw scanner channel
17:14:59 <GeKo> if someone fills me in with details
17:15:00 <mikeperry> GeKo: that is fine for now. dgoulet and I are working on an overload proposal in https://gitlab.torproject.org/tpo/core/tor/-/issues/40158 that you may want to look at
17:15:12 <GeKo> yep
17:15:28 <ahf> GeKo: it's sbws#29710
17:15:33 <GeKo> i can look at that this week
17:15:46 <mikeperry> ahf: yeah the work for the reachability scanner needs to be cross-checked against sbws unmeasured relays. I assume juga can do that when they start
17:15:47 <dgoulet> it is also that if we find a set of relays failing that test, we can start contacting them and attempt fixing
17:15:59 <ahf> we could extend what i have and build something we can run from a node and have dgoulet's prometheus system collect the info so we can keep this in track over time
17:16:00 <GeKo> ahf: ack
17:16:19 <ahf> especially with dgoulet's upcoming changes to tor on oom/cpu/socket info in descriptors it could be good if we had something to look at
17:16:25 <GeKo> +1
17:16:25 <ahf> mikeperry: yes
17:17:08 <ahf> ok, very good. do we have more s61 things to dive into tonight?
17:18:25 <mikeperry> those are the main ones from the planning meeting
17:18:31 <mikeperry> I think I am good
17:18:35 <ahf> excellent!
17:18:45 <ahf> we have no new things from other teams that want us to look at
17:18:56 <ahf> do we really not have any discussion items?
17:19:05 * dgoulet is good
17:19:06 <ahf> that is a new thing
17:19:28 <ahf> well let's call the meeting then and get back to what we were doing before!
17:19:31 <ahf> thanks all o/
17:19:32 <ahf> #endmeeting