16:58:56 <ahf> #startmeeting Network team meeting, 23rd May 2022
16:58:56 <MeetBot> Meeting started Mon May 23 16:58:56 2022 UTC.  The chair is ahf. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:58:56 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:59:02 <ahf> no more holding your breath!
16:59:02 <nickm> hi all!
16:59:03 <Diziet> o/
16:59:05 <ahf> hello all o/
16:59:09 <mikeperry> o/
16:59:35 <ahf> https://pad.riseup.net/p/tor-netteam-2022.1-keep is our pad
17:00:05 <jnewsome> o/
17:00:16 <GeKo> hi
17:01:05 <ahf> how are folks doing with their boards:  https://gitlab.torproject.org/groups/tpo/core/-/boards ?
17:02:22 <juga> o/
17:03:01 <nickm> doing ok. trying to wrap up as many things as possible before moving forward
17:03:22 <Diziet> OK I think.  My loose ends are getting smaller.
17:03:43 <ahf> nice
17:04:47 <ahf> things looks OK to me - can't spot anything off
17:05:09 <ahf> ok, gonna skip release topic since dgoulet is out with public holiday today i think
17:05:59 <ahf> don't see anything incoming from other teams. also don't think we generated any new tickets from the relay ops meeting saturday
17:06:14 <ahf> if people have the time (they are absolutely not mandatory) then these meetings are VERY nice to meet some of our community folks
17:06:19 <nickm> ahf: I have a couple of questions for you that don't need to be at this meeting. Can I PM you after?
17:06:25 <ahf> and i think everybody from the team will be able to answer some of the questions there
17:06:29 <ahf> nickm: sure!
17:06:35 <nickm> ok, will do :)
17:06:59 <ahf> ok! two announcements
17:07:09 <ahf> [2022-05-23] [ahf] Cutoff day for Thursday meeting agenda items is Wednesday at 18 UTC :-)
17:07:29 <ahf> so since we now need to add stuff for our thursday meetings to the pad we use in this meeting too, please make sure to do so before Wednesday at 18 UTC so people can plan their thursday after that
17:07:35 <ahf> i will try to remind people on wednesday on this
17:07:52 <ahf> this means if we have nothing to talk about, you don't have to join the meeting (and thus context switch)
17:08:43 <ahf> i am away wed+thu+fri this week due to public holiday (and me taking off an extra day there)
17:08:52 <ahf> i am online, but much less stable than normally
17:09:07 <nickm> next monday is a US public holiday, I believe
17:09:28 <ahf> eta is also going afk later this week for some real holiday, so we need to make sure to sync between the arti gang and remove eta from the rotation (she can do that by marking herself as busy on gitlab)
17:09:31 <ahf> i will prod her about that
17:10:00 <ahf> nickm: ah, then our next meeting is on tuesday at 17
17:10:06 <ahf> thank you for telling me that. i will update our schedule
17:10:25 <ahf> ok!
17:11:21 <ahf> so one discussion item: i don't know if we have remembered this, but when we talk about the arti relay functionality, it would be very good if we can think openmetrics/prometheus access into it from the start. it seems like the relay ops is very much aware of the metricsport functionality in C Tor
17:11:48 <ahf> anything else? otherwise mike can do s61
17:14:00 <mikeperry> ok. Good News Everyone! the Firefox 0day gods smiled upon us, and Tor Browser Stable 11.0.13 is coming out early, with 0.4.7.7, today/tomorrow
17:14:58 <mikeperry> juga says our exit upgrade rate was 78% by consensus weight last week; 80% today
17:15:26 <mikeperry> this means 80% of the time, those 11.0.13 TBB stable users will get a fast congestion control circuit
17:15:35 <ahf> LOL! don't tell the browser folks the thing with the gods smiling
17:16:10 <GeKo> they believe in the wrong ones
17:16:28 <mikeperry> lol
17:16:38 <hiro> lol
17:16:43 <ahf> XD
17:16:56 <mikeperry> anyway 80% is pretty good, though we will want to keep an eye on that consensus percentage and start poking that long tail if it stops increasing
17:17:27 <ahf> very exciting to get it out in people's hands
17:18:29 <mikeperry> hiro also got op-*7 onionperf instances up, but it looks like a different build issue caused it to still use 0.4.6
17:18:46 <hiro> yes I am rebuilding it now... almost done
17:18:49 <mikeperry> which is why these numbers look the same between op-*6 and op-*7:
17:18:51 <mikeperry> https://metrics.torproject.org/onionperf-throughput.html?start=2022-05-20&end=2022-05-23&server=public
17:18:53 <hiro> then I will delete the old data
17:19:30 <mikeperry> yeah, so going forward, the op-*7 instances will be our metrics source for what 0.4.7 clients experience, and op-*6 will bewhat 0.4.6 clients experience
17:19:36 <mikeperry> which is very handy to have for this period
17:20:16 <donuts> lmao
17:20:29 <mikeperry> juga: how is the next sbws 1.5.x release for the sbws looking?
17:20:30 <donuts> (@ the browser gods)
17:21:06 <mikeperry> (oh is juga still afk?)
17:21:28 <GeKo> don't think so
17:21:33 <juga> mikeperry: is done
17:21:39 <juga> *it
17:22:03 <juga> already in debian testing and bullseye backports queue
17:22:46 <mikeperry> nice.. so we could have sebastian upgrade this week, and if we don't see complete madness, upgrade the rest next weekish
17:22:59 <juga> yup
17:23:25 <mikeperry> juga,GeKo: again the thing we should see is the "Consumed Bandwidth" goes up as TBB users upgrade - https://metrics.torproject.org/bandwidth-flags.html
17:23:34 <mikeperry> assuming that happens, we can upgrade all sbws after that
17:23:38 <GeKo> right
17:23:44 <GeKo> sounds good
17:24:08 <GeKo> we do have some increasing overload starting this week
17:24:15 <mikeperry> I guess we will also learn if TBB users are actually the primary consumers of network bandwidth...
17:24:18 <GeKo> i am not sure if that is spoiling the party
17:24:24 <GeKo> https://share.riseup.net/#VzMV9F-VuP5Y-X8OJvydDg
17:24:41 <GeKo> it seems that's to a large extend related to overload at guard relays
17:25:11 <GeKo> mikeperry: should we look closer at that?
17:25:20 <mikeperry> hrmm.. every time Guard overload happens I think "Onion Service DoS attack"
17:25:26 <mikeperry> that has been the historical cause
17:25:32 <GeKo> i mean i could check whether that's actually related to 0.4.7.7 upgrades (which i suspect)
17:25:40 <GeKo> yeah
17:26:02 <GeKo> but maybe there is something else we want to check wrt cc upgrade
17:26:04 <mikeperry> could be some high-use onion service upgraded and its guards are being hit.. but what would be hitting them at this point?
17:26:26 <GeKo> right
17:26:42 <GeKo> not sure
17:26:57 <mikeperry> did we hear anything from any of these relays yet?
17:27:10 <GeKo> not that i know
17:27:57 <mikeperry> ok yeah well we should keep our ears to the grindstone on that or whatever
17:28:12 <GeKo> yep, will do
17:29:29 <mikeperry> hrmm.. anything else? I think that covers it from my side
17:29:37 * ahf good
17:29:48 * GeKo too
17:29:53 * juga too
17:30:37 <ahf> let's call it then!
17:30:39 <ahf> thanks all o/
17:30:41 <ahf> #endmeeting