16:58:36 <ahf> #startmeeting Network team meeting, 26th September 2022 (take two)
16:58:36 <MeetBot> Meeting started Mon Sep 26 16:58:36 2022 UTC.  The chair is ahf. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:58:36 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:58:39 <ahf> hello o/
16:58:43 <nickm> hi all!
16:58:44 <ahf> 2nd attempt to get the time right
16:58:56 <nickm> i think you've done it!
16:58:57 <ahf> our pad is at https://pad.riseup.net/p/tor-netteam-2022.1-keep
16:58:58 <Diziet> o/
16:59:07 <ahf> yeah, thanks lol. could just try once an hour until other people showed up XD
16:59:29 <mikeperry> o/
16:59:47 <ahf> okay, let's start with the board
16:59:52 <juga> o/
17:00:34 <ahf> i don't see anything entirely off there. the arti gang seems to be diving into the bridges/pt work for arti 1.1.0 and the c-tor things looks right too
17:00:35 <GeKo> hi
17:00:46 <nickm> nothing here. I should at some point break up the 1.1 arti stuff into a bunch of tickets instead of just having arti#69
17:01:19 <ahf> nickm: sounds good. i saw you and Diziet was coordinating a bit earlier on irc, that is very good. then eta can join when she's back \o/
17:01:24 <ahf> on tor release update: we may have to roll a release update with a bump of fallback dirs bump
17:02:06 <nickm> arti also ships the fallback list; please make sure we get it too
17:02:10 <ahf> roger was very interested in this right after limerick, but i am not entirely sure when we decide it is worth the efforts of rolling a full set of releases there. we currently have 1:3 fallback dirs that are not available. i will talk with GeKo this week about this and figure out if we need to roll it out
17:02:13 <ahf> ya
17:03:33 <ahf> i do not see anything incoming from other teams. roger expressed interest in diving into some of the PT stuff (i think most of those were unrelated to any of the s30/s28 tickets we have today) that has been bothering anti-censorship team for a while, so i will try to coordinate a bit with him on what he was gonna dive into there
17:04:21 <ahf> gaba and i will be talking a bit about Q4 roadmap this week, so expect something on that next week
17:04:50 <ahf> otherwise i think we can jump to s61 things with mikeperry
17:05:19 <mikeperry> ok
17:06:01 <mikeperry> so the main thing is that the network seems to be getting hit by yet another dos attack of some kind.. started on Sept 13th
17:06:37 <mikeperry> which I believe pre-dates the iranian traffic spike. but it is impacting exit performance more than onions this time, which is odd
17:06:41 <mikeperry> and unfortunate timing
17:07:49 <mikeperry> geko and I changed a queue limiting parameter to try to reduce memory pressure on guards a bit more
17:07:49 <ahf> we are pretty deep in the DoS inception right now
17:09:05 <mikeperry> I ran some sims over the weekend to test more frequent updates of congstion control in steady state, which actually also reduces queue pressure according to shadow
17:09:17 <mikeperry> so that is good. we can try those new params this week
17:09:36 <GeKo> ping me when you think it's the right time to push those changes
17:10:02 <ahf> nice
17:10:02 <mikeperry> ok
17:11:00 <mikeperry> have you noticed anything more about the dos? have we heard anything from exit operators? I am wondering if it is another DNS thing like what we saw before that messed up unbound
17:13:53 <GeKo> i have nothing heard from exit operators
17:14:08 <GeKo> but it does seem to affect our reachability tests badly
17:14:33 <GeKo> many exits failing
17:14:39 <GeKo> with connection timeout etc.
17:14:49 <GeKo> which is probably not surprising...
17:15:08 <mikeperry> yeah, this one could be a dos through the network this time
17:15:17 <GeKo> not so much unbound crashing though
17:15:35 <GeKo> so, i suspect it's slightly differnt this time
17:15:42 <mikeperry> ok
17:15:49 <GeKo> still need to evaluate this week's scan results though
17:18:23 <mikeperry> alright.. I still need to follow up on the sbws cc changes and other stuff too, but man, things keep happening
17:19:09 <ahf> sounds good
17:19:20 <ahf> for some value of good of course
17:20:27 <ahf> do we have anything else for today?
17:21:46 * jnewsome got nuthin
17:22:03 <GeKo> <- fine, too
17:22:04 * ahf neither
17:22:54 <ahf> mikeperry: you good?
17:23:21 <mikeperry> yeah, for now. I will update the pad for our monthly meeting next week
17:23:23 <mikeperry> 15utc
17:23:28 <mikeperry> oh wait no
17:23:34 <mikeperry> gaba asked for 18utc
17:23:41 <ahf> huh
17:23:56 <ahf> i have 1:1 at 18 utc with nickm
17:24:29 <ahf> 15 was picked back then specifically such that it didn't collide with any of the other team meetings we have on mondays :s
17:24:37 <nickm> (next week yeah?)
17:24:40 <ahf> yes
17:24:48 <gaba> I can not make 15utc ahf
17:25:07 <mikeperry> yeah, in the last one gaba asked to make it later. my notes say ppl were fine with 18utc but I guess not everyone was there that time either
17:25:26 <ahf> ok... let's coordinate this out of this meeting
17:25:34 <mikeperry> ok
17:25:37 <ahf> thanks all o/
17:25:40 <ahf> #endmeeting