16:59:13 #startmeeting Network Team meeting, 24 october 2022 16:59:13 Meeting started Mon Oct 24 16:59:13 2022 UTC. The chair is ahf. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:59:13 Useful Commands: #action #agreed #help #info #idea #link #topic. 16:59:15 o/ 16:59:16 hello hello, welcome 16:59:36 let's see who's here today 16:59:50 o/ 17:00:17 o/ 17:00:29 hey! 17:00:31 o/ 17:00:55 o/ 17:00:57 o/ 17:01:02 (thanks for the ping ahf) 17:01:06 pad is at https://pad.riseup.net/p/tor-netteam-2022.1-keep 17:01:08 np! 17:01:48 okay, so gaba and i went pretty deep in the paint last week on the arti board and moved a lot of things around, hope that didn't cause any confusion. when we are done with this 1.1.0 crazyness we can take a more serious look at the issues we have there and how to get the board setup for 1.2.0 17:02:02 are folks otherwise doing fine with their boards: https://gitlab.torproject.org/groups/tpo/core/-/boards ? 17:02:27 ahf: I have been blissfully ignorant due to paying attention only to PT MRs (mine and others') and the PT ticket arti#69 17:02:43 same here :) 17:02:51 thanks to everybody for reviewing my stuff btw 17:02:54 (plus of course the 3 visits from Dell and the dead washing machine and ...) 17:03:00 yeah, i said to gaba that if we ignore arti#69 they won't notice us messing around with this :S 17:03:11 ahf: lol v truje 17:03:15 :D 17:03:26 dgoulet: should i poke roger about https://gitlab.torproject.org/tpo/core/tor/-/issues/40622 ? is that something we need for this release or? 17:03:55 it hasn't seen activity in this quarter 17:03:59 ahf: wouldn't hurt to have it for sure! 17:04:13 ok, will do 17:04:16 excellent 17:04:21 otherwise i think stuff looks good 17:04:45 ahf: yes, please 17:04:45 dgoulet: anything to releases here? 17:05:28 we have this weird situation now that we need work around the eol for 0.4.6x for moria because it's still on some 0.4.6.x alpha due to that issue :( 17:05:28 so not much, we are still very close to a new release but I found many HS problems last week that need to be merged 17:05:41 we have a long list of things to get in for next release so dunno if this week but definitely "soon" 17:06:23 GeKo: ah, interesting. let's hope the moria operations department can talk with the assignee of the ticket then 8) 17:06:27 that is about it... "we are working on it" :) 17:06:31 otherwise we can take a look at the cherry-picked patch 17:06:37 dgoulet: excellent 17:07:30 for external teams, we got https://gitlab.torproject.org/tpo/core/tor/-/issues/40698 on our lists from nethealth 17:07:51 dgoulet: is this something on your radar or do i need to do something there? 17:07:54 I've given a +1 to roger's patch there 17:07:57 it's good to merge 17:08:08 ah, perfect 17:08:17 then nothing to do there. even better 17:08:21 I've also approved it 17:08:34 perfect perfect 17:08:57 dgoulet: are you okay with doing the merges on approved tor patches? I'm afraid it is getting likelier and likelier that I will screw up as I try to do so, for so long as they're on gitolite :/ 17:09:18 yes ofc, I plan to do the merge 17:09:21 thanks! 17:09:22 perfect 17:09:25 arti gang, do we want to do an irc checkin this week like you have done the other weeks in the current madness we are in? 17:10:22 yes, wfm 17:10:34 right when the s61 part of this meeting begins? 17:10:50 well, i was thinking tomorrow at 14 or something like we did last week 17:10:57 ahhhh 17:11:01 even smarter 17:11:01 but you are also welcome to do it at that time, then i will just be in here instead 8) 17:11:09 i cannot do two things in parallel on irc :-/ 17:11:24 Diziet, eta, nickm: 14:00 UTC tomorrow check-in ? 17:11:28 on irc 17:11:28 OK 17:11:33 tomorrow @14UTC is great 17:11:51 sure 17:11:56 perfect 17:12:00 This replaces the bbb which I have in my calendar 17:12:20 i think we can do that, yeah. i am not gonna start anything new up that i had planned while you are all dealing with the current project 17:12:31 if we need to do bbb, we can do that more ad-hoc 17:12:35 ok, good 17:12:50 people who needs to do s61 can stay, everybody else can go do the things they were gonna do before they got interrupted by the meeting 17:12:58 mikeperry: wanna go for s61 stuff? 17:13:33 cool, ttfn 17:13:34 yeah, so last week was the report. I need to check that over again 17:13:39 * Diziet returns to the compiler's complaints 17:14:07 hiro had some questions about dos dates for the indicators. I suggested she ask geko for more, but the main was was September 13th onward 17:14:49 yeah, this is sorted out 17:14:58 juga: I had a couple questions in reply to https://gitlab.torproject.org/tpo/network-health/sbws/-/issues/40146#note_2844634 17:15:13 we only have like 2 really good weeks at the begin of july... 17:15:28 mikeperry: i saw, just haven't check what you ask 17:15:30 the others miiiight be usable with some caveat, though 17:15:32 ye 17:15:35 yet 17:15:36 the beginnning of September was also ok, iirc 17:15:49 but the network had already suffered guard loss by then 17:16:33 juga: the slow start thing is very weird.. it should not switch back and forth ever.. that is either a tor bug, or a bug in sbws event tracking for CIRC_BW 17:17:10 mikeperry: i'll keep a log for a bit more time to see whether it happens again and i'll let you know 17:17:29 i also started code to listen for CIRC events, none so far 17:17:32 it's just getting worse from mid-july ramping up slowly (that's for onions): https://metrics.torproject.org/torperf.html?start=2022-05-01&end=2022-10-24&server=onion&filesize=5mb 17:18:07 the non-onion data is similar until 09/13 17:18:31 but as i said it might still be good enough to do at least something with it 17:18:38 ah, yeah 17:20:05 very sad. you read these metrics and you could conclude s61 is destroying the tor network 17:20:25 :/ 17:20:26 right 17:22:38 so this week, I will be reviewing dgoulet's onion service and queue fixes. the fix to https://gitlab.torproject.org/tpo/core/tor/-/issues/40680 along with the upload changes jnewsome did can help us investigate queue use a bit closer in shadow 17:23:07 jnewsome did the shadow migration issues from last week get resolved? are both large and very-large working for 10% sims yet? 17:23:11 (oh yeah that is ready to be merged. Been running on our relays as well successfully) 17:23:20 mikeperry: i think so, yes 17:23:50 dgoulet: you said it was getting triggered for 1250 on the test relays? do we want an option to disable the banning by setting it to 0? 17:24:35 for that, we can disable DoS circuit creation defense 17:24:39 or introduce a new param 17:25:02 or just make it use "0" as disabled 17:25:08 exactly yeah ^ 17:25:15 did we do that in the patch or we need an update? 17:25:27 but 0x7FFFFFF is also effectively disabled.. but it just keeps tracking.. which is maybe good, actually? 17:25:51 yeah basically 17:26:15 ah ok no, we need to change this so 0 disable it. I'll make a note 17:27:09 I expect most of my time this week will be code reviews and testing this stuff in shadow 17:28:38 nice 17:30:05 anything else? 17:30:13 not from me 17:30:20 from me neither 17:30:25 next week is still in october, so our next bbb here is in 14 days from today 17:30:26 * dgoulet is good 17:30:58 let's wrap it up then. thanks all! 17:31:02 #endmeeting