16:58:52 <ahf> #startmeeting Network team meeting, 16th May 2022
16:58:52 <MeetBot> Meeting started Mon May 16 16:58:52 2022 UTC.  The chair is ahf. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:58:52 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:58:55 <ahf> hello everyone
16:58:59 <Diziet> o/
16:59:03 <nickm> hihi
16:59:12 <jnewsome> o/
16:59:16 <ahf> our pad is at https://pad.riseup.net/p/tor-netteam-2022.1-keep
16:59:21 <GeKo> wow, early meeting start today
16:59:27 <mikeperry> o/
16:59:50 <ahf> yes, i am having a bit of a chaos day today, so i was also off by a few seconds for the meeting
17:00:33 <ahf> did someone go over our pad and fix all my incorrect usages of is vs are ? :-s
17:00:53 <ahf> how are folks doing with their boards: https://gitlab.torproject.org/groups/tpo/core/-/boards ?
17:01:11 <dgoulet> hi
17:01:51 <nickm> I believe okay.  Finaally making progress on arti#439.
17:01:56 <dgoulet> board is fine
17:01:57 <nickm> hoping to wrap up arti#329 soon
17:02:04 <nickm> then maybe finally something else :)
17:02:11 <ahf> looks like people have become very good at using the Q2 label! thank you all for that
17:02:53 <ahf> nice
17:02:59 <ahf> i don't see anything obviously off
17:03:06 <ahf> dgoulet: anything on releases for tor.git ?
17:03:45 <dgoulet> not much no, still want to go soon for 045 and 046 but might not be this week due to nsec.io
17:03:54 <dgoulet> that is about it
17:04:21 <ahf> sounds good
17:04:24 <ahf> enjoy nsec!
17:04:57 <ahf> don't see anything incoming from other teams
17:05:00 <juga> o/
17:05:19 <ahf> core/tor#40597 is funky
17:05:24 <ahf> juga: o/
17:05:33 <ahf> juga: you have something you want to talk about or you arrived? :-p
17:06:09 <juga> arrived, sorry :)
17:06:16 <ahf> ah, good good, just wanted to be sure
17:06:29 <ahf> i see no reminders or announcements
17:06:34 <ahf> mikeperry: you wanna talk about s61?
17:06:53 <mikeperry> yeah yeah
17:07:20 <mikeperry> so overall everything is looking OK with respect to congestion control deployment
17:07:47 <mikeperry> we put out the blog post and tor-relays post to get exits to upgrade
17:07:53 <mikeperry> https://blog.torproject.org/congestion-contrl-047/
17:08:00 <mikeperry> https://lists.torproject.org/pipermail/tor-relays/2022-May/020535.html
17:08:17 <mikeperry> about 1100/1500 exits have upgraded according to ggus
17:08:42 <mikeperry> I am curious what the fraction of exit consensus weight this is
17:08:42 <ahf> nice
17:09:06 <mikeperry> we're not seeing much increase in bandwidth consumption.. only advertised bandwidth
17:09:16 <mikeperry> see: https://metrics.torproject.org/bandwidth-flags.html
17:09:16 <ahf> i could imagine higher than 2/3 of the total weight if these are the people that track our releases well
17:09:36 <mikeperry> this likely means that our users are, for the most part, not using Tor 0.4.7
17:09:47 * ahf nods
17:10:36 <mikeperry> hiro is working on upgrading the onionperf instances to 0.4.7. last I heard, those will be new instaces, so the difference is clear there
17:10:39 <mikeperry> or will be
17:10:46 <hiro> yes
17:11:00 <mikeperry> hiro: what is the ticket for that?
17:11:11 <hiro> https://gitlab.torproject.org/tpo/network-health/metrics/onionperf/-/issues/40040
17:11:25 <jnewsome> is 0.4.7 in a released TBB yet?
17:11:37 <GeKo> alpha
17:11:42 <mikeperry> jnewsome: no. alpha TBB only
17:12:25 <mikeperry> the next scheduled TBB stable is May 31
17:12:38 <mikeperry> but maybe we get lucky and there will be firefox 0day ;)
17:12:44 <ahf> it is worth trying the alpha right now. i have gotten some nice downloads speeds there
17:12:57 <juga> mikeperry: with a quick script right now i get exits with 2 in flowctrl weight/ total consensus weight:
17:12:58 <jnewsome> nod. so probably as expected then? IIUC most usage is via TBB?
17:13:00 <juga> >>> 30305090 / 117741140
17:13:03 <juga> 0.25738743484223103
17:13:06 <ahf> i'd think so
17:13:15 <GeKo> yeah
17:13:23 <mikeperry> juga: so only 25% of the consensus has upgraded to 0.4.7?
17:13:26 <mikeperry> of exits?
17:13:45 <juga> weight wrt total net
17:14:04 <mikeperry> oh can you do weight vs total exit weight?
17:14:08 <ahf> you have the same number with exit only?
17:14:10 <ahf> ye
17:14:19 <juga> ok, will do
17:15:58 <mikeperry> there's some trickyness with the sbws upgrade to congestion control, but juga is looking at that. https://gitlab.torproject.org/tpo/network-health/analysis/-/issues/36 is the analysis ticket
17:17:09 <juga> mikeperry: yeah, hopefully last patch will make all cases work
17:17:47 <mikeperry> eta,nickm,dgoulet: I have a test vector WIP branch for clock heuristics, RTT, and plumbing for vegas: https://gitlab.torproject.org/tpo/core/tor/-/issues/40443#note_2803259
17:17:55 <mikeperry> for arti impl
17:18:10 <juga> with new quick script looks like exits with flowctrl2 weight is 0.8 of all the exits
17:18:30 <mikeperry> \o/. not bad for 3 weeks since stable
17:18:35 <ahf> nice, that better than i would have thought
17:18:53 <juga> (30305090KB / 38240006KB)
17:19:01 <nickm> nifty
17:19:19 <ahf> nice
17:19:43 <ahf> it is good we have so much reachout to these operators. it seems to be making our life a lot easier
17:20:17 <mikeperry> GeKo: any additional signs of overload or any other issues wrt the upgrade visibile so far?
17:20:28 <mikeperry> (or any other network health updates?)
17:20:43 <GeKo> the overload monitoring looks good so far
17:21:03 <GeKo> i helped two operators in #tor-relays with 0.4.7.7 overload last week
17:21:10 <GeKo> pointing to the blog post etc.
17:21:20 <GeKo> but otherwise there is not much we got
17:21:35 <GeKo> toralf reported seeing a new notice line on his bridges:
17:21:36 <GeKo> Unexpected path length 4 for exit circuit <snip>, purpose 5
17:21:43 <GeKo> is that expected?
17:22:00 <GeKo> that's nothing that showed up pre 0.4.7.7 i think
17:22:00 <mikeperry> oh my. the maze
17:22:09 <GeKo> The Maze?
17:22:12 <GeKo> :)
17:22:30 <GeKo> i leave that to you filing a ticket then, i guess
17:22:35 <mikeperry> circuit construction maze. it should not be able to build 4 hop circuits, but it does.
17:22:47 <GeKo> if that's indeed to be a ticket
17:22:54 <mikeperry> I noticed this with onion services in the picture.. bridges would be a new one
17:23:11 <GeKo> it's Real
17:23:34 <GeKo> mikeperry: i could need some help with replying to sebastian on the dir-auth list
17:23:44 <GeKo> even better if you just replied :)
17:24:17 <GeKo> oh, and someone needs to decide what to do with the 0.4.7.7 compilation failure on openbsd andreas hit
17:24:26 <GeKo> dgoulet: is that for a ticket?
17:24:31 <GeKo> i guess so?
17:24:58 <ahf> ... what was the failure? it has been added to openbsd ports
17:24:58 <GeKo> otherwise nothing from my side
17:25:46 <GeKo> ahf: https://paste.debian.net/1241089/
17:26:00 <GeKo> he compiles against libressl i think
17:26:02 <dgoulet> it is LibreSSL...
17:26:22 <GeKo> but it's blocking him from upgrading the dir-auth
17:27:12 <ahf> is it an up-to-date openbsd installation? have he looked in their ports file if the tor build does something special there like specifying a different libssl/libcrypto ?
17:28:07 <GeKo> yeah, it's up-to-date openbsd 7.1
17:28:52 <GeKo> not sure about the ports related question
17:30:02 <ahf> it's like this: https://github.com/openbsd/ports/blob/master/net/tor/Makefile#L20-L24
17:30:19 <ahf> it looks like they are setting a specific ssl-dir as part of their configure argument
17:31:49 <mikeperry> so if they use libressl it finds that and breaks?
17:32:00 <GeKo> seems so
17:32:03 <ahf> that's one guess, but i am not 100% sure
17:32:18 <ahf> it doesn't look the port pulls in openssl here
17:32:26 <ahf> unless wantlib = ssl means openssl, but i doubt that
17:32:59 <ahf> err, actually
17:33:05 <ahf> they carry around a lot of patches it seems: https://github.com/openbsd/ports/tree/master/net/tor/patches
17:33:26 <ahf> https://github.com/openbsd/ports/blob/master/net/tor/patches/patch-src_lib_crypt_ops_crypto_rsa_openssl_c seems related
17:33:57 <GeKo> right
17:34:17 <GeKo> okay, i can continue the convo with andreas
17:34:24 <ahf> maybe he should be hooked up with gman for some support here? it seems like stuff is packaged for their ports tree, but maybe he wants to compile it by hand himself?
17:34:33 <GeKo> yeah
17:34:54 <GeKo> if we can't solve it i'll get gman into the thread
17:35:27 <ahf> sounds good, feel free to cc me too. i still have an openbsd install somewhere, but i am especially a bit sensitive to deal with the libressl/openssl crap situation the openbsd gang have put themselves in..
17:35:49 <ahf> but can try a build there locally on a openbsd vm if he can't get it to work ta all
17:36:02 <GeKo> understandable, thanks
17:36:14 <ahf> np! thanks for following up on that
17:36:19 * ahf didn't even know one of the dirauth's was openbsd
17:36:28 <ahf> anything else we want to talk about today? :-)
17:36:44 <mikeperry> geko: wrt sebtasian, let's discuss after
17:37:07 <mikeperry> I am good otherwise
17:37:16 <GeKo> ack
17:37:39 <GeKo> (although i need to get some food first)
17:38:15 <ahf> arti people, remember we have public meeting on wednesday if we have something we want to talk about there 8)
17:38:22 <ahf> and dgoulet is out for nsec a lot this week
17:38:30 <ahf> otherwise, let's return to the non-work setting
17:38:33 * ahf also need food
17:38:34 <nickm> ahf: I think it would be neat to check in on ongoing projects for arti meeting
17:38:34 <Diziet> ahf: Noted.  FYI I will be afk on Thursday.
17:38:44 <Diziet> nickm, eta ^
17:38:49 <nickm> ack
17:38:53 <ahf> Diziet: perfect! i saw you already updated the calendar, that is good
17:39:03 <ahf> nickm: what is ongoing projects here? ongoing tasks?
17:39:07 <Diziet> Someone told me I should do that, so :-)
17:39:12 <ahf> Diziet: ya, is very good
17:39:26 <ahf> nickm: is this something we can do in our 1:1 tomorrow?
17:39:41 <Diziet> nickm: ongoing projects> +1
17:40:09 <Diziet> I was hoping for a task ~with moving actual bytes around" as you put it ...
17:40:19 <nickm> ahf: I meant, at the arti meeting, it would be neat to  go through the stuff for 1.0 and say how it's all going
17:40:28 <nickm> Diziet: woo
17:40:52 <ahf> ah
17:40:53 <Diziet> nickm: Too many of my existing things are "have done all the tractable bits"
17:41:01 <ahf> sounds good, so something we do in the meeting, not something we have to prepare for the meeting?
17:41:18 <nickm> yeah
17:41:27 <ahf> sounds good, let's do that
17:41:30 <ahf> ok, thanks all o/
17:41:34 <ahf> #endmeeting