16:57:50 #startmeeting network team meeting, 1st of march 2021 16:57:51 Meeting started Mon Mar 1 16:57:50 2021 UTC. The chair is ahf. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:57:51 Useful Commands: #action #agreed #help #info #idea #link #topic. 16:57:54 hello hello 16:58:06 our pad is at: https://pad.riseup.net/p/tor-netteam-2021.1-keep 16:58:26 david is afk for a bit and will miss the meeting, but otherwise we can get started 16:58:38 hello all 16:58:44 hello asn 16:58:46 hihi 16:59:57 let's get started. it's first meeting of the month 17:00:17 are people doing alright with roadmappy work? 17:00:31 i think so, roadmappy-wise 17:00:34 all good here 17:00:50 * ahf hopes to crunch through his ios things this week and end this so he can get ~2 days back per week after outreachy and this is done 17:00:54 great 17:01:02 remember this week is where we need to do peer review submissions 17:01:06 the outreachy interns have been great this time around 17:01:25 ahf: when is peer review due? Friday? 17:01:30 friday the 5th 17:01:53 nickm: yes, they have. i am so impressed and it has also been so much fun to be part of. i am gonna miss that 17:02:18 * asn writing status report 17:02:30 and we also have the hackathon on thursday. let me know if i need to do antyhing different than i usuall do. not many have signed up this time i think 17:03:02 dennis jackson will join us 17:03:14 all: If you asked me for peer feedback and I did not just message you to say I am on track, I have completely forgotten about doing your peer feedback. Please remind me :) 17:03:17 and we plan on recording the first 1h of the session this time, so we have it for future need 17:03:24 ack! 17:04:43 anything we need to talk about around 0.4.6-freeze and 0.4.5-post-stable ? 17:04:58 i think we looked a bit at it thursday, so maybe not so much needed to do here? david did some clean up 17:05:02 045-post-stable is all fine, and okay with me. Better check in with dgoulet about that after 17:05:19 i will poke him later when he returns. he is afk for a bit 17:05:19 046-freeze has unassigned tickets IIRC 17:05:32 we should see if any are truly necessary for the freeze, and assign them if so 17:06:35 hm, yes 17:06:41 9 issues 17:07:35 perhaps we should spend a short amount of time post-hackathon on thursday to quickly go over those 9 issues? 17:07:44 it will take forever here on irc and take 10 min in audio 17:08:05 post-hackathon we are usually wrecked 17:08:13 but we can try 17:08:18 ok 17:08:30 should we try to find a time on wednesday for a quick mumble call? 17:08:35 not sure... 17:08:42 perhaps post-hackathon is easier logistics! 17:08:49 i don't want to allocate more meetings to people right now, we have so many meetings going on right now 17:08:54 agreed 17:08:56 sounds good 17:08:57 people are doing interviews and prep for interviews 17:09:07 and last time we managed to get a bit done post hackathon, but i agree it's not ideal 17:09:24 sounds good let's do it 17:09:28 ack, good good 17:09:49 let's also look at those tickets independently and maybe take ones we think that we should do. 17:10:02 yes, i think that is a good point 17:10:15 if there are some we can see now we absolutely need to do and want to take lead on, just assign them 17:10:32 ok 17:10:40 anything we need to talk about regarding: https://gitlab.torproject.org/tpo/core/team/-/wikis/NetworkTeam/CoreTorReleases 17:10:45 we talked a bit about it thursday already 17:12:05 I have a discussion item to check in whether we still think 0457 is necessary before Mar 16 17:12:23 also TB just updated their release schedule, so we should check whether we should change any of our dates 17:12:25 we can take that now if it fits 17:12:40 ok 17:12:50 what's on march 16? 17:13:04 046 freezes on 15 March 17:13:23 and we should do a set of releases to fix the DoS in TROVE-2021-001 17:14:16 and you are thinking of waiting with 0.4.5.7 until ~mar 16 ? 17:14:24 i think that's ok? 17:15:13 I think so too but I want dgoulet's opinion about the stuff he's been fixing in 045. 17:15:27 once he's back he can let me know :) 17:15:36 i have no strong objections to that. i think the relevant people have info on TROVE, but maybe dgoulet can prod you if he thinks otherwise? 17:15:53 ok 17:15:53 yeah, please let him know. he is not in the channel, but he can read the meet logs in meetbot via http 17:16:06 ack 17:16:26 ok, so next item is "team issues" 17:16:29 we have a new one! 17:16:34 https://gitlab.torproject.org/tpo/core/team/-/issues/17 17:16:58 it's a great idea; does it have an owner? 17:17:05 it's from juga, it seems like it's an open discussion still, but if people have things they want to throw in, we should do so 17:17:47 i think maybe figuring out the goal is before getting an owner? what should this lead to? we want a wiki entry on what we think of stuff like this? 17:18:05 i have the feeling we have different ideas about this, but we have some uniform thoughts about things like using yubikeys and such 17:20:08 i can take ownership on it, but figuring out what we think is useful to get as a result is key i think 17:20:22 ok 17:20:37 * asn has no strong opinions on this topic 17:20:54 i will ask some questions on the ticket and poke @tpo/core for some feedback 17:21:03 unblocking this now and moving to next item 17:21:23 no new items from other teams 17:21:36 ok, discussions! woh 17:21:58 [1 Mar] [nickm] Which of our CI builders are unreliable? I keep seeing stem failures. Are they real? Should we do anything? 17:23:01 i had a bit of a hard time parsing this one, but i think the important parts is it does seem like our stem tests are flaky. i don't think we have multiple builders enabled right now, so i don't think it's the builders themselves here 17:23:32 do we have a ticket with the errors that are usually happening? it doesn't happen very often to me when i retry builds with different images 17:23:50 I don't think so. Next time we see this issue, let's open a ticket? 17:24:58 yes, please, and assign it to me 17:25:04 sounds sensible 17:25:14 ack 17:25:23 i think while CI is still a bit flaky with the disk issues and such, just open an issue with a link to the failed build and assign it to me if you see something 17:25:53 ok, the next item on the discussion list is the one where i think nickm wants to hear dgoulet too, so i will skip that as we talked about it above 17:26:00 [1 Mar] [nickm] Proposal: Triage-Bot should not assign reviewers to Merge Requests that are labeled with "Needs Revision" or "Merge Ready". 17:26:22 i am fine with that, people just need to remember to remove those labels 17:26:25 I can talk rationale there, if you want. 17:26:47 But the issue is that my review queue is full of tickets that don't actually need review right now :) 17:26:50 i think i see it - it is to reduce assignee list? 17:26:51 yeah 17:26:55 that makes entirely see 17:27:08 please create a ticket to ahf/triage-ops and i will do it! while we are at this 17:27:11 ok 17:27:13 we talked about thursday to have a list of labels to ignore 17:27:22 you had some thoughts on this, maybe add that to the ticket too? 17:27:31 we talked about this for issues, nor MR's 17:27:49 yeah, that was for triage, not MRs. 17:27:49 this was for milestone handling 17:27:52 yep 17:27:58 let's talk through that one after the meeting? 17:28:03 wfm 17:28:04 ok! 17:28:11 [1 Mar [nickm] Any prep we should do for this week's arti hacking? 17:28:33 * asn is trying to find a useful one-day project 17:28:36 i think we should run it like the last times, but we did talk about recording the intro so we have it for future use too 17:28:54 * ahf wants to return to the bridge project 17:29:23 I'm feeling like refactoring something, but I'm not sure what. Possibly making something more testable. 17:29:48 Also there is now a (not stable) API in tor-client: it would be cool to add stuff to it, if you have ideas. 17:30:43 sweet 17:31:08 i should hear kushal where he is on the compression work with lifting that into a crate on its own, i haven't followed up on that since last time 17:32:26 does arti have a syscall sandbox yet? that might be fun to look into. or tokio. or testing. or... :-D 17:32:30 anything i have to do? i was gonna send out a reminder email to tor-internal@ tomorrow 17:32:45 ahf: i dont think so 17:32:46 jnewsome: I don't want to do a syscall sandbox at this phase, but it could be neat 17:32:58 later 17:33:07 jnewsome: testing would be awesome but in many cases it will need refactoring 17:33:21 yeah, that is often the bulk of the work with testing 17:33:24 jnewsome: I want to get tokio done soon; if it isn't done by thursday, that's a great idea 17:34:33 what is the tokio goal? to switch to it? or to have it as a secondary async runner? i don't know if it's a bit like if tor had support for both libevent and libuv ? 17:35:04 I want to support both async_std and tokio for now. 17:35:23 I don't know which should be primary; let's figure it out again after we've got them both working 17:35:36 interesting 17:35:54 Fortunately, tokio and async_std both expose some version of the futures API, so in theory there is not a _lot_ of stuff that we'd need to isolate 17:36:05 ok 17:36:11 and I _think_ that the async-std-specific parts of the code are currently isolated in tor_rtcompat 17:36:17 but htey both come with their own TcpStream and so on, right? 17:36:28 yeah 17:36:30 like libuv have tcp/ip connection functions but libevent just uses fd's 17:36:31 ok 17:36:31 IIRC 17:36:39 very interesting 17:36:47 you are gonna hack at this before thursday? 17:36:48 the APIs are much more similar than libevent vs libuv, for the parts that arti needs 17:36:51 maybe, depends on time 17:37:07 ack! very cool. 17:38:01 ok folks, i think we are done with our loooong monday meeting 17:38:13 we have no s61 sync today, as that was earlier over bbb 17:38:21 see you all on the other irc channels o/ 17:38:25 peace all! 17:38:27 cheers o/ 17:38:29 #endmeeting