22:59:09 <ahf> #startmeeting network team meeting, 19 february 2020
22:59:09 <MeetBot> Meeting started Wed Feb 19 22:59:09 2020 UTC.  The chair is ahf. Information about MeetBot at http://wiki.debian.org/MeetBot.
22:59:09 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
22:59:15 <ahf> yoyoyo
22:59:21 <nickm> hihi
22:59:27 <gaba> o/
22:59:27 <catalyst> hi
22:59:28 <ahf> our pad is at https://pad.riseup.net/p/tor-netteam-2020.1-keep
22:59:34 <ahf> o/ nickm, gaba, catalyst
23:00:31 <ahf> mikeperry: argh, sorry for removing your section earlier
23:00:48 <ahf> let's start with:
23:00:51 <ahf> Let's check and update our roadmap: What's done, and what's coming up? Use: https://dip.torproject.org/torproject/core/tor/boards
23:01:09 <ahf> i have the feeling most people are in 0.4.3 land, so this will probably not be the biggest focus
23:01:15 <mikeperry> ahf: no worries. I am in and out; still trying to figure out when to task switch between teams
23:01:24 <ahf> mikeperry: :-)
23:01:24 * nickm moves #26768 to review
23:01:50 <ahf> jnewsome: cool with all the shadow items!
23:01:53 * nickm moves #31851 to doing, since I made some progress on that
23:02:19 <ahf> sweet
23:02:29 <jnewsome> ahf: :)
23:02:40 <nickm> the description for #5304 in "review" looks wrong?
23:03:15 <ahf> yeah
23:03:19 <ahf> it needs some love first
23:03:37 <nickm> some of the things in review aren't currently needs_review
23:04:08 * teor is here
23:04:12 <nickm> hi teor !
23:04:13 <teor> Sorry, I lost track of time
23:04:22 <ahf> o/ teor
23:04:43 <ahf> okay, if everybody can go over their items in the review queue there and remove those that shouldn't be there, that would make things more readable
23:04:55 * ahf will do his when he is on a computer with his pw manager for GL on..
23:05:07 <ahf> we have so many reviews this week :o
23:05:21 <ahf> #33374 is the only one with nobody assigned as reviewer
23:05:32 <teor> I'm sorry, I'm confused by the use of trac and gitlab at the same time
23:06:12 <ahf> teor: no worries, i think we all mess that up every now and then
23:06:41 <ahf> hiro is working on the new GL instance to tor this week, which timewarps us one step forward with all of that
23:06:59 <nickm> mikeperry: if you're able to access dip, there are some things you worked on in the "review" column that I think are actually under needs_revision?
23:07:37 <mikeperry> oh is dip now canonical for needs_review? I was looking at trac queries for that
23:07:46 <nickm> trac is canonical
23:07:54 <nickm> but we try to keep the columns in that dip query up to date
23:08:01 <nickm> nobody much likes it, but it helps ahf and gaba
23:08:20 <mikeperry> https://trac.torproject.org/projects/tor/query?reviewer=~mikeperry&status=needs_review says no tickets
23:08:44 <nickm> Are you able to see the dip query on the pad? these aren't ones you're listed as reviewer on
23:09:00 <nickm> these are ones that you worked on the code on where dip misclassifies them right now
23:09:36 <mikeperry> you mean the full boards view?
23:09:40 <nickm> right
23:09:51 <ahf> are folks OK with their reviews this week?
23:10:25 <nickm> yup
23:10:51 * ahf grabs the last one without a reviewer
23:10:56 <mikeperry> the ones I see there seem to not actually have anything for review.. is there a way we can bookmark query urls for stuff like this into dip?
23:11:24 <ahf> we bookmark it by having a link in the pad
23:11:33 <ahf> and the pad url doesn't change except for once a year
23:11:44 <teor> I have fixed two tasks which had the wrong status in dip
23:11:48 <mikeperry> the boards view does not tell me my responsibilities at a glance
23:12:23 <gaba> mikeperry: you need to filter it by assigned
23:12:23 <mikeperry> I am not sure what to do other than click on every single ticket?
23:12:29 <teor> We should work out how to do something like "my reviews" in gitlab: https://trac.torproject.org/projects/tor/wiki/user/teor#Reviews
23:12:36 <gaba> yes
23:13:29 <teor> ahf: I'm happy with my reviews, I've done the urgent ones that nickm asked me to do
23:13:34 <ahf> teor: sweet
23:13:49 <teor> Others may wait until Friday or Monday, because I'm buried deep in reachability testing right now
23:13:50 <mikeperry> when I type @mikeperry in the box, the board becomes blank :/
23:14:00 <mikeperry> when I do "Reviewer: @mikeperry" it is also blank
23:14:30 <gaba> it seems you do not have anything other than 29494 and 30992 in dip...
23:14:44 <ahf> it's assignee: @mikeperry, but i don't see anything for you
23:15:04 <teor> Your trac is also empty: https://trac.torproject.org/projects/tor/query?status=needs_review&reviewer=~mikeperry&col=id&col=summary&col=status&col=type&col=priority&col=milestone&col=component&order=priority
23:15:15 <mikeperry> ok. that's what I thought
23:15:31 <ahf> okay, let's do 0.4.3 status: https://trac.torproject.org/projects/tor/wiki/org/teams/NetworkTeam/CoreTorReleases/043Status
23:15:51 <ahf> hm, #33069 is in review but has no owner
23:16:03 <ahf> ah, it's a volunteer
23:16:03 <ahf> cool
23:16:05 <nickm> ahf: no problem; it's from a volunteer
23:16:18 <nickm> the big question for 0.4.3 here is: if we called maint-0.4.3 a release candidate tomorrow, what would we regret not fixing?
23:17:00 <ahf> none of mine except the TROVE would be a catastrophe, but it is annoying bugs that happens for mobile folks
23:17:16 <ahf> but a lot of them are in the nice to have department
23:17:27 <ahf> are you considering that? :-)
23:18:13 <nickm> Well, sure!  Gnerally i would like us to be realistic about 0.4.3.x tickets, and defer stuff that we aren't going to do in the next week or three
23:18:29 * ahf nods
23:18:29 <catalyst> i think the only one of mine is an 043-should and i already determined it's probably an old bug (that we should fix anyway but at least it's not a regression)
23:18:52 <nickm> #32622 ?
23:18:56 <catalyst> yeah
23:19:19 <nickm> How much time do you think is remaining on that, and do you think it makes sense to move it into 044 if it's not a regression?
23:20:07 <catalyst> still needs diagnosis, and might need a more invasive change than we'd be willing to take on 043 (not sure though)
23:20:33 <nickm> ok. I trust your decision on that, but feel free to ask if you're not sure
23:20:46 <nickm> catalyst: question for you on one of my 0.4.3 tickets...
23:21:15 <ahf> i think my TROVE and #32165 are my needs-to-be-fixed tickets for 0.4.3
23:21:17 <nickm> catalyst: In #31078 you requested some documentation and clarity improvements. I outlined some ones that I think would improve the code clarity on that ticket, and teor acked them.
23:21:51 <nickm> catalyst: Do you think I should move ahead with the renaming/documentation proposed there, or wait for you to give feedback?
23:22:17 <nickm> https://trac.torproject.org/projects/tor/ticket/31078#comment:7 is the comment where I suggest the changes; teor and I discuss them below on the ticket
23:22:44 <catalyst> nickm: it looks like teor had further comments that you hadn't responded to?
23:23:10 <nickm> I agree with teor's suggestions
23:23:38 <nickm> and I'm happy to rename things again later if we come up with even better names :)
23:24:25 <catalyst> sorry, this is too big for me to comfortably answer right now
23:24:43 <nickm> okay. So, I should wait on this?
23:25:09 <nickm> I'm not in a hurry to do this, so I can wait if need be
23:25:10 <catalyst> nickm: teor probably shares overlapping concerns with me here so their suggestions are probably something i consider an improvement
23:26:03 <nickm> ok. I can also go ahead and go with the plans, realizing that you may have additional comments later after you've thought about this for a while
23:26:36 <nickm> I'm not asking for a permanent waiver here; just wondering if you think it's worthwhile me making what I (and teor I think) believe is an incremental improvement
23:27:55 <nickm> but like I said, I'll wait a bit longer if you'd rather I did.
23:28:15 <catalyst> nickm: yeah, go ahead with teor's suggestions. next time i have a reason to touch that code, i can try to take notes about what parts of the naming make it hard to keep in my head
23:28:25 <nickm> okay
23:28:29 <nickm> thanks, catalyst!
23:28:38 <ahf> cool!
23:28:50 <ahf> are we OK with 0.4.3 things for now, nickm?
23:29:13 <nickm> I think so
23:29:25 <ahf> cool!
23:29:28 <ahf> we have a discussion item
23:29:32 <ahf> Want to help make Tor's CI faster?
23:29:32 <ahf> I'm mainly focused on macOS Chutney, but making other jobs faster would also help:
23:29:35 <ahf> https://lists.torproject.org/pipermail/tor-dev/2020-February/014162.html
23:29:35 <ahf> from teor!
23:30:00 <teor> Oh I think that's maybe last week?
23:30:28 <ahf> oh!
23:30:34 <ahf> maybe!
23:30:36 <teor> I'm not working on that right now, but I'm happy to review others changes
23:30:46 <ahf> sorry then, i thought i caught everything when i went over the pad diff
23:30:58 <teor> Also the chutney improvements I'm doing right now should speed things up. Fewer timeouts and retries.
23:31:15 <ahf> is
23:31:16 <ahf> asn: We seem to have accumulated a nice collection of circpad-related tickets/bugs: #33354, #33352, #32140, #30992 . But both me and Mike seem knees deep in other sponsored stuff. We need to create space and time to handle those issues.
23:31:22 <ahf> also from last week then? i don't remember this item :o
23:31:26 <teor> We haven't had a meeting for about 10 days :-)
23:31:36 * ahf enables the time warp mode
23:31:45 <nickm> #4631 is quite clever :)
23:32:06 <ahf> no
23:32:08 <ahf> this one is new!
23:32:26 <ahf> okay, mikeperry maybe you are the best one in here to talk about this. it's very late in greece right now, so asn isn't around :-/
23:33:24 <mikeperry> yes; basically I messed up the shutdown path of circpad while trying to optimize machine turnover.
23:33:42 <mikeperry> a complete fix is rather involved. I can do it, but the question remains about what to do in the meantime
23:34:03 <mikeperry> demoting the loglines is safe, or at least it was for the cases last time I looked
23:34:14 <mikeperry> it is something that should be properly fixed. More bugs may be hinding in there
23:34:46 <nickm> who is seeing this warning -- circpad experimentors, or regular users/operators?
23:35:18 <mikeperry> it might be the latter, esp if it is showing up in chutney. I think assuming it is the latter and treating it accordingly is a good plan
23:35:24 <teor> tor logs the warning in various chutney networks, and I think people are seeing it on the live network
23:35:48 <mikeperry> I consider this pretty high priority at this point, for me, esp since regular chutney is triggering the warns
23:35:50 <nickm> Hm.  In that case, maybe you and asn can figure out if there is a fix in between "the complete fix" and "demoting the log messages"?  Rather than demotion I'd even prefer rate-limiting
23:35:58 <nickm> ok
23:36:05 <nickm> should it be 043-should?
23:36:06 <teor> dgoulet sees it on a client on master: https://trac.torproject.org/projects/tor/ticket/33354#comment:2
23:36:13 <ahf> mikeperry: how is it in relationship to 0.4.3?
23:37:00 <mikeperry> it might require a protocol bump to fix properly.. if it does, we might want to still only demote the messages for 043
23:37:45 <mikeperry> but yeah, this is why I put myself back on the pad for this meeting
23:38:30 <jnewsome> re #4631 would be nice if we could use shadow for test cases of these tricky-to-repro issues
23:39:19 <nickm> jnewsome: I don't know if shadow can actually handle this one -- does shadow handle the case where a relay is overloaded in CPU, and so acts more slowly than it should?
23:39:25 <nickm> it would indeed be cool though
23:39:55 <jnewsome> I think you can configure networking delays; would that be sufficient?
23:40:13 <nickm> jnewsome: not exactly -- clock skew might help though if you can make a relay's clock incorrect
23:40:20 <nickm> mikeperry: let's call this ticket 043-should, then, where the 043 part is "figure out what the real fix is, and whether it applies to 0.4.3"?
23:40:32 <nickm> (and at least do something to quiet the situation in 0.4.3)
23:40:32 <mikeperry> ok
23:40:43 <nickm> mikeperry: is #33354 the ticket to do this with?
23:40:56 <jnewsome> Seems like something it could plausibly be made to handle, if it doesn't already
23:41:27 <teor> jnewsome: Reproducible errors would be helpful. Chutney isn't really designed for that.
23:41:58 <teor> Networking delays can make large votes upload late, so you shouldn't need clock skews. But testing clock skews is also important.
23:42:01 <mikeperry> nickm: I suspect all these stem from cases I enumerated in #30992, and is triggered by machines spinning up and shutting down repeatedly on the circ. for #33354 and the others
23:42:19 <mikeperry> so I'll probably use that one and dup the rest to it if so
23:42:32 <ahf> what is the actionable item in this? mikeperry and asn talks together and makes a plan for this? asn is right now in s27 land for a bit longer
23:43:01 <mikeperry> I'll dig into it. it's my mistake due to my optimization
23:43:34 <nickm> ok, i've set the milestone and added 043-should on #30992 then
23:43:53 <ahf> cool, yeah, sounds like something that we should have in mind with 0.4.3
23:44:23 <ahf> i don't see anything in bold in the pad
23:44:28 <ahf> does anybody have anything else we need to chat about?
23:45:10 <nickm> mikeperry: please check the comment on #30992 and make sure I got it right :)
23:45:17 <teor> I'm modifying chutney a lot right now, please log failures if you see any
23:45:37 <ahf> teor: cool! :-)
23:45:42 <mikeperry> nickm: looks good!
23:45:44 <teor> I'll put that in the pad
23:45:57 <ahf> now that i think of it the way i use chutney is that i pull the repo and then never update it
23:46:08 <ahf> i guess that isn't the smartest thing to do
23:47:06 <teor> Tor will make you update soon, because tor uses a new chutney network
23:47:47 <ahf> ah
23:47:50 <ahf> that makes it easier
23:47:59 <ahf> cool folks. i'm gonna close the meeting down now then
23:48:07 <teor> I don't deliberately do breaking changes ;-)
23:48:07 <ahf> thank you all for showing up and happy hacking for the rest of the week!
23:48:11 <ahf> #endmeeting