23:00:32 #startmeeting weekly network team meeting, 28 may 2019 23:00:32 Meeting started Tue May 28 23:00:32 2019 UTC. The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot. 23:00:32 Useful Commands: #action #agreed #help #info #idea #link #topic. 23:00:44 Pad, per usual, is at https://pad.riseup.net/p/tor-netteam-2019.1-keep 23:00:45 o/ 23:00:58 hi ahf, gaba, teor4 ! Who else is here? 23:01:12 o/ 23:01:17 hi 23:01:25 hi catalyst, mikeperry ! 23:01:29 hi! 23:01:40 hello world 23:01:42 I hope asn is taking it easy; it's really late for him 23:01:51 (really late for you too, ahf) 23:02:12 is 1, but i'm in the sofa with family guy on the tv 23:02:15 so is good 23:02:34 I just sent a /msg to dgoulet 23:02:45 david wrote he'd be afk in the notes 23:02:50 ah! 23:02:51 thanks 23:03:07 then that's everybody 23:03:51 so, current priorities are S19 wrap-up and 041-must/should, still. 23:03:54 (Right, gaba?) 23:05:03 I'd like to get to other sponsored stuff again, but we need to get 041 more stable before we can move on to 042 work 23:05:25 yes 23:05:44 anything PT related or snowflake related go straight to the next sponsor after s19 23:05:45 so, first thing to look at is CI issues 23:05:56 #29437 and #20235 seem like the bigger ones 23:06:02 um no 23:06:07 #29437 and #30235 seem like the bigger ones 23:06:15 anything else we're hitting? 23:06:49 There's the issue where apparently our freebsd builders don't exist 23:06:52 what was the decision on the test behind #25791 (and other dupped tickets) 23:07:04 and should we close that ticket? 23:07:13 * catalyst hit another instance of #28229 23:07:41 mikeperry: we looked a bit into that evening and the CI for the BSD's are not working at all 23:07:51 there are no builders for it. i logged onto jenkins and tried to restart it 23:08:20 i think i'd close the ticket (and maybe we should get rid of the BSD builders in the jenkins list since they are defunct?) 23:09:15 yes, let's remove those builders 23:09:16 I think that makes sense 23:09:35 unless fixing the builders is trivial. but IIRC we don't run them, so probably not. 23:09:55 it's the pixelminers ones 23:10:10 if fixing the builders is trivial, we can restore them once they are fixed 23:10:14 these: https://buildbot.pixelminers.net/#/ 23:10:55 and none of their workers seems to be working: https://buildbot.pixelminers.net/#/workers 23:11:33 ok. let's close that one as "wontfix" with a note about lack of CI, and open a ticket to remove the bsd builders from jenkins 23:12:15 ack 23:12:19 ty 23:12:30 for the other CI failures -- are we on track to get them fixed? 23:13:36 I'm on CI rotation this week, so I should be helping out here 23:14:12 i think test_rebind.py has been failing enough that we should disable it or figure out why it's failing intermittently 23:15:36 ok, let's mark that 041-should 23:16:01 do we actually know what is happening there? 23:16:14 teor4: also, have we made any progress on the stem failure? 23:16:23 I saw the ticket to add logs 23:18:12 next item is https://trac.torproject.org/projects/tor/wiki/org/teams/NetworkTeam/CoreTorReleases/041Status -- the 041 release 23:19:07 do we think we can get all the assigned stuff done ... this week-ish? 23:19:13 we also need to deal with assigning the new items 23:19:22 nickm: yes, we have made progress on CI. atagar made some stem changes yesterday, and I have a list of changes to make today. 23:19:23 when is the next point relase planned for? 23:19:36 I think there are also some CI tickets pending review. 23:20:04 great 23:20:16 mikeperry: I try to do them about 2-3 weeks apart when we're in alpha 23:20:31 but we should try to fix all the regressions we found first if we possibly can 23:21:49 re CI: it seems like this is the first time we have tried to debug stem in this way, so it has been slower than expected 23:22:20 yeah, it will take some work, I think 23:22:34 actually, there was the thing that I fixed with TLS 1.3 issues in python? 23:23:05 but yeah, for that one, I found a way to repro locally 23:23:30 I am not sure if I will get to mine (#30649) this week, but it is also a simple fix. should take only 30 minutes or so and be easy to review 23:24:09 If you can find those 30 minutes, it would be really appreciated 23:24:36 But I agree the other stuff you're listed on for this week is important too and can't really be deferred 23:24:47 that's when I asked the deadline. it would be a lot easier to find those 30 minutes if it is just as good if its done early next week 23:24:56 sure 23:25:12 thanks for asking! 23:25:51 one last thing on https://trac.torproject.org/projects/tor/wiki/org/teams/NetworkTeam/CoreTorReleases/041Status before we move on -- for the things in the "owner:(none)" category at the head of the page that are not needs_review... 23:26:26 ... is there a logical person to handly any of them in particular? 23:27:09 I've tried to make sure that all the regressions are assigned 23:27:25 *handle 23:27:45 I'm also using the "041-regression" tag to track 041 regressions in particular, so that we can do a postmortem and figure out how we broke things :) 23:29:00 ok. I'll try to draw up a tentative mapping of these tickets to people 23:29:07 there are some tickets there that has people working on, like #28793 with teor4 23:29:45 teor4: should you be owner on that? please accept if so. Any time between now and July 31 should work on that one 23:30:09 ah, thanks! you did 23:30:24 I'd like to get it done soon, because the fallback down warnings are annoying 23:30:54 also I'm planning to remove all the items at the end of the page this week (the ones that are not 041-MUST or 041-SHOULD) so maybe everybody have another look at those before then :) 23:30:59 teor4: ack 23:31:47 ok. I don't think we had any roadmap progress last week, but please let's check the roadmap in case :) 23:31:59 we might have stuff to remove from the "in progress" column 23:33:02 On reviews: is anybody stuck on anything, especially for 0.4.1? 23:34:50 Are we happy with reviews & kanban status? If so let's move on to discussion time? 23:35:01 *nod* 23:35:02 ok 23:36:06 ok. first is the rotation updates; they look pretty clean 23:36:31 second is the DisableAllSwap question. teor4, is that moot as I think per our discussion from a couple of days ago? 23:36:54 Yes, I thought I deleted that part of the pad. oops 23:37:04 no worries 23:37:27 next thing is arma's question about getting us to use valgrind more 23:37:48 30629 is something our sanitizers really should have caught; it might be worthwhile looking into why they didn't 23:38:03 #30629 23:38:16 yep. mostly i realized that i periodically find good stuff with valgrind, so getting other people to consider that, and figure out if there's a way to put a method around it, could be smart 23:38:34 teor4: Could #29528 be the issue here? 23:39:38 Possibly, but asan should catch double-frees as long as we free and NULL the pointers 23:39:55 So are we NULLing all the pointers to the freed memory? 23:40:05 part of this is happening inside libevent, which is probably not NULLing the pointers 23:40:34 and may not be instrumented with asan 23:41:09 ok. I'll open a subticket to figure out what CI we want here 23:42:02 i just added a question 23:42:05 Our control refactor caused at least two regressions: #30580 #30646. How can we make sure there aren't any others? 23:42:38 good question. We should look at our coverage on control_cmd.c. 23:43:02 is this something that we can assign to sponsor 31 and do it there? 23:43:05 maybe we can use stem to write more tests for the un-tested commands 23:43:40 If you provide me a list of test scenarios I'd be happy to help. 23:43:57 (or happy to merge patches if others want to get their hands dirty) 23:44:00 atagar: I could try to write a list of controller commands without coverage? 23:44:05 would that help? 23:45:29 starting with those two commands :-) 23:45:34 Yup. Bonus points if it includes how you'd care for them to get exercised (ie. what kind of input or assertions you'd find interesting). 23:46:39 ok. just made #30676 23:48:28 catalyst -- you might also have thoughts on this point 23:49:26 other than that, I think we are out of questions for the pad. Anything else for us to talk about today/tonight? 23:49:55 * ahf is good 23:50:01 I've assigned myself a pretty big pile of 041-should/must stuff, since I think I know how to fix it, but if anybody would like to take anything from me when they run out of stuff to do, please feel free :) 23:50:15 nickm: we don't currently do test-stem coverage in Travis do we? 23:50:18 no 23:50:41 would you like me to generate a gcov run with test-stem? 23:51:33 maybe? looking at #30676 23:51:42 ok, will do 23:52:08 some of them might be doable with unit tests; who knows? 23:52:33 true 23:54:48 catalyst: I'll upload a coverage run somewhere once I have it; please let me know if you want to do a pass on #30676 at that point 23:54:57 nickm: ok 23:55:18 both of the regression bugs that we found so far would have been stopped by any test that tried _any_ successful case of those commands, IIUC 23:55:42 anything else for this week? 23:56:06 also please feel free to ping me or gaba any time you want about anything you want. If I'm not around, we can set up a time where I will be 23:56:25 thanks o/ 23:56:54 hearing nothing else ... peace, all! I'll linger on IRC for another 10 min or so before I have to put the kid to bed :) 23:57:03 thanks! 23:57:05 have a good night and/or day, everybody! 23:57:06 o/ 23:57:07 #endmeeting