mozilla :: #conduit

14 Sep 2017
00:37mcoteexcuse the naive question, but what exactly is the difference between the bmo-extensions and the demo repos?
00:37mcoteI think demo is meant to hold lando as well, but it doesn't at the moment, does it?
01:02mcoteanyway, finished a first crack at some general contribution docs:
01:02mcoteI'll put some specifics about lando-ui in the README in the repo
01:57dklmcote: the bmoextensions should just be the phab and bmo related containers andnothing else
01:59mcoteIsn't that all the demo repo is now?
02:54dklno the demo repo has all of the lando stuff too afaik
12:15janxsmacleod: Hello :) we're successfully publishing static analysis reviews on MozReview! (Well, manually at least, our bot on staging is misbehaving for some reason)
12:17janxsmacleod: However, as you can see in, the bot opens issues about lines that were not actually touched by the patch (see last 5 issues). Do you think this will be a problem? We wouldn't want to block patches because of code issues that are unrelated with the patch, so maybe we should silence these
12:17janxunrelated issues?
13:22smacleodjanx: IIRC the code I sent you should have a way to drop comments on unmodified lines or something. If it's not, I can find where that is because I know it exists
13:23janxsmacleod: and any issues opened by a reviewer need to be resolved before the patch can land, right?
13:23smacleodThe other option is to still comment on those lines, but not open an issue, only have an issue with the comment on modified ones
13:23smacleodjanx: ya
13:23janxaha, great idea!
15:32* janx apologizes for dropping out, I have another meeting
15:50globmcote: i have an extra day in my weekend, and i do like a good irc bot. i can slap something together until the real solution is in place
16:02smacleoddavidwalsh: still haven't gotten a new laptop?
16:04marti_hello imadueme
16:04marti_could you check my PR
16:05davidwalshsmacleod: My last *2* macs have had this problem; I emailed our tech peeps to ask if they had updated Macs, haven't heard back yet
16:05davidwalshBeen head down working
16:05davidwalshNow's not a good time to switch machines I guess
16:07imaduemeHi marti_: Great! The changes in that PR are perfect. The only problem is that, as per mcote's suggestion, we would prefer to do all code reviews in Phabricator - our new review tool which we are building. Have you ever used Phabricator before? If not I can give you a quick start on how to make a PR in Phabricator.
16:08marti_no could you explain me please
16:09imaduememarti_: Do you have a Bugzilla account? If not you can make one at
16:09marti_yep I have
16:17imaduememarti_: Ok, awesome. You can start following the instructions here to setup a Phabricator account You can see our Phabricator instance here:
16:18imaduemePhabricator is a little bit more complicated, especially if you are used to GitHub, but with time it starts to make sense.
16:19imaduememarti_: Please let me know if you get stuck because we all have felt that when we first started using it. The arc commands can be hard to understand especially at first. Thanks for the contribution!
16:23marti_ok :)
16:33davidwalshdkl: Reviewing your PR/Revision(s) now
16:34davidwalshSorry I didn't get them in for today arcanist updated to libphutil updated to phabext updated to phabricator updated to
17:21bolg^ lies; that's just it setting its initial state
17:29mcoteThat was some fast bot creation
18:00imaduemePassing on the meetings today to finish my work, just ping me if something comes up. Thanks
18:30smacleodckolos: for what I'm about to talk about:
18:30smacleod(/ bobm ^ )
18:35mcotechartjes: ops meeting is running a tad long
18:35mcotefeel free to hang out :)
18:36chartjessure. I have no qa updates -- in the requirements gathering stage for what I need to help y'all
18:36mcotechartjes: any questions for us?
18:38bobmSo, I think was mistaken in the granularity of the documentation for data classification.
18:38bobmGeneral data classification is outlined here:
18:38bobmIt doesn't touch on what should and shouldn't be in logs on that level.
18:39bobmMore general data here:
18:39bobmBut, yeah, not anything at that level I was discussing in the meeting :
18:43chartjesmcote: none for you folks yet, still tracking down stuff on using GitHub login in tests etc. No worries
19:11davidwalshdkl: Per , returning `null` from `getMaximumRetryCount` retries daemon tasks indefinitely
19:14davidwalshmcote: So doorkeeper's default getWaitBeforeRetry is all math-y:
19:14davidwalshmcote: Not sure I feel comfortable with that
19:14davidwalshLike..say we get a transient error, then more events happen to said revision, and those go through
19:15smacleoddavidwalsh: do you know what the units of the number returned are? seconds? smaller?
19:15davidwalshThen an hour later overlapping transactions get pushed through
19:15smacleodthat's a loooong retry window :/
19:15davidwalsh60 is the default for other worker items
19:15davidwalshfor some reason the default for doorkeeper is more complex
19:16smacleoddavidwalsh: that's (5 minuites) * (8 to the power of count)
19:17smacleoddavidwalsh: after 2 failures that's 320 minutes
19:17davidwalshMy dashboard here says some of the tasks are re-queued for 4 hours
19:17davidwalshThat's no good
19:18smacleodnot at all...
19:18smacleoddavidwalsh: wait, you can redefine that method for our task though right?
19:18davidwalshsmacleod: Right; I can return 60 for every 60 seconds
19:18davidwalshWe don't need to do anything complex
19:18smacleodah okay, we should probably do exponential, but have a maximum
19:20smacleodlike `min(pow(2, $count), 5 * 60)` or something
19:21smacleode.g.: 2, 4, 8, 16, 32, 64, 128, 256, 300, 300, 300 for above ^
19:24davidwalshMy god
19:24davidwalsh2 failures and the wait becomes monsterous
19:25smacleodya, pow(8, $count) * 5 minutes is no joke
15 Sep 2017
No messages
Last message: 7 days and 13 hours ago