[00:33] ok axw ready when you are === akhavr1 is now known as akhavr [01:01] * redir eods. may be back later. [01:07] hi folks - anyone available to help troubleshoot what looks like a bug in user removal? https://pastebin.canonical.com/169528/ [01:11] blahdeblah: not sure about troubleshoot - looks like a straight bug. when users are removed, they are actually marked as inactive. looks there's there's some logic issues there [01:12] axw: remind me - for bootstrap, the majority of machine config is done via ssh; cloud init is used to get the ssh keys onto the machine I think. is that also the case for workload machine provisioning ie we do the bulk of setup via ssh and not cloud init? [01:17] wallyworld: we don't use ssh at all for non-bootstrap machine provisioning [01:17] axw: ta, i thought that was the case but wasn't certain [01:17] wallyworld: (apart from one or two providers which may use it internally) [03:13] axw, menn0: do you want me to come into the tech board meeting to talk about the dependencies I emailed about, or are you good? [03:16] natefinch: it might be useful if you could talk about them when we get to that bit, but we can also manage without you if you're finishing up for the day [03:16] menn0: I have some time [03:16] natefinch: cool. i'll ping you when we get to it (or join whenever you like) [03:16] menn0: cool, thanks [04:00] natefinch: now? [04:00] menn0: omw [04:26] Bug #1254402 changed: error bootstrapping on ec2 duplicate security group

[04:26] Bug #1456725 changed: TestActionEvents fails

[04:26] Bug #1502139 changed: Panic in TestUpgradeJuju

[04:36] anastasiamac: yt? [05:58] redir: i m. why r u? (was doing a school pick-up)... [06:31] anastasiamac: do you recall - i think --metadata-source for bootstrap was broken around beta15 timeframe and it needed to be fixed before 2.0GA. Right? [06:32] wallyworld: sounds about right.. [06:32] wallyworld: why r u in beta15? [06:32] i'm not === frankban|afk is now known as frankban [09:44] voidspace: ping [10:10] dimitern: pong [10:10] voidspace: so I've been testing your gomaasapi change: https://github.com/juju/gomaasapi/pull/62 [10:11] voidspace: it looks like it's doing the right thing (compared to the maas cli at least) with one or more negative spaces constraints [10:11] dimitern: great [10:12] voidspace: however, I can see that 'unorderable types Subnet() < Subnet()' in some cases [10:12] voidspace: is there a bug for that in maas? [10:12] dimitern: I've never seen that [10:12] dimitern: is that an error from MAAS or in the gomaasapi code? [10:14] voidspace: from maas - e.g. when trying `jujy bootstrap --no-gui --debug vmaas-21 --config=~/.j/conf-x-fast.yaml --constraints 'arch=amd64 spaces=^compute,^external'` - the config just sets trace logging, default-series: xenial, disables apt update/upgrade [10:14] dimitern: ah, so definitely sounds like a MAAS bug [10:15] voidspace: there's 1 machine that matches those spaces, and here's the error we get: http://paste.ubuntu.com/23415499/ [10:16] voidspace: yeah, IIRC mpontillo did mention something about that and they seem to be fixing it [10:16] dimitern: cool, so I don't think we should hold up this PR with it - I haven't seen a MAAS bug for it [10:16] dimitern: worth filing even if it turns out to be a duplicate [10:19] voidspace: I'll look into filing one later today; just testing with 2.0 now - 2.1 tested OK [10:19] dimitern: I have a very short PR fixing not_networks for 1.9 I'd like you to test too if possible [10:19] dimitern: will ping in a couple of minutes [10:20] voidspace: sure - can do - is it about the way we serialize multiple negatives? [10:22] dimitern: yep [10:23] dimitern: changing from comma separated to multiple parameters [10:32] voidspace: ok [10:32] mgz: ping [10:33] mgz: do you know which mongo version are we using for the windows CI slave running this job: http://reports.vapour.ws/releases/4546/job/run-unit-tests-win2012-amd64/attempt/3002#highlight [10:33] mgz: if it's 2.6.x, the test might be failing due to this mongo bug - https://jira.mongodb.org/browse/SERVER-14840 [10:52] voidspace: tested on 2.0 successfully and LGTM-ed [10:53] dimitern: you rock [10:55] voidspace: it took me a lot longer to setup the spaces / subnets on maas than to run bootstrap OK :) [10:56] dimitern: I can imagine [10:56] dimitern: my maas 2 is not playing ball and won't enlist new nodes [10:56] burned a few hours on it :-/ [10:56] dimitern: might have to reinstall from scratch [10:57] got inveigled in another conversation [10:57] about to create the PR for maas 1.9 [10:57] and I'll do PRs to bump the gomaasapi dependency in juju [10:57] both for 2.0 and develop branches [10:58] voidspace: I'm now using frobware's kvm-maas-add|remove-node with great success to add nodes - see https://github.com/frobware/kvm-maas/pull/1 [10:58] it takes <5m to create, enlist, and commission 6 KVM nodes with multiple NICs [10:58] dimitern: yeah, I've used that before - don't know why I didn't think of it. [10:58] dimitern: will try it now [10:59] dimitern: it uses the api so should work fine [10:59] dimitern: I'll just delete my existing nodes that MAAS won't recognise and try again [10:59] voidspace: it works quite well, but I'd suggest trying my modified version (re that PR), as it improves quite a few things :) [11:00] voidspace: like not hard-coding 4 NICs per node but as many as you've passed as args [11:00] dimitern: where's your version? [11:01] voidspace: this branch essentially: https://github.com/dimitern/kvm-maas/tree/multiple-networks [11:01] dimitern: cool, thanks [11:01] dimitern: I only have two spaces/subnets defined in maas - but that's easy to change and enough to test this anyway [11:02] voidspace: +1 [11:28] * dimitern steps out for ~1h [11:31] does anyone know if juju restore actually works? [11:55] dimitern: see bug 1637294 btw [11:55] Bug #1637294: Windows host needs a current mongodb (3.2.x) [11:59] mgz: cheers! [11:59] * dimitern is just getting out the door - bbiab hopefully in ~1h === icey-travel is now known as icey [12:05] is aws efs volumes something that is on the horizon or will it be a while before network shared storage will be available in juju? [12:05] it would be really great being able to deploy multiple units that share the same filesystem [12:26] dimitern: sorry for the delay I got distracted: https://github.com/juju/juju/pull/6528 [12:26] dimitern: bugger, I need to retarget that to develop not staging [12:27] dimitern: done [12:27] unfortunately means pre-check failed [12:27] all unit tests pass though [12:37] dimitern: and https://github.com/juju/juju/pull/6529 [12:37] dimitern: gah, conflicts [12:42] dimitern: rick_h___: so, when a PR forked off staging but targetted against develop has conflicts [12:42] dimitern: rick_h___: I assume the correct approach is to merge develop into the PR and fix the conflicts? [12:42] I have both "go fmt" and build failures (probably related to go version) with current develop [12:42] so I've fixed both those as part of the conflict resolution and they'll now show up in my PR [12:43] running all related tests [12:43] voidspace: the best thing is to cherry pick the conflict commit and resolve the conflicts [12:43] voidspace: imo [12:43] rick_h___: so fork off develop and just make the new commit [12:44] rick_h___: I think that has the same end result - and I still need to fix the two issues [12:44] rick_h___: thanks [12:44] voidspace: so checkout develop, update it from upstream, go to your branch, and cherry pick the conflict from develop [12:44] rick_h___: ah, I see [12:45] rick_h___: that would be better, yes [12:45] rick_h___: I'm still concerned that we have a "go fmt" error and a build error on develop! [12:45] rick_h___: (the build error is in a test) [12:45] rick_h___: and is likely to be go version related - but with my fix tests all pass [12:46] rick_h___: this is now my PR (very short) https://github.com/juju/juju/pull/6529/files [12:47] rick_h___: (self assignment to args.Host was the build error I saw) [12:53] mgz: thanks for the magic incatation on my branch - I've repeated it [12:54] voidspace: I'm a little surprised it didn't listen to mine [12:55] voidspace: also, lgtm [12:55] mgz: cool, thanks [12:55] mgz: I'll wait for the check and for dimitern to QA [13:08] * dimitern is back [13:10] a small names change to make it easier to work with empty cloud credentials: https://github.com/juju/names/pull/77 [13:10] reviews appreciated: ^ [13:11] voidspace: looking at both of your PRs, and QAing the 1.9 one [13:13] actually QAing both of them as it's trivial [13:14] rick_h___: is our 1:1 going to be on? [13:19] rick_h___: I have no burning questions, so happy to skip this week [13:24] Bug #1638560 opened: 1.25 broken on MacOS sierra

[13:26] rogpeppe1: Hi. I've been asked to do one of the things we talked about yesterday: test the latest juju against a released version of JIMM, to ensure juju doesn't break what JIMM uses. Do you have any tips on how to go about that? [13:35] voidspace: skip please [13:35] rick_h___: kk [13:47] voidspace: both PRs LGTM [13:48] dimitern: you rock, you really do [13:48] dimitern: about to propose same PRs against 2.0 [13:49] voidspace: ;) cheers [14:03] natefinch: dimitern: frobware: standup time [14:03] katco: skipping, at the cloud sprint. [14:04] frobware: ah ok... ta! === daniel is now known as Guest19937 === Guest19937 is now known as Odd_Bloke [16:00] dimitern: ping for call [16:08] dimitern: the 500 error happens in lieu of a "no machine matched constraints" error. positive space constraints should happen in the interfaces constraint; they will not work properly in the 'subnets' constraint. [16:08] voidspace: ^ [16:09] mpontillo: ok, interesting [16:17] * dimitern well, it's been good 4 years... take good care ;) [16:19] dammit [16:21] mgz, voidspace, dooferlad: any of you know dimiter's home email? [16:22] got it, thanks [16:30] .... [16:30] what just happened? [16:34] I do not know. [16:35] :-( [16:38] check your email [16:39] natefinch: I believe that their reactions are mail related [16:40] ahh, right === admcleod1 is now known as admcleod [17:01] okay, I think I finally have a solution I might like with this ssh stuff [17:02] and I can do it in two somewhat sane prs [17:02] mgz: did you see jam's reply there? [17:02] mgz: he sent an email on the ssh notes a couple of days ago I think? [17:02] rick_h___: yeah, that sent me off down a rabbit hole [17:02] ah ok nice [17:03] which, to be fair, gets to a much better final result, but involves deeper magic [17:58] changing locations all, will be back online is a few minutes [18:02] Bug # changed: 1541482, 1597601, 1626304, 1634289 [18:02] heh, that was a drop off [18:03] yeah [18:03] i think the event ends when the person ends it, not when the time-delayed stream ends [18:03] i've had that happen a few times now [18:07] easy review anyone? https://github.com/juju/jsonschema/pull/2 [18:08] +19 -19 [18:08] natefinch: I'm in reviewing mode while waiting for build so okay [18:09] natefinch: heh - er, are we going to use this for actual translations ever? === frankban is now known as frankban|afk [18:09] because if so singular/plural is wrong [18:10] I'm not sure what you mean. It's used for generating interactive UI [18:10] natefinch: https://www.gnu.org/savannah-checkouts/gnu/gettext/manual/html_node/Plural-forms.html [18:12] ahh, the old "multiple plural forms" [18:12] if we only ever want this to do english it's fine [18:12] you uh, let me know when we plan to translate Juju, and I'll get right on making this better :) [18:12] well this is great, I only have decaf and I live too far from the coffee shop... I really need a second car or a bike safe city [18:14] perrito666: I feel you, I ran out of coffee today too. luckily I have caffeine pills in the cupboard which let me function until I could get to a coffee shop [18:15] natefinch: I live way closer to "civilization" than you do, I just have one car only which my wife takes to work so even though I live closer its not all that close [18:15] mgz: anyway... yes, for now, I think it's sufficient. At some point, if we want to localize juju, we'll need to update this. But that's a much bigger job. This is just so you can take jsonschema and produce reasonable english-looking queries out of it [18:15] natefinch: code changes approved. [18:15] mgz: thanks :) [18:16] perrito666: I don't really believe that you can't use the internet to cause a man on a scooter to bring you coffee [18:17] mgz: never been to south america eh? [18:18] perrito666: I simply trust your ingenuity [18:19] well I guess I could kidnap one of mi neighbours and ask their family to bring me coffee but its a bit excessive [18:19] :D [18:20] lol [19:02] dooferlad: hey are you still around? [20:10] perrito666, nice job on the quick turnaround on the backup/restore test failure [20:13] thank me when CI passes, jenkins does not always agree with me ;) [20:13] anyway, ill step out for a moment [21:40] natefinch: yt? [21:59] wallyworld: unit tests appear to failing on non-ubuntu test hosts and it could be a recent change by you [21:59] damn ok, will look [21:59] wallyworld: http://reports.vapour.ws/releases/4551/job/run-unit-tests-win2012-amd64/attempt/3007#highlight [21:59] how did it land i wonder [22:00] i have meeting for the next hour will look after [22:01] wallyworld: b/c the windows tests for merge only seem to test packages under cmd/ [22:01] which is a bit shit [22:01] well that sucks [22:02] menn0: that sucks a lot [22:02] and by that I dont mean wallyworld but the limited scope of windows tests [22:03] oh i suck too [22:03] menn0: luckily looks like it will just be a skip [22:07] wallyworld: sweet [22:14] menn0: To do a migration, do I need to have bootstrapped both controllers with the feature flag? [22:14] yep [22:15] that would explain it then! [22:15] I have this in my shell config: [22:15] okthx [22:15] export JUJU_DEV_FEATURE_FLAGS=migration,developer-mode [22:17] cool - what's developer-mode? [22:18] it enables commands like dump-db and dump-model [22:18] I *think* it also enables the introspection worker [22:18] or maybe that runs all the time, not sure [22:18] babbageclunk: the world becomes pixelated around you and you hear 8 bit music, a bit like what happens in tron :p [22:18] perrito666: so it's like DMT? [22:19] wouldnt know :) but I would adventure things are a bit more nerdy [22:19] like this https://www.youtube.com/watch?v=KEkrWRHCDQU [22:20] I'm pretty sure that's how I learned to program! [22:23] babbageclunk: to be more accurate https://www.youtube.com/watch?v=KEkrWRHCDQU&feature=youtu.be&t=134 [22:27] perrito666: I had to have a one minute cad chill-out to recover https://youtu.be/mNTj05I4k5I?t=6m9s