[00:00] I added a constraint that a new environment should have the same provider type as the state server environment [00:00] well, default [00:00] not constraint [01:37] thumper: I've not been able to reproduce the bug 1412292 - are those errors happening consistently for you? [01:37] Bug #1412292: Intermittent test failure in ActionSuite.TestActionsWatcherEmitsInitialChanges

[01:37] jw4: about one time in three [01:38] thumper: hmm; do you have a fast machine, or a slow one - or is it PPC - or anything like that? [01:38] two year old fast i7, lots of ram and fast ssd [01:38] thumper: kk - I'll try picking apart the logic to see what I can find [01:38] ok [01:59] thumper: Fix it, then Ship it [01:59] thumper: two quick questions - 1) I assume the defect report has the full output of the test run. Is that true? 2) what version of go are you using? [02:00] thumper: fwiw I suspect the problem is the watcher isn't getting updated in time that the two actions were removed [02:42] menn0: ta [02:42] jw4: using what ever version is in trusty [02:42] jw4: I'm guessing it is a race condition [02:43] jw4: but I've not looked at the code [03:20] man... [03:20] getting endorsements on linked in by folk I don't remember linking to [03:21] oh... he is a juju gui peep [03:25] wallyworld_: about to send an email to fwereade for some advice on the state model, I've CC'd you in case you have ideas on it [03:25] ok [03:26] the relations to units and services are feeling a bit messy [03:26] * thumper finishes the apiserver httphandler before enfixorating the branch menno reviewed [03:50] menn0: I have another branch that fixes the rest of the apiserver [03:50] menn0: I just need to write tests for each of the api end points to exercise the additional environment [03:52] thumper: sweet [03:52] thumper: i'm deep into integrating envWorkerManager into the machine agent [03:52] awesome [03:52] I feel the pieces are slowly coming together [03:52] thumper: lots of awkward details to figure out [03:53] menn0: yep, I know [03:53] menn0: so many fiddly bits [03:53] thumper: e.g the singular runner [04:15] boo... hiss... [05:08] damn [05:08] I bet that two line tweak might break something [05:14] I'm done [05:14] these tests are taking ages to write [05:14] laters [07:05] morning all [07:13] dimitern: morning :) 5pm here tho [07:13] anastasiamac, afternoon then :) [07:14] dimitern: thnx for consideration :) [07:14] dimitern: one thing i miss about Nothern hemisphere - r long dusks in summer [07:14] anastasiamac, oh, they're not so long there? [07:15] anastasiamac, I'd have a hard time adjusting to summer in january :) [07:16] dimitern: with +40 and an awful humidity, u will not mistake this for winter! [07:16] dimitern: there is no dusk :( [07:16] anastasiamac, *lol* [07:16] dimitern: one minute it's day, the next night [07:16] dimitern: there r millions of other advantages tho :) [07:17] anastasiamac, that is a bit weird [07:17] dimitern: k... to be fair it takes about 20 minutes for day to turn to night :D [07:17] anastasiamac, I bet there are, although with all the creepy crawlers in AU.. [07:17] dimitern: and don't forget flying cockroaches :P [07:18] anastasiamac, oh we have those as well - nasty creatures [07:19] dimitern: certainly very exotic [07:23] axw: i've made some changes; i was taking some shortcuts previously to try and get it done more quickly; tests should have better coverage now [07:24] wallyworld_, axw, hey guys - can one of you review this goamz PR please? https://github.com/go-amz/amz/pull/20 [07:24] wallyworld_: ok cool [07:24] wallyworld_: looking [07:25] dimitern: will take a look in a bit [07:25] dimitern: i'm about to get ready for soccer, but can look when i get back if not done [07:25] axw, thank you [07:25] wallyworld_, cheers [07:25] wallyworld_, isn't it too hot for soccer? :) [07:26] dimitern: inly if you're soft :-) [07:26] wallyworld_, :D [07:33] axw: i gotta dash, but i might next add some basic framework for registering storage provider types eg loop with a skeleton implementation. then the blanks can be filled in while work is done on making the loop storage available to env providers, and fallback logic implemented etc [07:33] axw: tnks for review, i just used state in tests to get end-end tested for now, can mock out later [07:33] wallyworld_: that SGTM [07:33] thanks [07:57] dimitern: reviewed [07:57] axw, ta! [10:03] voidspace, standup? [10:03] dimitern: omw [11:13] dimitern: when you're free, can you please review https://github.com/juju/names/pull/37 ? [11:14] axw, sure thing [11:20] axw, lgtm [11:20] dimitern: do you mean vary the storage name? i.e. something other than "store"? [11:21] thanks [11:21] axw, yeah [11:21] dimitern: okey dokey, will do [11:21] axw, I only later read the description of what it is :) [11:50] morning [11:51] perrito666: morning... [12:00] dimitern: another small one http://reviews.vapour.ws/r/766/ [12:02] axw, ship it! :) [12:02] dimitern: thanks [12:21] bogdanteleaga: there is something wrong with http://reviews.vapour.ws/r/632/ [12:21] it lacks code [12:21] jam2: inet problems? [12:22] perrito666: its been happening for a couple of days now. They "upgraded" my local router in preparation for changing something about the networking to the building, and now every 15-30min I get a SSL session negotiation failure. [12:23] perrito666: hmm, it hasn't been created automatically so I tried to do it manually, apparently it didn't work [12:23] perrito666: the code is on github however can I update it from the web interface? [12:25] bogdanteleaga: if you did the post manually probably it worked but you need to go to reviewboard and publish it [14:04] voidspace, maas meeting? [14:07] * perrito666 actually had to prepare a brew to review fwereade's branches [14:07] perrito666, oh dear, is it that bad? [14:08] nope, I just checked the lenght and decided that i better prepare my brew before and not after [14:08] perrito666, good thinking :) [14:08] perrito666, if it helps, I think the test changes outweigh the code in both of them [14:12] abentley: this was the failing test of the last cursed email I got http://juju-ci.vapour.ws:8080/job/win-client-deploy/1319/console that is the test right? [14:13] perrito666: That looks like a problem with the machine running the test. [14:14] * perrito666 stops peeking at the local newspaper before getting an ulcera [14:26] Morning everybody [14:26] hey natefinch! [14:27] dimitern: howdy :) [14:27] hey natefinch back? [14:27] perrito666: back, yes [14:27] 058524 [14:27] natefinch: we wherent expecting you for a couple of weeks [14:28] natefinch: tx nice pass [14:28] lol [14:28] just my google authenticator number [14:28] yeah, this is about when I expected to be back.... 2 weeks after the company shutdown [14:39] dimitern: damn! [14:39] dimitern: sorry [14:40] voidspace, :) np [14:40] dimitern: just forgot :-( [14:40] dimitern: I need to tattoo it to my hand or something [14:40] voidspace, *lol* [14:41] voidspace, nothing major - non-installable nodes can be tested very soon - check the minutes doc if you like [14:41] natefinch, perrito666 dimitern : I don't see any 1.21-beta5 specific issues in play in the this bug. I don't think it should block the release of 1.21-rc1 https://bugs.launchpad.net/juju-core/+bug/1410876 [14:41] Bug #1410876: Error executing lxc-clone: lxc_container: utils.c: mkdir_p 220 Not a directory - Could not destroy snapshot %s - failed to allocate a pty; Insufficent privileges to control juju-trusty-lxc-template

1.22:Incomplete> [14:41] dimitern: I will do, cool [14:41] dimitern: sounds good [14:42] sinzui, looking [14:42] dimitern: we need an api to assign resources (like ip addresses) to them for non-installable nodes to be directly useful to us [14:42] sinzui: I agree thumper and Í documented our findings there and it seems it is outside juju [14:43] thank you perrito666 . I will try to remove the bug from rc1 [14:52] sinzui, yep, I was just checking the code - it seems 1.21-rc1 shouldn't be affected [14:53] fab [15:06] perrito666: ping [15:22] natefinch, dimitern:do you have a moment to review this branch http://reviews.vapour.ws/r/768/ [15:26] sinzui: ship it! [15:26] thank you natefinch [15:44] OCR PTAL http://reviews.vapour.ws/r/769/ - fix for bug 1412292 [15:44] Bug #1412292: Intermittent test failure in ActionSuite.TestActionsWatcherEmitsInitialChanges

[15:44] jw4, looking [15:45] dimitern: thanks [15:46] jw4, if the issue is with the watcher, why not fix it there? [15:50] dimitern: 1) I'm not sure it's a problem with the watcher 2) making a change there would have widespread effect [15:50] dimitern: the issue in the watcher is that it waits up to 10 ms every time after a change to see if there are more changes before returning [15:50] dimitern: that's probably not bad in real life, but in the tests we were expecting it to be a bit more deterministic [15:51] jw4, right, there might be an issue lurking in there, but the fix seems fine - almost ready with the review [15:51] dimitern: tx [15:55] jw4, done [15:55] dimitern: ta === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [18:29] just spent about an hour trying to work out if I can get the MACAddress out of a network definition in lshw [18:29] only to realise a few minutes ago that it's there, it's just called "serial" instead of mac address [18:29] d'oh... [18:30] at least I have it [18:36] is usermanagement so to be functional [18:38] nevermind === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [19:10] right [19:10] g'night all [19:10] EOD === kadams54 is now known as kadams54-away [19:44] when we create our lxc templates for cloning, we start them and then wait for them to stop. what causes them to eventually stop so cloning can continue? === kadams54-away is now known as kadams54 [19:51] katco: not sure [19:52] for reference: https://github.com/juju/juju/blob/1.20/container/lxc/clonetemplate.go#L162 [19:52] trying to dx https://bugs.launchpad.net/juju-core/+bug/1348386 [19:52] Bug #1348386: lxc template fails to stop

[19:56] are tags supposed to leak through the api? [19:57] hazmat: they are not [19:57] cause it seems like we started off without doing that, but now their leaking all over the place [19:57] actions, usermanager, etc [19:58] ssh key manager.. basically anything using Entities is expecting tags. [19:58] param.Entities that is [19:58] hazmat: in the api/* packages? or apiserver/*? [19:59] katco, in apiserver [19:59] hazmat: ah. that is designed to take tags [19:59] hazmat: tags are OtW format, api/* is what is client-facing [19:59] well that is the exposed api [19:59] hazmat: as i understand it anyway [20:00] api/* is an implementation of a client.. server is the api so we have a rather unfortunate situation on server that things are a mix of tags and not tags. [20:00] hazmat, katco as I recall from discussion w/fwereade the apiserver takes string representations of tags [20:01] hazmat: this could be wrong, but how it was explained to me is that tags are OtW format and so as jw4 said we pass the string-representation of tags into apiserver/* [20:01] hazmat: there was some confusion about tags vs. string representation of tags - are you referring to the string representation or the type names.Tag ? [20:01] hazmat: i.e.: tags are how clients communicate with the RPC server [20:02] none of the core api does (deploy/add-machine/remove-machine/add-units/remove-units/setcharm/etc), but all the things that have been added recently are using tags heavily [20:02] so we now have proper mismash of different ids in the api [20:03] hazmat: as I recall it that is because of a shift in focus too - the intent being that all ids are internal only and string representation of tags used ubiquitously [20:03] hazmat: we just haven't migrated everything yet [20:03] or maybe its just client facade shouldn't use tags [20:03] and other facades are all tag centric [20:04] hazmat: but I could be wrong - hoping fwereade notices and weighs in [20:04] hazmat, all facades should be tag centric -- but the client ones were mostly written before tags were a thing, and many have not been (needed to be) updated yet [20:04] hazmat, tags were introduced as a way to encode id and type for things like the annotations api [20:04] tags have been around a while, we just never exposed them through to client facade consumers [20:05] fwereade, which means they were in the first release of the api really.. ie ever since gui supported juju-core [20:05] hazmat, so we didn't need to ensure that every entity namespace was disjoint from every other [20:05] hazmat, near enough, yes, it was specifically for annotations that we added them iirc [20:07] okay.. i guess i'm going to just keep with other client impl not forcing users to care about tags, just automatically append per appropriate call [20:08] the only place there was ever a need for tag disambiguation was the annotations (and perhaps not there), i'd rather not have api consumers actually have to concern themselves with it unless by choice. === kadams54 is now known as kadams54-away [20:52] morning thumper [21:01] wwitzel3, ericsnow: can you guys post a review for the GCE stuff so I can get started on it? [21:02] natefinch: almost there [21:02] ericsnow: cool... feel free to make a PR with a big WIP on the front of it if that's helpful. [21:11] ericsnow: have a moment? [21:11] aznashwan: almost :) [21:11] aznashwan: gimme a minute [21:12] * thumper headdesks [21:12] damn... damn... damn.... [21:13] thumper: good morning [21:13] I've just realised that although the tests pass, this will fail under real load [21:13] hi katco [21:17] aznashwan: what's up? [21:18] ericsnow: i was the loonie with the systemd service management PR from Christmas Eve (namely this one: http://reviews.vapour.ws/r/671/) [21:19] aznashwan: yep :) [21:19] ericsnow: the thing is it's done exactly as upstart (namely I used exec calls to systemctl for managing the services), and since have been told that a more native approach would be more favorable... [21:21] I've recently been show this package from the coreOS guys: https://github.com/coreos/go-systemd which might be exactly what we're looking for. [21:21] aznashwan: yeah, that is what we were planning on using [21:22] It would not take too long to adapt my PR to use that, however i'm in the middle of exam season right now, and am not able to get to it sooner than this weekend [21:23] plus however much time responding to review requests and etc... would take [21:23] is that timeline ok for you guys? === kadams54-away is now known as kadams54 [21:24] aznashwan: it depends on what we end up getting done this week with regards to systemd [21:24] aznashwan: I expect that we will already have take care of things by this weekend [21:25] aznashwan: we were planning on incorporating changes from your patch in whatever we did [21:26] natefinch: http://reviews.vapour.ws/r/771/ [21:27] ericsnow: as you will, apart from changing the exec calls and (sadly) having to redo the tests, my patch is pretty much on par with the upstart one [21:27] aznashwan: cool [21:28] ericsnow: are you guys also planning to do anything about how the service management system is selected in the (very) near future? [21:29] aznashwan: yeah, that will be a part of what we are working on [21:30] aznashwan: we also have to refactor various bits of upstart-specific code throughout juju core to be init-system agnostic :( [21:31] ericsnow: yes, the places were services were used seemed very upstart-specific in the first place... [21:31] \me gears up for paying off some technical debt [21:32] natefinch: for fun take a look at the list of commits on that PR [21:32] ericsnow: well, best of luck to you guys. as i've said, I would be more than happy to do it on Saturday of you don't get around to it by then [21:33] aznashwan: sounds great :) [21:33] ericsnow: keep in touch. godspeed :D [21:34] ericsnow, wwitzel3: cool, looking [21:54] / === lazyPower is now known as lp|outy5000 [22:15] axw: i'm currently reviewing your charm storage PR [22:37] menn0: where is the statesuite that you created? [22:38] state/testing [22:38] nm [22:39] thumper: :) [22:40] thumper: I'm pretty sure I've fixed the watcher test - fyi :) [23:06] ericsnow: ping [23:07] wwitzel3: present [23:07] ericsnow: going to jump in to moonstone for a bit [23:07] k === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away [23:47] thumper: super command of super command.. [23:47] thumper: for eg. juju storage pool list.... [23:47] thumper: is there a precedence?..