[00:11] wallyworld: almost there, but I need to duck out [00:12] wallyworld: I'll push the code now, and fix the remaining bugs when I get back [00:12] axw: no problem, ty [00:13] wallyworld: https://github.com/juju/juju/pull/4469 [00:14] ta [00:26] Bug #1547268 opened: Can't bootstrap environment after latest lxd upgrade [00:29] fwereade: are you still lurking? [01:07] wallyworld: bugs fixed, PTAL [01:10] looking [01:11] axw: did you see my last question about map/slice? [01:12] wallyworld: I didn't. no, it won't work, goyaml will unmarshal it as the default types (MapItem is (interface{}, interface{})) [01:13] and we can't do a type assertion? [01:13] wallyworld: no... that's never going to work. goyaml just sees a MapSlice, it doesn't know you want it to be a region [01:14] ok [01:14] axw: also, change the version to beta1 in the PR so we land one change [01:15] wallyworld: sure, just about to rebase on master, will do that [01:15] axw: and just for sanity since we need this to get blessed with one run, test live? [01:15] wallyworld: will do [01:15] ta [01:17] axw: o/ [01:17] katco: heya [01:17] axw: don't need anything, just thought i'd say hey while the opportunity presents itself :) [01:17] katco: :) [01:18] katco: I was looking forward to having standups with you and the others, timezones suck [01:18] axw: that they do. i'm positive we'll be in one together sometime in the near future [01:19] maybe I'll get up before the crack of dawn one day [01:19] axw: haha, that will be a nice surprise, but certainly not expected [01:19] axw: wallyworld said, "we'll all just move to st. louis" which i think is an awful idea [01:20] axw: of course i'm not sure i'd like to live in the land of giant spiders either :| [01:20] katco: heh :p [01:20] they mostly just sit there and watch you [01:21] that's even worse! [01:25] wallyworld: what magic do you use to preserve the line ending in setup.iss when updating the version? or does it not matter? [01:26] axw: i think my IDE respects what is there [01:29] axw, in every editor I have used to edit setup.iss, all is fine, but git's diff makes the line ending look scary [01:29] sinzui: ok, thanks. I'll sed it and hope for the best [01:33] wallyworld: https://github.com/juju/juju/pull/4470 <- same changes + delta1->beta1 [01:34] looking [01:34] axw: lgtm === beisner- is now known as beisner [02:28] wallyworld: I'm going to copy&paste the bootstrap release notes as a reply to your email [02:28] wallyworld: so people know how to bootstrap... and thye can vet the notes at the same time [02:28] axw: ah yes, good idea, ty [02:29] axw: i am so used to the new syntax i forgtot others wouldn't be :-) [02:29] other devs [02:34] wallyworld: I can update the list-clouds output if you like [02:34] if you're not already doing it [02:36] wallyworld: list-clouds output order is not right :/ must be another map somewhere [02:37] damn [02:37] in the cmd i guess [02:37] wallyworld: yep, in cmd/juju/cloud/show.go [02:38] axw: but the default regon is correct. so we can add a note in the release notes [02:38] that list-clouds will be fixed next beta [02:38] axw: i'm out for a bit so will pick up any remaining release notes issues when i get back unless you get to them first [02:38] wallyworld: sure [02:39] enjoy lunch [02:50] gaaaah uniter tests [03:43] thumper and/or waigani: http://reviews.vapour.ws/r/3907/ [03:46] pending tests, I *think* I have relation imports working [03:46] well, haven't tried to compile yet [03:46] but the theory seems sound [03:47] menn0: I'll take a look now before writing tests [03:47] menn0: oh, was just going to look .... [03:48] thumper: I like your optimism :p [03:48] * menn0 facepalms [03:48] waigani: you can look too [03:48] that's fine [03:48] thanks :) [03:49] some people I know in Brisbane (acquaintances) just called their son "Pheonix" [03:50] at least check the spelling of a word before calling using it for your child's name... [03:51] lol [03:51] heh [03:52] are they going to "fix" it? [03:52] friends of mine called their kid Orion... she's a girl [03:52] or is this one of those new alternative spellings? [03:53] at least not "north" or another direction [03:53] haha [03:54] another friend named their kid Aoife... which is evidently irish and pronounced eefah ... but yeah, good luck kid. [04:12] menn0: done [04:13] thumper: thanks [04:21] thumper: man, I'm clearly tired. lots of stupid stuff [04:24] :) [04:37] All, what is the client facade in the api server ? [04:37] is the apiserver a client of itself ? [04:37] is this a version of the human centepede ? [04:49] no, seriously [04:49] what is the Client facade that the API exports [04:49] axw: do you know [04:50] davecheney: it's the facade that does all the things that the CLI client wants to do. we've been slowly splitting bits out into new, more specific facades [04:51] davecheney: what axw said. It has (predictably) grown into a kitchen sink of functionality. [04:51] davecheney: e.g. the Service facade for deploying. that used to be on the Client facade [04:51] davecheney: nothing new should be added to Client now [04:51] so apiserver/client [04:52] imports api/ to get api.CharmInfo [04:52] this is 3 inches away from an import loop [04:52] client is a shitty name for an api endpoint [04:52] it would be like me having a child and calling it "person" [04:57] thumper: review responded to: http://reviews.vapour.ws/r/3907/ [05:02] hey can anyone get to ap-southeast-2 at the moment ? [05:07] ok, so trying to hit aws.amazon.com is routing via tokyo [05:08] your packets have a hankering for sushi, I guess [05:08] om nom nom [05:09] well, I've officially added a new test to the uniter. And it only took 3 days. [05:10] that's not fair. The test only took a day, it was 2 days of wrestling with the existing tests to figure out why they were failing. [05:11] but the test is only 3 lines, which obviously saved me a lot of time. [05:12] :) [05:32] we have two CharmInfo structures delcared in the client [05:32] api.CharmInfo [05:32] api/charms.CharmInfo [05:32] !!!! [05:33] THEY ARE IDENTICAL [05:33] bah, extra tests can wait [05:33] and in fact they have the same method on their type === thumper is now known as thumper-afk [05:35] axw: what the F [05:35] https://github.com/juju/juju/blob/master/api/charms/client.go#L15-L43 [05:35] https://github.com/juju/juju/blob/master/api/client.go#L161-L178 [05:36] davecheney: I suspect the one in api should have been deleted [05:36] trying to figure out which one is unused now [05:37] davecheney: it should be in the params package anyway [05:37] wait... there is one in the params package [05:37] sorry [05:37] never mind. [05:40] the api/charms package's CharmInfo method is not used [05:40] if I remove the method [05:40] then api/charms no longer needs to import "gopkg.in/juju/charm.v6-unstable" as charm [05:40] which undermines the name of the whole package [05:42] davecheney: sounds good. that apiserver/client code for CharmInfo might still be required for the GUI, not sure. should be safe to remvoe the api code tho [05:43] if the gui needs it [05:44] all it cares about is the json form [05:44] which is what I was checking [05:44] but it turns out that no copies of CharmInfo have any json tags [05:44] so we can just replace them with an anon stryct [05:44] so we can just replace them with an anon struct [05:47] axw: besides i'm removing code from api/ [05:47] so the gui won't be affected [05:48] davecheney: yeah, I'm saying that's fine, but we should check before going further and removing the backend from apiserver/client [05:48] not that you said you were going to, just in case [05:53] axw: https://github.com/juju/juju/pull/4474 [05:53] yeah, i won't remove that method [05:53] just it's dependency on a type defined in it's client [05:53] just its dependency on a type defined in its client [05:56] davecheney: uh sorry, lost in translation. I thought you were removing them from the api/client.go code. I'm pretty sure the code should be changed to use api/charm instead of api [06:04] man, I'm going to start squashing all my branches down to a single commit... it makes rebasing and cherry-picking and all that garbage so much easier. [06:11] axw: on that review where I'm trying to get API params to show during tests... you said to tweak the logging instead. Do you know where that would be done? JujuConnSuite or something? [06:12] natefinch: if it's for a full stack test, which I assume it is if you need logs, that'd probably be the appropriate spot [06:45] axw: looks like CI is borked so lets sneak this one in http://reviews.vapour.ws/r/3911/ [06:46] wallyworld: ... http://reviews.vapour.ws/r/3906/ [06:46] damn :-) [06:47] axw: but does map slice work as expected for json [06:58] wallyworld: it'll come out as a [{Key:name, Value:{...}}] [06:58] wallyworld: not ideal. hadn't thought of JSON [06:58] axw: i think my solution works for json [06:59] i could add you test stuff to my branch [06:59] wallyworld: yes please [06:59] sure, will do [06:59] wallyworld: except mines based on formatting/parsing the YAML... [07:00] wallyworld: maybe the YAML format could use Regions instead of RegionsMap? [07:01] axw: to preserve order, yeah, good point [07:01] i'll rework it a bit [07:16] wallyworld, axw, just running make check now on master triggered my chrome to open up and ask for credentials - could it be related? [07:16] credentials for what? [07:16] well, the sso screen [07:17] dimitern: nothing to do with our changes. *maybe* related to macaroons? [07:17] axw, ah, could be [07:17] wallyworld: what version of lxd do you have? [07:18] I dist-upgraded and now lxd provider doesn't work. [07:18] axw: sadly i updated today and so i suspect i am borked [07:18] :/ [07:18] yep, known issue, casey files a bug [07:18] they don't test juju with their stuff it seems [07:19] bug 1547268 [07:19] Bug #1547268: Can't bootstrap environment after latest lxd upgrade [07:21] wallyworld: lxd_2.0.0~beta2 is OK, FYI [07:21] just uninstalled and installed the deb from cache [07:21] axw: yeah, i think i was running that one priot to the upgrade === urulama__ is now known as urulama [07:30] axw: PTAL, also fixed a bug where we were printing storage endpoint in region details even if it was the same as the cloud default storage endpoint [07:32] wallyworld: the same applies to Endpoint [07:33] doh [07:33] of course [07:42] wallyworld: are we pushing the register one through as well, or wait till after? [07:43] axw: it seems pretty safe doesn't it [07:44] will only fail in unit tests [07:44] wallyworld: yeah, also pretty safe to leave it out tho. assuming people don't expect to be able to upgrade from a beta to release... [07:45] we don't support upgrade, and models.yaml is changing also anyway [07:45] true [07:46] wallyworld: I'll just leave it till later. no need to rush it through [07:46] +1 [07:46] i hope this current change gets through CI in time [07:46] shoudl do [09:09] dimitern: ping [09:12] frobware, pong [09:12] dimitern: can we do standup now? james is out, as is michael, and I want to futz with my hosts /e/n/i. :) [09:12] frobware, sure, why not - omw [09:15] dimitern: ah, I suspect I'm in the wrong HO... tis Friday. === urulama_ is now known as urulama [14:04] https://bugs.launchpad.net/juju-core/+bug/1547268 [14:04] Bug #1547268: Can't bootstrap environment after latest lxd upgrade [14:05] latest lxd (2.0.0~beta3) upgrade seems to break the LXD provider ^^ [14:06] wily does not seem to be affected, but if you're installing lxd from the lxd-stable PPA on trusty, or testing xenial, you'll see it [14:06] also, i can't seem to juju init with latest master. gonna open a bug... [15:04] ericsnow: hey i left several comments on your download resource metadata pr... lmk if you want to chat about it [15:04] natefinch: how goes the uniter tests? [15:05] katco: got them working last night, finally. [15:08] natefinch: woohoo! ah i see your review now [15:08] natefinch: moved your card for you [15:08] natefinch: can you give ericsnow's prs a review? [15:09] katco: will do [15:29] katco: thanks [15:44] natefinch: review up [15:44] katco: thanks! [15:58] Bug #1547186 changed: Cannot create model with Azure provider

[15:59] does anyone know how livetests work? [16:04] perrito666: nope [16:05] they seem to be mocking ec2, not sure where [16:06] perrito666: hey i know a bit how they work [16:07] katco: can you give me a hint? :) [16:07] perrito666: no, i just wanted to let you know i knew :) 1 sec [16:07] perrito666: https://github.com/go-amz/amz/blob/v1/ec2/ec2test/server.go [16:08] perrito666: except you probably want latest, so make sure you hit the v3/v4 branch [16:08] frobware, hey [16:09] perrito666, mfoord do you guys know if LP 1435283 is entirely fixed? we are seeing a weird behavior with 1.25.0 (http://paste.ubuntu.com/15132753/) , basically is picking any address as the public-address instead of using a sticky one since first election. [16:09] frobware, I've proposed the next step: https://github.com/juju/juju/pull/4481, a few things still remain around tests, but I need to go out, so I'd appreciate if you have a look later [16:09] niedbalski: have no clue [16:10] dimitern, can you take a look ? ^^ [16:10] fwereade, if you're around, I'd very much like your thoughts on that as well ^^ [16:10] katco: thanks taking a look, I am pretty sure the filter I need is not implemented, because such is my luck [16:10] * dimitern reads scrollback [16:10] perrito666: =/ lmk if you need anything else [16:12] niedbalski, well, the fix for that bug in 1.25 IIRC was to ensure whatever gets picked for private/public address sticks for the lifetime of the machine [16:12] the story for 2.0 is a lot better though [16:13] * dimitern needs to step out, perhaps back later [16:16] dimitern, well, it doesn't seems to be the case with 1.25.0 http://paste.ubuntu.com/15132831/ [16:20] dimitern, sorry, I don't think I have any special insight -- did I michael EOWing the other day? [16:27] ericsnow: review on your other pr [16:27] katco: thanks! [16:27] katco: you're tearing it up! [16:27] ericsnow: uninterrupted blocks of time! [16:28] katco: get used to it :) [16:28] katco: (and the consequences ) [16:31] ericsnow: natefinch: master merged into the feature branch cleanly. merge PR CI run underway [16:32] (god bless the component oriented approach) [16:32] bbl, [16:32] katco: btw, thank you for the pointer it was very helpful. [16:32] perrito666: np, i'm hear for anything you need in the future [16:32] perrito666: i am the ian of the western hemisphere [16:33] perrito666: bleh... s/hear/here/g [16:34] katco: +1 (long live the component-oriented approach!) :) [16:36] katco: awesome about merging master [16:50] ericsnow: natefinch: 7 line change: http://reviews.vapour.ws/r/3916/ [17:10] fwereade, sorry for the mix up, I meant if you can have a look at this PR https://github.com/juju/juju/pull/4481 [17:16] hi o/ [17:16] did we nuke kvm as a provider option when we nuked the local provider? [17:16] I am trying to set up juju 2.0 and I really need kvm [17:16] mbruzek really wants to know [17:16] I get an error when bootstrapping kvm. [17:17] ERROR model "kvm" has an unknown provider type "local" [17:17] Does anyone know how I can use kvm in 2.0? We need that because containers currently don't run in lxd. [17:18] lazyPower: mbruzek: if we did, that was unintended [17:18] tych0: do you have any insight? ^^^ [17:18] mbruzek - app-containers *ftfy [17:18] katco: I tried changing the provider to "kvm" that did not work either. [17:19] katco: I was really happy with the kvm support in juju 1.25 it was nearly as fast as the lxc provider, and it let me run docker containers inside kvm machines. [17:20] mbruzek: good to hear. we'll figure out what's going on. it should still work, so either an oversight or we've changed how you interact with it [17:21] katco: Yeah if there is a new way to configure it I would be willing to test those instructions out. [17:22] katco: I checked the docs and seem to be configuring it properly [17:22] https://jujucharms.com/docs/devel/config-KVM [17:28] ericsnow: natefinch: master merged into feature-resources. i don't expect you to have conflicts, but might want to rebase all the same. [17:28] katco: k [17:28] * katco lunches [19:27] natefinch: I just reviewed your patch [19:27] ericsnow: thanks! [19:48] ericsnow: in theory, resources in the DB shouldn't ever change unless their bytes change, right? It would be kind of bad if a resource got updated with a different origin or type or something [19:50] natefinch: that's mostly true [19:50] ericsnow: there's two logical parts of a resource - the definition as exists in the charm metadata, and the bytes + metadata about the bytes [19:51] natefinch: if I remember right, there's also info for a resource that will have bytes stored soon [19:53] natefinch: e.g. charm store resource where no unit has called resource-get since the last time we synced up with the charm store info for that resource [20:25] ericsnow: you got a pile of comments dumped on you. working through them? disagree? [20:25] katco: working on it :) [20:25] ericsnow: kk [20:25] katco: I'll let you know [20:26] ericsnow: natefinch: just keep in mind we have 4 points to complete next week to stay on track. i have 2 of those in-flight, but waiting to see how ericsnow's poller patch turns out [20:27] ericsnow: also, i think i'll need to ask you to migrate the state stuff since you're most familiar [20:27] katco: k [20:32] katco: I think that other 2 pointer will be pretty quick. I think we're in good shape, but definitely if we can get ahead of the game and get more delivered than scheduled, we should try to do that. [20:33] natefinch: i agree on all counts [20:33] katco, natefinch, rick_h_: doesn't there need to be a way to tell the controller to use a different charmstore revision of a resource? [20:34] ericsnow: what's the use case for that? [20:34] I thought it was juju upgrade-charm, to use the latest in the charmstore... [20:34] natefinch: ericsnow: ditto [20:35] I'm not sure the charmstore even exposes anything other than the current tuple [20:35] natefinch: well, i think we're defining that [20:35] katco, natefinch: ah, the charm revision in the charm store isn't tied to a particular set of resource revisions [20:35] katco, natefinch: but on the controller it is [20:35] never mind then [20:36] ericsnow: i think we will be coupling those concepts in our new set of api endpoints [20:36] katco: k [20:36] ericsnow: honestly, i probably need to spend next week digging into that [20:41] urulama: ping [20:42] urulama: we need to talk about resources in the charm store API (e.g. that charmrepo PR I put up) [20:43] ericsnow: sure ... i've already pinged the team about that [20:43] urulama: k [20:44] ericsnow: sorry, but it's 10pm here and i really want to finish :) [20:44] urulama: no worries :) [20:44] ericsnow: so, talk to Francesco and Roger next week [20:44] k [21:08] Bug #1547665 opened: juju 2.0 no longer supports KVM [21:29] ericsnow: katco natefinch yes, this is all tied to the charmstore apis and updates required there. [21:32] ericsnow: standup time [21:55] seems that I drew the short straw for all neverending tests... [22:04] meh, hangout died no me [22:04] perrito666: i think it kicked us all out [22:04] oh I feel better now