[00:13] thumper: menn0 https://github.com/juju/juju/pull/4931 part 3 or 4 [00:13] part 4 coming up [00:31] thumper: i agree with your point [00:32] getShapshot, despite the huge comment, was never used except in one place ... [00:32] and the place where Ecfg is used [00:32] to take a "snapshot" of the config [00:32] it's called twice [00:32] so yeah, snapshot [00:32] twice == [00:32] v1 := evc.Ecfg().somethign [00:32] v2 := env.Ecfg().something else [00:32] ^ v1 and v2 are not taken from the same snapshot ... [00:58] davecheney: sorry, have tradespeople coming out my ears [00:58] looking [00:58] davecheney: never mind [00:58] davecheney: I see it's merging already [01:06] menn0: don't wrry, there are more changes coming [01:06] you won't miss out [01:07] whoo hoo :) [01:07] http://juju-ci.vapour.ws:8080/job/github-merge-juju/7109/console [01:07] ^ uh oh [01:07] the bot has stalled creating an aws instance [01:07] can anyone kick it in the head ? === natefinch-afk is now known as natefinch [02:01] hi ho, hi ho, it is off to the meeting we go [02:02] menn0, davecheney, anastasiamac ^^^ [02:02] and anyone else [02:02] omw [02:03] davecheney: I killed the stuck process [02:07] mgz: thanks! [02:13] davecheney: 5 of us in team meeting btw [02:16] davecheney: so... [02:16] # github.com/juju/juju/provider/gce [02:16] provider/gce/environ_policy.go:35: env.archLock undefined (type *environ has no field or method archLock) [02:16] provider/gce/environ_policy.go:36: env.archLock undefined (type *environ has no field or method archLock) [02:26] mgz: oops, I didn't mean to remove that [02:26] fixed [03:08] thumper: http://reviews.vapour.ws/r/4370/ pls === meetingology` is now known as meetingology === marlinc_ is now known as marlinc === xnox_ is now known as xnox [04:42] looks like AWS has gone to crap [04:42] builds are taking hours [06:13] voidspace, dooferlad, frobware: anyone around? [07:25] rogpeppe1: are you making any changes to the juju API to expose user access info? I'm going to need the same info for a new "show-model" command, so don't want to double up [07:26] axw: i haven't started yet and probably wasn't going to do it soon, so please go ahead [07:26] ok, will do [07:26] axw: perhaps we could align goals so we come up with a call that's sufficient for both of us?| [07:27] rogpeppe1: yes, sounds good. I'll forward an email to you that has the details I'll need [07:27] let me know if you need more [07:27] axw: in particular, i'm after a controller-level call that lets me retrieve the users that have access to a model [07:27] axw: thanks [07:28] axw: as well as finding out what level of access they have [07:28] rogpeppe1: yep, that's what I'll be adding - to the usermanager API [07:28] rogpeppe1: admins and owners can see all the users, and non-admin/owner can see only their own access [07:28] axw: isn't the usermanager API a model-level API? [07:28] probably don't need to send the email, that's basically all there is to it [07:28] * rogpeppe1 checks [07:29] rogpeppe1: no, you can ListModels, it would be similar [07:29] oh wait [07:29] that might be controllermanager ... [07:29] yep [07:29] axw: i think i'd expect to find it in modelmanager [07:29] axw: along with Grant and Revoke [07:30] rogpeppe1: yeah, I was lookign at the wrong thing, ListModels is in there [07:30] so it would go in there [07:30] axw: ok, cool [07:31] rogpeppe1: I'll poke you when I have something ready for review, so you can check if it'll fit your needs [07:31] probably not till tomorrow [07:33] axw: thanks [08:21] axw: reviewed http://reviews.vapour.ws/r/4371/ [08:26] Bug #1556293 changed: gce Invalid volume ID destroying environment

[10:14] voidspace, babbageclunk: care to cast an eye over http://reviews.vapour.ws/r/4364/ [10:14] frobware: looking [11:18] rogpeppe1: thanks, will address tomorrow. definitely agree about moving index closer to the package, that felt iffy [11:20] axw: cheers [12:39] Is it possible to define on a bootstrapped node which interface to bind the agent to ? [12:51] anyone know a decent way of telling if, when connected to the juju API, one has connected to a controller or a model? [13:15] Bug #1564395 opened: newly created LXD container has zero network devices

[13:15] Bug #1564397 opened: MAAS provider bridge script deletes /etc/network/if-up.d/ntpdate during bootstrap

[14:03] frobware: dooferlad: https://github.com/juju/juju/pull/4943 [14:04] frobware: dooferlad: in fact http://reviews.vapour.ws/r/4380/ [14:04] first PR for the maas2 branch [14:04] ericsnow: natefinch: morning [14:05] katco`: morning === katco` is now known as katco [14:05] natefinch: ericsnow: anything i can do to help? [14:06] katco: re-review my channels branch? http://reviews.vapour.ws/r/4269/ [14:06] natefinch: sure thing [14:07] katco: and if you're feeling ambitious, there's still some todos in that branch that need to be finished up, like storing channels in the DB for a charm [14:07] katco: oh hmm.. interesting, probably channel per service... interesting thought - you might have mysql-3 deployed from dev channel for one service and for stable for another service [14:08] blech [14:09] natefinch: i suppose that's possible... wondering why someone would do that [14:09] natefinch: i'd say definitely per models [14:09] per model rather [14:09] katco: who knows why people do things? maybe they were testing the service from dev channel, then publish it to stable and want to test again [14:09] but having 2 of the same service running, under different channels? [14:10] katco: the CLI will let you do it [14:10] natefinch: but that's what i'm saying, wouldn't you spin up different models for that? [14:10] natefinch: dev model, test, prod [14:10] katco: you should, but you don't have to [14:10] natefinch: i mean if you weren't a crazy person i mean [14:11] natefinch: but yes, if it's possible from the cli to do that, then we have to have data model integrity. no use second-guessing people's motivations [14:13] ericsnow, katco: man, I love that the resources tests are so fast.. go test github.com/juju/juju/resource/... takes 4.7 seconds on my machine. [14:14] :D [14:14] there's no reason not to run them all the time [14:14] i dream of a day when all of juju's tests run that quickly [14:14] i need to make it happen [14:14] can someone tell me how to get started with the embedded GUI? [14:14] my life will be incomplete otherwise [14:14] katco: you need a faster computer with many, many cores. [14:14] katco: yeah, definitely [14:15] dooferlad: no! we fix those tests! ;p [14:15] dooferlad: i was looking at this earlier: https://bugs.launchpad.net/juju-core/+bug/1555083 [14:15] Bug #1555083: Certificate error when visiting the embedded Juju GUI <2.0-count>

[14:15] dooferlad: you're still going to be limited to about 5 minutes as a lower bound... some test suites take that long by themselves [14:15] Bug #1564428 opened: destroying storage: destroying volumes: querying volume: The volume 'vol-214a99e7' does not exist.

[14:15] dooferlad: i think you just run "juju gui" [14:15] dooferlad: but have to compile the embedded-gui branch? dunno if it's in master [14:16] natefinch: that is where i am. 16 threads in this baby. [14:16] katco: I have compiled the embedded-gui branch and juju gui gives an error. [14:16] juju bootstrap --upload-tools lxd lxd [14:17] juju gui [14:17] ERROR Juju GUI is not available: GET https://10.0.3.165:17070/gui/3ef720bb-178d-4cb7-80e7-03b1a146cd14/: Juju GUI not found [14:17] dooferlad: i don't know, it's new to me as well [14:17] dooferlad: right now the gui doesn't pull the code from streams [14:18] dooferlad: work is still in progress on deliver [14:18] dooferlad: if you have a release tarball you can upload it manually [14:18] rick_h_: ah, will find a release. Thanks! [14:18] dooferlad: in #juju-gui you can hit up the team on the instructions for supplying manuall (equiv of --upload-tools for the gui) [14:23] * katco is so excited for 2.0 [14:23] so much goodness [14:27] Bug #1564428 changed: destroying storage: destroying volumes: querying volume: The volume 'vol-214a99e7' does not exist.

[14:29] katco: FYI, first download the latest GUI from https://jujucharms.com/juju-gui/ -- files section on the right, in /releases there is jujugui-2.1.1.tar.bz2 [14:30] dooferlad: awesome ty [14:30] katco: then juju upgrade-gui ~/Downloads/jujugui-2.1.1.tar.bz2 (or whatever the path is) [14:30] katco: then just run "juju gui" [14:30] it is magic! [14:30] dooferlad: <3 cool didn't realize you were on trunk. in beta3 there's some extra steps [14:31] rick_h_: I am running off the embedded-gui branch since that seems like the right place to fix the bug I am looking at [14:42] Bug #1564428 opened: destroying storage: destroying volumes: querying volume: The volume 'vol-214a99e7' does not exist.

[14:45] ericsnow: looks like bug 1560201 has been updated with a new repro script [14:45] Bug #1560201: The Client.WatchAll API command never responds when the model has no machines <2.0-count>

[14:45] Bug #1564428 changed: destroying storage: destroying volumes: querying volume: The volume 'vol-214a99e7' does not exist.

[14:45] katco: yeah, saw that [15:02] ericsnow: standup time [15:09] Bug #1564452 opened: panic: unexpected machine id "8"

[15:23] looking for help on how I run upgrade-juju these days with admin and default (see http://pastebin.ubuntu.com/15568439/) [15:23] everytime I try in :default I get the error about "a hosted model cannot have a higher version than the server model" [15:24] but the version number gets bumped... [15:25] ericsnow: you're OCR today I believe: http://reviews.vapour.ws/r/4380/ [15:25] frobware: dooferlad: if you get a chance ^^^ [15:26] voidspace: k [15:26] ericsnow: thanks :-) [15:26] voidspace: yep, will look in a bit. [16:17] katco, ericsnow: this a precursor to the macaroon PR, just making a bunch of functions take an args struct, so I can later add a macaroon to the struct: http://reviews.vapour.ws/r/4382 [16:17] ericsnow: look at me, pulling things out into separate PRs like a good boy ;) [16:18] Bug #1564500 opened: Cannot test packages in proposed or other archives

[16:28] ericsnow, katco: once that previous PR lands, this PR will clean up a lot (since it includes all those changes, too): http://reviews.vapour.ws/r/4381/ [16:31] Bug #1413050 changed: ec2 provider should not identify machines by security group

[16:31] Bug #1544662 changed: cannot bootstrap with mongodb3 and manual-provider

[16:31] Bug #1564503 opened: arm64 unit tests have never passed

[16:34] * ericsnow pats natefinch on head [17:01] Bug #1564511 opened: cmd/jujud/reboot tests fail with lxd container running [17:01] Bug #1564515 opened: help text for juju get-config needs improving

[17:18] natefinch: do you know about controller UUIDs by any chance? [17:31] Bug #1564524 opened: Unable to deploy openstack base using juju2 and maas 1.9

[17:46] can I get a review from someone? http://reviews.vapour.ws/r/4384/ [17:46] it's pretty easy [17:48] rogpeppe1: not really [17:48] natefinch: ok [17:50] cherylj: ship it [17:50] thanks, natefinch! [17:57] ericsnow: +100 to making the charmstore ClientConfig use url.URL instead of string.... I was just complaining about that with the csclient params [17:57] natefinch: :) [18:01] Bug #1553539 changed: Win client cannot exec bash <2.0-count>

[18:01] Bug #1557216 changed: MADE-model-worker stale master

[18:01] Bug #1559712 changed: Backup-restore: could not start prepare restore mode

[18:01] Bug #1556293 opened: gce Invalid volume ID destroying environment

[18:09] man, crazy crazy windy here today [18:09] and super warm, so all the windows are open, so it's more obvious [18:11] ericsnow: got some time to chat? [18:12] katco: sure [18:12] ericsnow: moonstone [18:35] I need a parody of "Free Falling" that replaces all the lyrics with stuff about git... the chorus could be ♬ And I'm re... I'm re-basing... ♬ [18:37] ♬ she's a git girl, loves her repos... ♬ [18:51] gah, this rebase is killing me [18:57] ericsnow: did you perchance delete resource/api/server/server.go? [18:57] natefinch: otp [19:13] Bug #1564577 opened: Cannot Bootrap MAAS 2.0 alpha 4. Could not access file ***-provider-state requested map, got

[19:25] Bug #1564577 changed: Cannot Bootrap MAAS 2.0 alpha 4. Could not access file ***-provider-state requested map, got

[19:30] natefinch: hey running about 5m late [19:31] Bug #1564577 opened: Cannot Bootrap MAAS 2.0 alpha 4. Could not access file ***-provider-state requested map, got

[19:36] natefinch: ok 15m late =/ [19:38] katco: ok, was hacking away and lost track of time anyway [20:31] ericsnow, katco: whenever you're ready, that new card can be pointed [20:33] natefinch: i'll hop in moonstone as soon as ericsnow responds [20:33] natefinch: I'm ready whenever :) [20:46] mwhudson: I want to talk to you about stripping [20:50] mgz: get a room [20:50] natefinch: but it's more fun in public... :P [20:51] mgz: bit of discussion just went down in the other channel [20:51] rick_h_: I've read it [20:51] ah ok yay then [20:52] rick_h_: I think the right call is to not change the beta3 packaging right now [20:52] mgz: rgr, requires a teste run at least [20:54] mgz where can I get more information about the multi-series stuff? [20:55] redir: a-linkin' [20:58] redir: https://docs.google.com/document/d/1ID-r22-UIjl00UY_URXQo_vJNdRPqmSNv7vP8HI_E5U/edit "Multi Series Charms" heading [21:11] CI zero for two this morning [21:11] BAD RECORD MAC! [21:13] mgz: tx [21:19] davecheney: well, that's juju/mongo zero for two... [21:20] we should talk about gating with go 1.6/mongo 3.2 shortly [21:31] Bug #1563585 changed: juju destroy-controller fails on existing controller after upgrade to Beta3 - ERROR getting controller environ: controller-uuid: expected string, got nothing

[21:32] cherylj: hey are you coming to the release standup or busy with the sprint? [21:32] katco: I am busy at sprint, but can come give some updates if you'd like [21:33] cherylj: np, it's ok [21:33] I forgot alexis is out and wouldn't be able to fill in today [21:33] sorry :/ [21:33] cherylj: np at all... don't worry about it [21:44] thumper, fwereade: I just noticed that MADE-model-workers has merged. woot! [21:44] thumper: would you mind removing the feature branch please? [21:45] yeah, by the way, that merged ;) [21:46] we're merging drop-maas-1.8-support-from-juju2 now [21:46] cherylj: thanks! (curtis too but he's not around) [21:46] he's right next to me, I'll forward your message :) [21:47] cherylj: great :) [21:48] cherylj: is there a plan to remove merged branches from http://reports.vapour.ws/releases ? [21:48] cherylj: we've accumulated a lot there [21:49] * cherylj brings it up with (nearly) the entire QA team [21:55] Bug #1563384 changed: could not exit restoring status: connection refused

[21:55] Bug #1563418 changed: TestModelDestroy fails

[21:55] Bug #1564622 opened: Suggest juju1 upon first use of juju2 if there is an existing JUJU_HOME dir

[22:02] menn0: yay [22:30] mgz http://paste.ubuntu.com/15571871/ [22:40] * menn0 wonders if the dep engine's cycle detection isn't quite right [22:49] Can I get a review from someone? http://reviews.vapour.ws/r/4389/ [22:56] cherylj: looking [23:05] dep engine cycle detection is fine... hoorary for tests and well written code [23:05] * menn0 had mucked up [23:06] ping https://github.com/juju/juju/pull/4905 [23:06] ^ bot didn't pick it up [23:06] Go 1.6 compile time really sucks [23:06] probably 'cos its on a dev branch [23:06] or something [23:10] cherylj: reviewed [23:11] thanks axw. And, I can't use ec2ErrCode there because the error is wrapped by an intermediary call [23:11] cherylj: ec2ErrCode(errors.Cause(err)) ? [23:11] axw: I can if I do it on the Cause [23:11] if that's ok [23:12] yes, please [23:12] kk [23:12] thanks [23:19] ec2ErrCode should probably do that internally [23:19] +1 [23:19] sounds good [23:45] Hello :-) [23:45] quick questions from #juju [23:46] is maas 1.9 and juju 2.0 supported, or known to work? [23:49] arosales: yes [23:50] mgz: ok, thanks [23:50] arosales: http://reports.vapour.ws/releases/3830/job/maas-1_9-deploy-trusty-amd64/attempt/2153 [23:51] arosales: or just http://reports.vapour.ws/releases/3830 for all the maas 1.9 tests (including deploying centos and windows workloads, bundles, etc) [23:51] mgz: thanks for the confirmation and links [23:57] arosales: I tested it earlier on OB24, it works [23:57] has anyone seen kill-controller fail? [23:58] or aware of an issue that kill-controller doesn't actually reap machines? [23:58] marcoceppi: yes [23:58] LiftedKilt: is there a bug for it? [23:58] marcoceppi: It just hung after confirming I wanted to delete it [23:58] LiftedKilt: I'm talking about it exiting and not cleaning up the environment [23:58] marcoceppi: I'm not sure - I wiped everything and rebootstrapped [23:59] I ran kill-controller, it killed one machine and then left the 22 others [23:59] marcoceppi: cool