[00:00] perrito666: it's buried away in the connection code in provider/gce/google [00:00] k tx :) ill dig it up [00:00] perrito666: k [00:01] perrito666: gotta run; catch me in the morning if you have more questions [00:01] tx a lot ericsnow [00:01] perrito666: np, good luck :) [01:37] ericsnow: ping on https://github.com/juju/juju/pull/3044 [01:42] marcoceppi: eric left a bit over an hour ago [04:10] master is unblocked again people [04:12] menn0: tyvm :D [04:14] Bug #1486675 changed: supportedSeriesWindowsSuite.TestSupportedSeries fails

[04:23] Bug #1486675 opened: supportedSeriesWindowsSuite.TestSupportedSeries fails

[04:32] Bug #1486675 changed: supportedSeriesWindowsSuite.TestSupportedSeries fails

[07:39] dimitern / TheMue: please take a look: http://reviews.vapour.ws/r/2423/ [07:40] dooferlad, sure, looking [07:42] dooferlad, so you went with returning an error with --networks given and these are the only tests that failed? [07:42] dimitern: yeo [07:42] yep [07:43] dooferlad, good! [07:45] * dooferlad tries testing juju with go 1.5 for fun [07:47] dooferlad, reviewed [07:47] dooferlad, btw do you have a windows machine around that you can try running juju unit tests on? [07:48] dimitern: not really. I have my gaming box, but I haven't got it set up for development [07:48] dimitern: https://aws.amazon.com/windows/ ? [07:49] dooferlad, it's fine then, np [07:49] dooferlad, after the yesterday's sync up with master net-cli *almost* got blessed :) [07:50] dooferlad, a couple of tests remaining, one of which is windows specific and "native" to net-cli - bug 1469807 [07:50] Bug #1469807: ListSuite.TestOutputFormats fails on Windows

[07:51] dooferlad, we need to sort this out before we can merge net-cli into master [08:00] cmars, tasdomas wallyworld https://github.com/wallyworld/juju/pull/29 [08:13] mattyw, tasdomas fwereade wallyworld PTAL, http://reviews.vapour.ws/r/2432/ [08:26] dimitern: are you sure you want me to remove the SupportsNetworking check in deploy.go? [08:26] dimitern: do we make sure that networking is supported somewhere else? [08:31] dooferlad, since we'll be using constraints, it's fine to drop that [08:31] dooferlad, because constraints are not guaranteed to be supported and are verified at provisioning time anyway [08:32] dimitern: thanks. Pushing changes now and will $$merge$$ [08:34] dimitern: I yesterday fixed the comments for my latest branch, only go logical trouble with the test of the FinishedWorker. the fact, that the runner stops a worker after Wait() returned nil is tested elsewhere, so due to the simpleness of the FW we need only to assure that it implements worker (simple static assignment) and if Wait() returns nil [08:35] dimitern: so I would like to drop the one I added and create a new real finishedworker_test.go [08:35] * dooferlad thinks that localServerSuite.TestNetworkInterfaces has a race [08:36] * TheMue fetches his chequered flag [08:46] dimitern: have you seen this? http://pastebin.ubuntu.com/12134188/ [08:52] TheMue, let's discuss it during standup? [08:52] dooferlad, looking [08:52] dimitern: sure [08:53] dooferlad, yes, but it's hard to reproduce on my machine [08:53] dimitern: I think I have seen the problem, but I wanted to check I wasn't duplicating work [08:53] dooferlad, there's a bug about it [09:31] dimitern: could you point me again to the test you mentioned? [09:39] TheMue, have a look how newWorkerStarter() is used [09:42] dimitern: do you mean newTestWorkerStarter()? [09:43] TheMue, yes, that's the one [09:45] dimitern: yeah, it works with the newTestWorkerStarter, which works with the testWorker which really has a goroutine in the backend. we don't have that. [09:47] dimitern: IMHO TestOneWorkerFinish() tests that any worker finishes if Wait() returns nil [09:48] dimitern: opposite to TestOneWorkerRestart() [09:49] dimitern: but here this is achieved by telling the goroutine to end with err == nil [09:57] TheMue, ok, then no need to test the finishedworker in runner_test? [09:58] dimitern: that's what I meant, yes. only one that says "yes, I'm a Worker" and "yes, I return nil on Wait()". [09:58] TheMue, just check newAddresser in cmd/jujud/agent/machine.go returns FinishedWorker{} when disabled? [10:00] dimitern: that's in the worker_test right now, but can see how to do it in machine_test too [10:02] TheMue, in machine_test, I'd suggest saving the original agent.NewAddresser value (from export_test), then patching it with another which calls it internally and asserts the result is finishedworker [10:02] dimitern: ok [10:03] all: core team meeting? [10:04] I'll skip it this time I'm afraid [10:04] wallyworld, axw is this basically what you wanted? https://github.com/wallyworld/juju/pull/31/files#diff-abee59b3dd781e907de49c2651b6a332R32 [10:09] hey people I need to take my wife to work I cant make it to the meeting sorry [11:07] dimitern: goamz has changed a lot :-| at least I found a real bug in juju (TestNetworkInterfaces relied on map ordering) [11:08] dimitern: ah, go-amz, not goamz [11:29] dooferlad, yeah [11:29] :) [11:29] dooferlad, what's the ordering bug about? [11:42] Bug #1486965 opened: Destroying the leader sets the service status to terminated [11:54] Bug #1486965 changed: Destroying the leader sets the service status to terminated [12:00] Bug #1486965 opened: Destroying the leader sets the service status to terminated [12:21] dimitern: that is https://github.com/juju/juju/pull/3052 [12:21] dimitern: no idea why I didn't create a pull request before lunch [12:22] it was a map ordering problem [12:22] https://github.com/go-amz/amz/pull/61 is about thread safety [12:22] and now, windows! [12:22] dooferlad, awesome! looking [12:30] dooferlad, both approved and the goamz one merged [12:30] dimitern: thanks! [12:40] tasdomas, cmars: I think we should talk about uniter availability again, I don't quite get how that last branch solves the problem [12:40] cmars, fwereade sure [12:56] fwereade: https://github.com/wallyworld/juju/pull/32 [12:56] axw: https://github.com/wallyworld/juju/pull/32 [13:14] dooferlad, TheMue: dimitern is battered by storms and without power [13:14] fwereade: thanks for the update. Yikes! [13:18] yet he manages to communicate with the guy in an island :p [13:20] perrito666, and whose own power is out again, the house is being rewired :) [13:21] lol [13:21] wallyworld, could you take a look at https://github.com/wallyworld/juju/pull/31 and let me know if it seems reasonable? [13:21] fwereade, you might be interested as well ^^ [13:28] ericsnow: when you ahev time, could you please take a look at https://github.com/juju/juju/pull/3035 ? it's a mp for a long-running branch that we are using to implement the embedded gui story. thanks! [13:29] hi i've got a juju-core pull request on github and reviewboard. with a 'shipit' vote from davecheney. does it require a second? when ready to merge, do you just press the 'shipit' button on reviewboard? [13:32] sinzui: Want to talk about weekly test candidates? [13:33] abentley: indeed. have you been listening to my thoughts this morning? [13:33] ericsnow: ping [13:35] sinzui: Err, not intentionally? [13:35] dooferlad: or anyone working on net-cli, ping who when bug 1469807 has a fix merged. I want to enure we test it [13:35] Bug #1469807: ListSuite.TestOutputFormats fails on Windows

[13:37] sinzui: I am looking at it now. Will let you know [13:48] TheMue: Could you take a look at https://github.com/juju/cmd/commit/4a5e8abce36462f89d92d28bd69b54c7845db818 please? [13:49] TheMue: I pushed to the wrong branch, but I have tested that this works :-( [13:56] dooferlad: yep [13:59] dooferlad: a direct usage of target with the defer of it's closing isn't possible? [13:59] TheMue: no, because the type of target doesn't have a Close [13:59] but File does [13:59] dooferlad: ah, and the assigning is a kind of downcast, yes [13:59] dooferlad: ok [14:04] natefinch: standup? [14:05] katco: oops, coming [14:06] frankban: yeah, I'll take a look [14:07] ericsnow: ty! [14:07] Ah, my last strange tests are now working too, phew. [14:07] wallyworld, another one https://github.com/wallyworld/juju/pull/34 [14:21] Grmpf, still an issue. Those two tests running together by pattern work, also isolated, but inside the whole suite they don't work. [14:24] Error is "connection is shut down" when using the api in the MachineSuite of jujud/agent/machine_test.go [14:29] TheMue: http://reviews.vapour.ws/r/2437/ [14:30] dooferlad: *click* [14:30] Just a rev bump to bring in the change I just made to cmd/juju [14:30] dooferlad: done [14:31] TheMue: thanks! === xwwt_ is now known as xwwt [14:31] dooferlad: yw [14:32] dooferlad: I'm still looking for my strange test failures. normally one test shouldn't effect another one. my both work when called single or together, but with the whole suite one fails [14:32] TheMue: timeouts? [14:33] TheMue: races? [14:33] dooferlad: connection of the API [14:33] dooferlad: here I have an idea [14:34] dooferlad: I exchange a function variable, so this has to be done atomically [14:54] cmars: you have a patch up to run the feature tests only upon request? [14:54] ericsnow, correct. though, that will be set by default in jenkins [14:54] ericsnow, so it is on request in your dev env [14:55] cmars: a big +1 from me, but why the change from opt-out? [14:56] cmars: BTW, this might have helped :/ http://reviews.vapour.ws/r/1647/ [14:56] ericsnow, unit tests are appropriate for development, integration tests are appropriate for CI [14:56] cmars: totally agree; perhaps I'm assuming too much to think there was a rationale for opt-out originally [15:02] cmars: PTAL @ my review on http://reviews.vapour.ws/r/2432/ [15:22] dooferlad: seen the mail? do you know the branches? [15:22] TheMue: just replied [15:22] TheMue: no, there is nothing on his github [15:23] dooferlad: just seen your reply [15:23] ;) [15:24] TheMue: I assume he is working on https://canonical.leankit.com/Boards/View/101652562/116729781 and maybe https://canonical.leankit.com/Boards/View/101652562/116652760 [15:26] dooferlad: yeah, and I don't see branches for it in his account [15:58] dooferlad: but at least I found my failure, a misunderstanding of the helpers of the test suite [16:20] ericsnow: ping [16:21] marcoceppi: hey [16:21] ericsnow: I had some questions on https://github.com/juju/juju/pull/3044 [16:21] marcoceppi: yep [16:22] ericsnow: > for clarification does this mean that if I have a script called `juju-foo` in my path `juju foo` will no longer work? [16:22] marcoceppi: no different concept of plugins [16:22] cool, thanks [16:22] marcoceppi: this has nothing to do with juju plug-ins [16:22] just making sure [16:22] * marcoceppi puts all the red flags away [16:23] dooferlad, TheMue, o/ [16:24] marcoceppi: yeah, what katco said :) [16:24] dimitern: o/ [16:25] marcoceppi: we are disabling bundles and the config-changed hook, though [16:25] dooferlad, so I need to merge your changes from net-cli into my branch and push it [16:25] marcoceppi: I should have made it more clear that this relates to the workload processes feature [16:26] dimitern: ok, I will look out for a review request [16:27] dimitern: will let you get on with it for now. Ping me if there is anything you want. I am idling/hangout out with the daughter (first steps!) until given somethine else to do. [16:27] dooferlad, cheers, I'll try to be quick [16:27] (since it is EOD) [16:27] dooferlad, :) ok, sounds good [16:28] I will see what I can do this evening once she is in bed. [16:28] (if anything is needed) [16:50] natefinch: wait...what [16:52] marcoceppi: lol [16:52] marcoceppi: just messing with you [17:33] ericsnow: you around? [17:33] natefinch: yep [17:35] ericsnow: have you tried the process removal code? I call it and returns no errors and... then the process still exists. Maybe I'm missing something somewhere [17:36] natefinch: there are tests, but we certainly may have missed something between the layers [17:42] cannot for the life of me get log level = trace to work [17:50] dimitern: ah, back online. I'm idling here too for any task. latest branch is ready for 2nd review (after fighting with testing troubles this afternoon) [17:51] TheMue, cool, I'll have a look then [17:51] cmars: ping [17:51] I'm ready with mine as well - all tests pass, now need to merge [17:52] top [18:06] TheMue, reviewed [18:24] dimitern: thx, will take a look in a few moments [18:29] gah, rolling over a 401k is such an outdated process... mailing paperwork and then mailing paper checks... what is this, 1970? [18:29] * natefinch is going to receive a check in the mail worth approximately half the cost of his house :/ [18:30] old brokerage: "But we're sending it priority mail, so we can track it." Fantastic. [18:39] dooferlad, TheMue, if any of you are still around I'd appreciate a quick look on this; it's quite big, but it's also well tested, including a feature test which confirms CLI works end-to-end to state for spaces list|create [18:39] dimitern: can do [18:41] dooferlad, TheMue, tyvm, http://reviews.vapour.ws/r/2440/ [18:41] *click* [18:58] dimitern: I almost had several comments, but it looks good! [18:58] dooferlad, awesome! [18:59] dooferlad, I have a similar one for subnets CLI, but it's already very late [18:59] dimitern: up to you. I can come back after I have eaten [19:00] dooferlad, well, if I'm not done within the next half hour with it, it won't happen, but there are a few tests still left [19:16] dimitern: reviewd [19:18] TheMue, thanks! [19:58] ericsnow: hey going to be just a little late [19:59] katco: no worries [20:16] Bug #1487191 opened: TestAddOkay mismatch

[20:25] Bug #1487191 changed: TestAddOkay mismatch

[20:35] did the procedure for using development tools change? I can't seem to be able to bootstrap anymore, it complains about tools not found [20:36] http://paste.ubuntu.com/12137669/ [20:37] Bug #1487191 opened: TestAddOkay mismatch

[20:46] bogdanteleaga: have you tried using --upload-tools? [20:47] nope, I have them uploaded somewhere and I'm using agent-metadata-url [20:48] it was a permissions issue though I fixed it now [20:48] bogdanteleaga: ah, cool [20:48] though, it still seems like it defaults to devel even though I tried specifying releases [20:49] bogdanteleaga: iirc, wallyworld was doing some work around that, but I'm not sure where things landed. [20:49] cherylj, I found this https://github.com/juju/juju/pull/2985 [20:50] I'll try again soon, it may have tried to read devel anyway for no reason and fail like that [22:03] anyone else having problems with reviewboard not creating review requests for PRs? [22:30] cherylj: I just put one up, worked for me [23:07] abentley: call? [23:08] waigani, thumper: Sure. I'm in https://plus.google.com/hangouts/_/canonical.com/jes-feature?authuser=1 [23:08] abentley: we are in the one defined in the calendar [23:08] jes-testing