[00:01] thumper: wallyworld_ axwalk (absent): trying to get you ppc testing resources asap [00:01] oh? [00:02] we doing some ppc work? [00:06] wallyworld_: yes, it turns out I know things that you don't know (yet) [00:06] that wouldn't be hard to do [00:38] wallyworld_: oh hai [00:38] yello [00:39] * thumper is munching [00:39] wallyworld_: you'll get all the goss at our meeting [00:40] wallyworld_: can I pick your brain? I'm finding it hard to understand TestWatchEnvironConfig in state_test.go [00:40] ok [00:40] might be easier via a hangout, but I'll try via irc first [00:40] It sets "default-series" to "another-series" and then uses s.State.WatchEnvironConfig to grab the config object. [00:40] * wallyworld_ looks [00:41] then it tests to that the change DID NOT take effect [00:41] this fails as it did take effect, as would be expected [00:42] WatchEnvironConfig doesn't get the config, it gets a watchr [00:42] so then, I changed the test to assert that the change did take place and that failed too [00:42] s.State.EnvironConfig() gets the config [00:43] yep [00:43] you said above it got a config object [00:43] i just wanted to clarify :-) [00:43] so WatchEnvironConfig returns 'got' [00:43] which is the (changed?) config [00:44] it then compairs that against state: [00:44] c.Assert(got.AllAttrs(), gc.DeepEquals, cfg.AllAttrs()) [00:44] no, w.Changes() returns goy [00:44] got [00:44] w := s.State.WatchEnvironConfig() returns a watcher [00:44] sorry, yes the point is the same though [00:44] sort of [00:45] so Changes returns a changed config object right? [00:45] *w.Changes() [00:46] yes, in this case i think so. it doesn;t always, it depends on the watcher. sometimes it returns an empty struct. let me check [00:46] so why would the test want to assert that there was no change by compairing the changed cfg with state cfg?!? [00:47] it doesn't [00:47] the assertNoChange method is different [00:47] case got := <-w.Changes(): [00:47] c.Fatalf("got unexpected change: %#v", got) [00:47] I'm talking about the c.Assert(got.AllAttrs(), gc.DeepEquals, cfg.AllAttrs()) line [00:47] is the no change check [00:48] that's for testing changes, not no changes [00:48] you said "no change" [00:48] it fails because the changed cfg != state cfg [00:49] in which case the changes haven't been applied [00:49] but they have and it fails [00:49] and/or sent in the event [00:50] you have some code I can look at? [00:50] yep [00:51] umm, should I put a comment next to this test, finish my branch and propose it (even though it has failing tests?) [00:51] you can create a wip merge proposal [00:51] and ask me to look at that [00:51] oh, okay [00:51] use the -wip option for lbox i think [00:52] don't worry about finishing anything, just propose [00:52] and i can easily see the current state of it [00:52] yeah? Okay [00:52] although i have a meeting in 8 minutes [00:52] oh, well, shall we aim for after our standup? [00:53] because it is wip, it won't be marked as ready for review [00:53] after standup works [00:53] great, thanks :) [00:53] propose whenever though and i might get to see if before standup [01:58] wallyworld_: could you please take a look at TestEnvironConfigWithoutAgentVersion ( you can now set an agent version - to allow upgrades) and TestWatchEnvironConfig. [01:59] ok [01:59] wallyworld_: thank you :) [02:02] waigani: i can't see TestEnvironConfigWithoutAgentVersion in the diff [02:02] what file is it in? [03:47] waigani: hi, you back? i can't see TestEnvironConfigWithoutAgentVersion in the diff. what file is it in? [03:47] wallyworld_: go/src/launchpad.net/juju-core/state/initialize_test.go [03:48] oh, i was looking in the diff [03:49] so that is an existing test. does it fail? [03:49] wallyworld_: yep - I was just writing that [03:49] where does it fail? [03:50] wallyworld_: [03:50] initialize_test.go:142: [03:50] c.Assert(err, gc.ErrorMatches, "agent-version must always be set in state") [03:50] ... value = nil [03:50] ... regex string = "agent-version must always be set in state" [03:50] ... Error value is nil [03:51] i.e. you can no set the agent-version [03:52] that test is checking that there's an error if agent-version attempts to be deleted [03:52] so i would expect the test to still pass? [03:52] ie that it is failing is a problem [03:52] there are two parts arn't there? I think just the second part fails [03:53] where it says something about setting a bad version [03:53] line 138 [03:54] there's two tests in there: initialise with missing agent-version should fail, initialise with agent-version and try to delete it should fail [03:54] the line numbers are different for you vs me, but i think i see where you mean now [03:55] waigani: link to the CL please? [03:55] axw: ah thanks [03:55] waigani: i'm not seeing what your question is [03:56] the test as written is fine [03:56] waigani: if I were a guessing man, I'd say something happened to checkEnvironConfig in your CL [03:56] axw: https://codereview.appspot.com/70190050/ [03:56] wallyworld_: axw's outline of the intention of the tests is what I was after [03:56] oh [03:56] sorry [03:57] i thought you were asking something else [03:57] np, they are failing, so I just needed to understand what the expected / correct behaviour was [03:57] in that case what andrew says is correct [03:57] so I can see if 1. tests need to change or 2. my code needs to change [03:58] and it looks like it is no. 2 [03:58] yeah, in this case [03:58] i misunderstood what you were after [03:59] no no, all good. Thanks for taking the time :) [03:59] well, i didn't help, so no need to thank me :-) [04:00] i think you should ave said "thanks for nothing" :-) [04:00] lol [04:00] axw: the main business logic is in state/state.go [04:01] yep, looking now [04:01] I'm still polishing tests etc but the general gist is there [04:01] ah, my girl just jumped off the ice - I'll be back online tonight [04:02] eep. later [07:17] wallyworld_: ping [07:17] wallyworld_: unping === Tribaal is now known as Tribaal_ [08:38] axw: thanks for the review. It all makes sense and I agree. [08:38] waigani: cool, no worries [08:38] thumper and I looked at ChangeEnvironConfig today and we both thought it could go [09:17] fwereade: https://codereview.appspot.com/69600043/ [09:28] https://codereview.appspot.com/71490045 [09:44] voidspace, description please? [10:06] morning all [10:06] morning nate! [10:06] join the weekly hangout? [10:07] mgz: thanks [10:08] wallyworld_: joining? [10:21] https://bugs.launchpad.net/juju-core/+bugs?field.tag=regression [10:21] https://launchpad.net/juju-core/+milestone/1.17.5 [10:22] https://launchpad.net/juju-core/+milestone/1.18.0 [10:22] fwereade: i'd like you to look at these because there's a subtle assumption about how major upgrade version is determined https://codereview.appspot.com/69890053/ https://codereview.appspot.com/71920043/ [10:24] dimitern: do you mean on the code review or somewhere else? [10:24] voidspace, on the code review yeah (or CL as it's usually referred to) [10:24] dimitern: what does CL stand for? [10:25] voidspace, that's a good question - commit .. something change log perhaps? [10:25] dimitern: heh, ok - thanks [10:48] axw: i forgot to ask - i think there may be some deprecated config options now that the storage stuff has been removed? eg share-storage-port? === Tribaal_ is now known as Tribaal [11:01] axw: can I run something past you? [11:16] axw: did you get a chance to look at https://bugs.launchpad.net/ubuntu/+source/juju-core/+bug/1284127 ? [11:30] voidspace: http://godoc.org/github.com/juju/ratelimit [11:32] rogpeppe: thanks [11:32] dimitern: https://codereview.appspot.com/71490045 [11:37] voidspace, you've got a review [11:55] wallyworld_: shared-storage-port can be deleted [11:55] waigani: you can, but I'm making and about to eat dinner [11:55] jam: was just about to [11:55] axw: ok, i'll add it to the upgrader. anything else? [11:56] wallyworld_: not that I can think of [11:56] ok, ta [11:56] axw: ok, feel free to answer tomorrow but here is the question [11:58] in firewaller_test.go setGlobalMode rog (rogpeppe?) has a TODO that it should not be possible to set the firewalling mode after an environment has been bootstrapped [11:58] that TODO is now done. The validation in State does not allow it. [11:59] now, there are a bunch of tests in firewaller_test.go that start by calling setGlobalMode [12:01] so is there a valid way to setGlobalMode or do I delete all the tests that relied on that setting? The latter seems drastic and I don't understand the details of the tests to feel confident doing so. [12:01] rogpeppe: you need to update the readme on ratelimit [12:01] axw: ^ (and enjoy your dinner) [12:01] waigani: thanks, will take a look after dinner. [12:03] Good luck 1.8 sprinters! [12:08] jam: I've updated the bug, we can just close it. not sure what status to give it... can you please take a look when you have a moment? [12:15] jam: re sshstorage, we don't necessarily need to create the temp dir the same, but it should ideally ignore the .tmp that filestorage creates [12:19] rogpeppe: ratelimit also needs a license file [12:20] axw: so the list get/put stuff is pretty superfluous to be honest. List() doesn't list dirs anyway, so we don't accidentally walk there [12:32] axw: so, I'm pretty sure my trapping of .tmp is pretty "we don't actually need to do this because it isn't exposed already". Do you feel its really worth following up on? [12:33] (as in, someone knowing the secrets might be able to see a file that was in progress before it finished the atomic put, but it doesn't seem to actually leak anything) [12:35] jam: yeah I suppose so, I'm being a bit anal. ignore that then - I will have a look at your updates after dinner [12:35] axw: thanks [12:37] rogpeppe, https://codereview.appspot.com/70010045/ [13:13] hazmat: playing again wuth juju-lxc, installed git and cpu-checker on the base img, but still fail to come up, get this in the lxc logs: [13:13] lxc-start 1393886691.822 ERROR lxc_cgfs - Device or resource busy - failed to set memory.use_hierarchy to 1; continuing [13:30] rogpeppe: mornin.. what ever happened with the solution you thought that you had for the watcher bug? [13:31] cjohnston: i hope to have something done today - it requires setting up something to poll the connection, so it's not totally trivial [13:31] rogpeppe: ack [13:34] bloodearnest, haven't seen that before [13:39] hazmat: ack, will keep digging [13:40] bloodearnest, i'd raise that hallyn or stgraber on #ubuntu-server [13:40] raise that to [13:40] hazmat: yep, digging a little more first [14:21] natefinch: sorry if I DC'd, I accidentally hit the turn-off-wireless when looking for the "turn-on-monitor" button [14:26] jam: heh no problem [14:36] mgz: lp:~natefinch/juju-core/037-EnsureMongo [14:48] mgz: I need to push up one more change.... evidently missed updating a test. won't take long. [14:49] fwereade, https://codereview.appspot.com/70010045/diff/20001/worker/deployer/simple.go?column_width=80 [14:58] mgz: fixed the test. My test run looks pretty good, just mongo tests that are failing that always fail for me, so it's probably fine to merge. [15:01] natefinch: ta [15:02] mgz: no, thank you. [15:11] rogpeppe, https://codereview.appspot.com/70010045/ [15:25] voidspace, would you take a look at https://codereview.appspot.com/71920043/ please? looks like it may conflict [15:25] fwereade: looking [15:36] anybody from core available for the cross team meeting? [15:38] mramm2, arrgh, might be tricky to find a room [15:39] natefinch, would you represent us please? [15:39] fwereade: was just going to offer [15:39] I need answers from you guys [15:39] like when will we get 1.18 [15:39] mramm2, fwereade: not sure I'll have all the answers [15:40] natefinch: can you join anyway? [15:40] mramm2: certainly [15:40] mramm2: link? [15:41] fwereade: I can just ask you to feed me answers ;) [15:41] mramm2, natefinch: I *think* we're on track for 1.18 tomorrow [15:41] mramm2, natefinch: it remains our highest priority [15:42] fwereade: awesome [15:42] fwereade: are you planning to break manual bootstrap for 1.18? [15:42] mramm2, we are not [15:42] cool! [15:43] great! [15:43] mramm2, agreed with abentley that a warning when we bootstrap using config from a .jenv is sufficient mitigation [15:46] cool! [15:51] https://projects.puppetlabs.com/issues/1848 go \n === Ursinha is now known as Ursinha-afk [15:55] anyone able to give [yet another] final look over at: https://codereview.appspot.com/71490045/ [15:59] mramm2: did you seem my request for wwitzel3 to get a Kanban account? mfoord already has one, but we'll need to plan some capacity around new hires [16:00] I added wayne to kanban [16:00] but for some reason haven't been able to get the add users to board screen to load [16:00] so have not yet given him permissions on the core board [16:01] and now he has board access [16:03] also RE steve gorge's query LKK is managed by Nick Tait === Ursinha-afk is now known as Ursinha [16:09] can you make sure jam sees that the LeanKitKanban admin is Nick Tait? [16:09] (in the main juju channel) [16:10] mramm2: I'll make sure he sees it [16:10] cool [16:11] I meant to type that in another window... [16:11] talking to too many people at once! === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [16:31] rogpeppe, updated https://codereview.appspot.com/70010045 [16:34] dimitern: https://bugs.launchpad.net/juju-core/+bug/1284183 [16:34] <_mup_> Bug #1284183: jujuclient.EnvError:

[16:57] natefinch: you appear to have a text conflict in your branch, otherwise you'd be my guinea pig [17:09] mgz: I saw, working on it. Getting test failures after I fix the conflict and trying to figure out why [17:31] Hi natefinch . I have a branch that addresses some of my incompetence and naivety. Do you have an few minutes to review it and check my thinking? https://codereview.appspot.com/72100043 [17:31] sinzui: I didn't realize you could fix that with code ;) [17:32] dimitern, frankban, if you have a moment would you please take a quick look at https://codereview.appspot.com/49960047/ ? most has already been reviewed, but I hit some of the api server stuff [17:33] fwereade, looking [17:37] oh dear. natefinch I think I need to understand this bug. https://bugs.launchpad.net/juju-core/+bug/1288873 . We compile a 32bit juju client on a 64bit 2012 Windows server. We install it on the same machine. Is win7+64bit unable to install 32bit? [17:37] <_mup_> Bug #1288873: Installing juju windows binarys fails on windows 7

[17:37] sinzui: looking [17:38] sinzui: 32 bit binaries should work regardless of the bitness of the OS they're built on [17:38] sinzui: and 32 bit will work on 32 or 64 bit windows [17:39] It certainly does on win 2012 server === lazyPower_ is now known as lazyPower [17:39] I will start a win7 64 bit instance and try the install. [17:39] sinzui: the error message could be a red herring. Windows isn't exactly known for being terribly accurate with its messages [17:43] fwereade: looking [17:53] sinzui: makefile changes reviewed btw [17:57] thank you natefinch [18:00] sinzui: I killed the bot merge, am still playing [18:01] mgz, understood [18:01] fwereade, reviewed === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [19:02] natefinch: I want to land you branch, but it still seems to have conflicts [19:03] (possibly not your fault, things have been landing) [19:07] natefinch: either way, the bot is up so you should be able to just set the approved bit when ready [20:40] sinzui: hiya [20:41] hi wallyworld [20:41] sinzui: do you have joyent creds you can share with me so i can help diagnose a live test failure? [20:45] just take it, you stupid bot [21:21] * thumper is now confused as to how to add a dependency to the bot [21:33] thumper: for now, the bot doesn't read the dependencies file - you need to do it by hand [21:34] if i recall correctly [21:34] the email from mgz said that it can be done by config [21:34] but I didn't find it clear [21:34] oh, i haven't read that fully yet - you used to have to do it by hand [21:48] gah... [21:48] I need to start landing shit [21:49] * thumper just racks up some dependent branches [21:52] thanks for nothing [21:53] waigani: eh? [22:00] thumper: I was trying to find a conversation between wallyworld_axw and I yesterday. My keyword search was "thanks for nothing", but I put it into the wrong box! [22:00] haha [22:00] yeah, lol [22:01] just randomly being rude on IRC, you know me (sigh) [22:02] thumper: do you know what time axw comes on? I've got a few questions for him. [22:02] waigani: normally around 2pm our time [22:02] oh, that late. Can I run the question past you? [22:03] sure [22:04] http://pastebin.ubuntu.com/7046704/ [22:04] line 13 [22:05] as there is no policy, there is no validation and agent-version can be reset [22:07] ah no, ignore me. It should be deleting the agent-version that fails. I'll update test. [22:10] pleased to be of service :) [22:20] thanks for nothing ;) [23:00] waigani: sorry, no idea on your stuckness [23:00] thumper: :( [23:01] thumper: I've got a bunch of other tests to fix while I wait for axw, thanks for taking a look [23:04] * thumper nods