[00:39] http://edition.cnn.com/2015/09/14/travel/aso-volcano-erupts-japan-flights/index.html [00:39] welp, i guess we're not going to be sprinting in japan [00:54] well prices will most likely plummet, we might [01:08] * perrito666 works a bit more to procastinate writing tests for his personal projects [03:26] jam, rogpeppe: it's "in progress" in the distro sense of the word [03:27] i'm not sure when the fix will hit trusty-updates [03:27] treat this as an opportunity to learn and go ask a distro person :-) [03:27] (i'll do the same tomorrow when i'm working again) [04:36] Bug # changed: 1493887, 1494356, 1494441, 1494864, 1494913 [08:07] morning all [08:08] mfoord: or night :D [08:08] anastasiamac: o/ [08:08] anastasiamac: it's always morning on the internet... [08:08] :) [09:22] * dimitern rebooting [09:53] fwereade: if I create a new error, ErrNoAddress, does that belong better in network or in errors [09:53] fwereade: my guess is errors alongside all the others [09:54] mfoord, heh, I am desperately ambivalent there :( [09:54] fwereade: hehe, cool - thanks [09:56] mfoord, I do worry that the error types in errors tend to become functionally useless re: detecting specific scenarios, because they often and up getting used in more and more contexts and having their meaning diluted [09:56] fwereade: interesting [09:57] mfoord, NotFound, for example, now means so many things that you'd be crazy to take any potentially-destructive action based on encountering one [09:57] right [09:57] NoAddress seems pretty specific, but then perhaps that's more of an argument for putting it in the network specific package [09:58] mfoord, indeed :) [09:58] returning an ErrNoAddress from machine.PublicAddress() now causes 20 unit test failures in the state package [09:58] mfoord, ouch [09:58] so I'm still not sure it's a good idea... looking at how easy to fix they are [09:58] most of those places were *probably* checking for an empty address anyway, so it shouldn't be too much work [09:59] they certainly *should* have been [09:59] mfoord, right, how many of my comments are actually addrssable is very much a judgment call, and you're right there by the code [09:59] yup, thanks [10:29] dammit... upgraded to vivid and it broke my ethernet :/ [10:37] god, I miss trusty [12:49] Bug #1495952 opened: Return specific error cause when trying to add existing services or relations [12:52] Bug #1495952 changed: Return specific error cause when trying to add existing services or relations [12:59] perrito666: could you please take a look at http://reviews.vapour.ws/r/2633/ ? thanks! [13:01] dooferlad, are you in the webex conference? [13:04] frobware: I have the java thing loaded. Failed on the phone bit [13:05] dooferlad, I dialled in too. Odd. [13:07] Bug #1495952 opened: Return specific error cause when trying to add existing services or relations [13:07] frobware: got it :-| [13:08] dooferlad: do you have a hangout sharing your screen? [13:09] dooferlad: the java icedtea plugin is just hanging for me [13:09] voidspace, it would be a silent movie [13:09] ah... [13:09] voidspace: no, but you can dial in and look at the presentation. Slides are in the meeting invite [13:10] dooferlad: frobware: dialling in [13:34] voidspace, dooferlad: the amount of (my) data they want to own is scary. [13:34] frobware: yea, they just want to watch your every move including what you eat. Nice. [13:42] frobware: dooferlad: :-) [13:52] dimitern, want to sync up w.r.t to subnet/zones? [13:52] Bug #1495978 opened: Juju does not deploy CentOS images in LXC

[13:53] frobware, yeah - in ~7m (at the hour) ? [13:57] * fwereade will be back for a bit sometime this evening, but has a doctor's appointment in an inconvenient place [13:57] ping me if you need me and I'll see them later [13:58] fwereade: good luck [13:58] Bug #1495978 changed: Juju does not deploy CentOS images in LXC

[13:58] btw, "doctor's appointment in an inconvenient place" is a beautiful euphemism for a proctologist [14:00] haha [14:00] frobware, let's use the standup HO? [14:01] perrito666: like the back of a volkswagon? [14:02] dimitern, in there now... [14:03] frobware, coming [14:04] Bug #1495978 opened: Juju does not deploy CentOS images in LXC

[14:07] Bug #1495978 changed: Juju does not deploy CentOS images in LXC

[14:10] Bug #1495978 opened: Juju does not deploy CentOS images in LXC

[14:28] dimitern: fancy a review? http://reviews.vapour.ws/r/2656/ [14:35] or anyone else please? It's been almost 24 hours and I'd really like one review from someone in juju-core. http://reviews.vapour.ws/r/2656/ [14:37] perrito666: (you're OCR, right?) [14:38] rogpeppe: perrito666 was OCR last week instead of me (but it's Wed 1am my time now) [14:38] anastasiamac: ok, np [14:38] rogpeppe: sorry i did not get to ur PR - was delaing with a blocker ;( [14:39] anastasiamac: that's good - the blockers are a pain! :) [14:39] rogpeppe: if noone will have a look, i rpomise to look at it in about 9hrs :D [14:39] promise* [14:39] that sounds like the correct order, as rog can't land his branch till master is unblocked anyway [14:39] anastasiamac: thanks for the offer :) [14:40] mgz: the problem is that i'm trying to coordinate upstream branches and if this gets rejected then it's a right merge pain [14:44] rogpeppe: ship it [14:44] natefinch: ta! [14:44] rogpeppe: welcome :) [14:46] rogpeppe: sorry I was at the console, I can review your patch anyway :D [14:46] or instead nate did it [14:46] :p [14:46] perrito666: "at the console" - wow, you're using a mainframe? [14:50] rogpeppe: oh ffs, one cannot translate directly :p [14:50] I was paying attention at my terminal emulator (which incidentally is konsole, almost the same) [15:06] ericsnow: hey, looks like we need to ping mgz re: bug 1495681 sooner rather than later. can you handle that? [15:06] Bug #1495681: quickstart delployments broken in 1.24

[15:06] katco: sure [15:07] ericsnow: ty [15:12] ericsnow, please work with mgz on any questions regarding test results [15:12] alexisb: will do [15:13] mgz: ping [15:22] ericsnow: sorry, waving hands at people, I am around to help [15:22] mgz: thanks [15:24] ericsnow: so, we have three changes on 1.24 with the regression, none of which jump out at me as the cause [15:25] mgz: from what I can see initially, the config-get hook command isn't working right (or something related to it) [15:26] Bug #1468815 changed: Upgrade fails moving syslog config files "invalid argument"

[15:26] Bug #1496016 opened: jujud uses too much memory [15:26] mgz: bb38bf13c00283f572f05139428961b259275079 is pretty closely related [15:27] mgz: I meant 629fc0507c4b61be93fd79453244cb8fc853f413 [15:27] natefinch: ^^^ [15:29] hmm [15:29] ericsnow: so, we have a couple of options then [15:29] can build 1.24 locally with that change removed, run the test locally, see if it passes [15:29] then just back out the change for releases [15:30] mgz: i don't think we can release 1.24.6 w/o that fix [15:30] mgz: it is for a major vendor [15:33] the other option means getting the fix fixed :) [15:34] mgz: either way, sounds like first step is testing w/ that fix reverted? meanwhile ericsnow can continue examining the error? [15:34] katco: sounds good [15:34] mgz: you'll test with that change removed? [15:36] this seems quite likely to be something broken from my change... I think the problem is that the default value for settings aren't getting set [15:36] ericsnow: okay, I'll test that locally [15:36] natefinch: yeah, that's what I was just thinking too [15:36] and we'll reserve CI for testing a 1.24 with a fixed fix [15:36] which is why the dict is missing the value, instead of having a default value [15:36] natefinch: you read my comment on the bug, didn't you [15:37] ericsnow: nah, just read the stack trace in the log and then thought about what changes I'd made to the settings. [15:37] natefinch: :) [15:41] Bug #1496032 opened: backups restore won't create bootstrap on GCE [15:47] Bug #1496032 changed: backups restore won't create bootstrap on GCE [15:48] saw a quote once from Rob Pike about working with Ken Thompson... that Ken could usually figure out a bug faster by just sitting and thinking about the code, than Rob could by running it through a debugger. [15:52] natefinch: but debuggers in the 70's were so fast! :) [15:53] Bug #1496032 opened: backups restore won't create bootstrap on GCE

[15:54] k ppl bbl [16:08] natefinch: looks like there's a semantic difference between setting the service settings directly and calling UpdateConfigSettings()... [16:09] natefinch: ...that they treat "unset" setting differently, which has a direct bearing on how defaults are applied later for each unit [16:09] s/treat// [16:10] ericsnow: ug.... I hate all this implicit code that is so side-effect driven... so easy to break things that don't work in obvious, straight forward ways [16:11] natefinch: I'm double-checking that UpdateConfigSettings discards "unset" settings, which is what I expect it does [16:28] ericsnow: thanks, currently wrestling with my 8 month old (harder than it sounds), so not able to do much for another 45 mins or so until my wife gets back [16:28] natefinch: np [16:49] fix for LP:#1495591 http://reviews.vapour.ws/r/2668/ [16:49] Bug #1495591: TestRunCommand fails on windows

[16:50] cmars: thanks [16:52] cmars: that bug looks like the listener is getting closed twice(?) is that the right place to fix it? [16:53] Bug #1449210 opened: cloudsigma index file has no data for cloud

[16:55] bogdanteleaga, where else is it getting closed? [16:56] cmars: that's the first thought I had when I saw element not found, no idea where it could get closed [16:56] dimitern, you still around? [17:05] bogdanteleaga, i don't think it is getting closed twice.. if it were, i'd expect to see duplicate "juju-run listener stopped" messages [17:20] Bug # changed: 1493623, 1494121, 1494912, 1494917 [17:21] cmars: https://github.com/juju/juju/commit/1fbd501e88ee29456b7c471ce617877f6f1393ad#diff-7842644c47e14851e8a4649bfe2ade13R125 [17:21] cmars: looks like we've been ignoring them for quite a while now [17:22] still doesn't seem like the right approach is to continue ignoring it [17:26] cmars: I think it's worth asking natefinch about this since the error is most likely coming from inside npipe [17:32] natefinch, regarding above ^^, can you review http://reviews.vapour.ws/r/2668/, or advise on a better solution? thanks! [17:33] mgz: did that updated deployer test finish yet? [17:53] cmars: I can look, yeah. [17:53] natefinch, much thanks! [18:19] mgz, sinzui: where can I find the bundle that is used for the quickstart tests? [18:22] ericsnow: it's a version of https://jujucharms.com/u/landscape/landscape-scalable/9 in their test repo. [18:22] rick_h_: thanks [20:35] Bug #1496127 opened: [juju-log] help is poor and results are confusing

[20:39] rick_h_: quickstart seems to be hard coded to use /usr/bin/jujud rather than just using the first one in the path? [20:40] natefinch: hmm i know you can env var it but not on path list [20:41] natefinch: checking [20:41] rick_h_: hey [20:42] thumper: party [20:42] rick_h_: after saying thursdays are good (which they are generally), I can't do tomorrow [20:42] kid school thing to be at [20:42] thumper: ah ok [20:42] thumper: tomorrow is wednesday ;) [20:42] * natefinch is hilarious [20:42] I can move the normal meeting [20:42] sorry [20:42] ugh [20:42] thumper: I've got a think fri (my thurs) but wanted to chat soon to get things in order pre-seattle if we can [20:42] brain futz [20:42] rick_h_: I have 15 minutes now [20:43] thumper: sure [20:55] wallyworld, I am going to be late to the sts call [20:57] when you add a new facade verison is there anything else besides making sure you embed the previous, register the new version, and update facadeversions? === natefinch is now known as natefinch-afk [21:06] alexisb: me too, i just got back from school drop off [21:08] niedbalski: there now [21:08] natefinch-afk: sorry for the delay, yes it doesn't look at the path list: http://bazaar.launchpad.net/~juju-gui/juju-quickstart/trunk/view/head:/quickstart/platform_support.py#L117 [21:08] natefinch-afk: if you'd prefer it to look there next please file a bug and can see what can be done, but that's the way it works currently. [21:08] Bug #1331151 opened: 'juju destroy-environment' sometimes errors [21:08] Bug #1496143 opened: juju 1.24.5 no-proxy does not work as expected. [21:22] thumper: temp fix to unblock master http://reviews.vapour.ws/r/2669/ [21:23] wallyworld: hm.. have to ask the q: is that in the spirit of blocking master? [21:24] katco: ? [21:24] it's a temp fix to unblock [21:24] wallyworld: unblocking by skipping tests seems like it's just side-stepping the problem [21:24] katco: the functionality is dsiabled [21:24] wallyworld: when the purpose of blocking is so that we don't layer onto broken code [21:24] so the tests need to be too [21:24] with current master I am jsut getting this error over and over : machine-0: 2015-09-15 21:24:21 ERROR juju.worker runner.go:223 exited "toolsversionchecker": cannot update tools information: cannot get latest version: canot find available tools: no matching tools available [21:24] wouldn't it be better to have something like ignore- when pushing rather than doing this? [21:25] wallyworld: ah ok. makes sense [21:25] anyone seen that before? [21:25] katco: the functionality will be fixed today - a closed mgo session - this just buys time [21:25] wwitzel3: I've been getting it a lot, but I've just attributed it to "no upgrades available", but a bad error message [21:26] wwitzel3: sorry, haven't seen it. master's in a bad way atm though [21:29] wallyworld: shipit [21:30] ty [21:31] cmars: I'm going to have to skip today [21:47] thumper: a selection of bugs, there are going to be some dupes in here [21:48] mgz: ta [21:48] most of the test panics are one new form of panic on vivid/wily [21:57] Bug #1496159 opened: test suite runs fail to clean up all artifacts

[22:00] thumper, k [22:03] Bug #1496159 changed: test suite runs fail to clean up all artifacts

[22:05] katco: thumper wallyworld alexisb: Ci will pause for at most 1 hour when a voting test fails its reties. [22:08] sinzui: 1 hour is a long time :-) [22:08] wallyworld: we do need time to look at os-deployer which will trigger 1h. If the failure is in the first 2 hours of testing, there will be no delays at all [22:10] wallyworld: this is somewhat complicated, in practice most failures will not cause a delay to retesting [22:10] because it's based on time after failure, and if there are other tests ongoing, that time is used anyway [22:10] makes sense [22:12] Bug #1467712 changed: cmd/jujud/agent: data race in test [22:12] Bug #1467753 changed: cmd/jujud/agent: multiple data races detected [22:12] Bug #1496159 opened: test suite runs fail to clean up all artifacts

[22:12] Bug #1496161 opened: cmd/jujud/agent: data race detected [22:15] Bug #1496161 changed: cmd/jujud/agent: data race detected [22:15] Bug #1467712 opened: cmd/jujud/agent: data race in test [22:15] Bug #1467753 opened: cmd/jujud/agent: multiple data races detected [22:19] mgz: sinzui: one fix landed, one more to go, let's hope tmac bad rec stays away, fingers crossed [22:24] Bug #1467712 changed: cmd/jujud/agent: data race in test [22:24] Bug #1467753 changed: cmd/jujud/agent: multiple data races detected [22:24] Bug #1496161 opened: cmd/jujud/agent: data race detected [22:54] Bug #1496166 opened: juju ssh connects to incorrect units [23:54] Bug #1494947 changed: Panic keyManagerSuite.TestImportKeys on wily

[23:54] Bug #1494948 changed: Panic charmVersionSuite.TearDownTest on wily

[23:54] Bug #1494949 changed: Panic charmVersionSuite.TearDownTest on wily