[01:08] davecheney: you around? [02:00] thumper: sorry i missed you [02:00] i thought the 1:1 was tuesday [02:00] ah :) [02:00] just last week [02:01] I have some time in an hour if that works for you [02:06] thumper: i don't have anythig to talk about [02:06] tomorrow is my last day [02:06] then i'm taking a few personal days [02:06] then off to vienna [02:06] see you in nuremberg [02:07] um... ok [02:10] or we can 1:1 [02:10] if you're free [02:10] if there is stuff you have to talk about [03:25] wallyworld_: hi :D [03:29] hi [03:33] jam: ping for when you are around [03:38] wallyworld_: u keep dorpping from canonical ... [03:38] wallyworld_: 1.23 and master forwards r up for review [03:38] wallyworld_: also map-ordering milestone is 1.24 which is currnt trunk so no backporting is required :D [03:40] ok, ta looking [04:20] jam: reping [04:31] hey thumper [04:31] jam: just doing expense paperwork [04:31] jam: hangout in 5 minutes? [04:32] thumper: as you'd like, what about? [04:32] container stuff [04:35] jam https://plus.google.com/hangouts/_/canonical.com/containers === JoshStrobl is now known as JoshStrobl-AFK === lp|sprint is now known as lazyPower [07:00] morning everyone [07:03] mattyw, o/ [07:05] hey dimitern, want to meet up a bit early? [07:05] jam, hey - yeah, that works for me [07:06] dimitern: k, I'm there [07:06] jam, omw [07:37] morning o/ ≋ (windy) [07:38] TheMue, morning [08:22] wallyworld: ping [08:31] voidspace, dooferlad, morning [08:31] dimitern o/ [08:31] dooferlad, I have a PR for you to have a look - https://github.com/juju/juju/pull/1976/ [08:31] dimitern: looking [08:32] dooferlad, in retrospect, I should've done that a week ago, but there it is [08:33] dimitern: If you had, I would probably have done a lot less digging into the Juju code. It is all an education. [08:33] dooferlad, that's my consolation :) [08:33] dimitern: :-) [08:40] dimitern: morning [08:41] voidspace, nice branch btw [08:42] voidspace, i'll approve it shortly I think [08:42] dimitern: fixing the provisioner issue? cool. [08:42] dimitern: wallyworld reviewed it, but he missed the test I think. [08:42] voidspace, the maas thing yeah [08:42] dimitern: I have a trunk branch as well, but I haven't submitted it [08:42] voidspace, was it a diff mismatch on RB or it shows that test? [08:42] dimitern: in case I need to make changes, when the first one lands I'll submit the second [08:43] dimitern: it shows the test, but it's in container_test.go and wallyworld was expecting something in provisioner_test.go [08:43] voidspace, ah, right [08:44] dimitern: I'm pretty sure that's it anyway, he's not around to ask :-) [08:52] voidspace: hi, sorry as having dinner [08:52] wallyworld: you're allowed to have dinner... :-) [08:52] voidspace: sorry i missed the test, i'll take another loko [08:52] wallyworld: np [08:53] voidspace: actually, i gave you a shipit so go for it, was more of a question seeking clarification [08:53] wallyworld: ah [08:53] voidspace, reviewed [08:54] wallyworld: well, it was a "Fix it, then ship it!"" with an open issue. I wanted to check with you before just dropping the issue. [08:54] voidspace: ah i see what you mean, thanks for asking [08:54] dimitern: thanks [08:54] wallyworld: I have some minor issues from dimitern to fix anyway... [08:54] :-) [09:02] TheMue, standup? [09:34] dimitern: [09:34] ... value *errors.Err = &errors.Err{message:"", cause:(*errors.Err)(0xc2100790f0), previous:(*errors.Err)(0xc2100a50f0), file:"github.com/juju/juju/leadership/leadership.go", line:43} ("leadership claim denied") [09:34] dimitern: util_test.go:1406: [09:35] ocr: is anyone available for reviewing http://reviews.vapour.ws/r/1324/ ? thank you! [10:07] dooferlad, my branch landed btw [10:07] dimitern: sweet. [10:07] dooferlad, I'm starting on CLI for subnets following the same approach [10:37] natefinch, morning [10:37] natefinch, judging by twitter I'd say you've been watching the news/ reading a paper this morning ;) [10:37] mattyw: been watching the news for a while, just finally figured out how to put my views into 140 characters :) [10:58] * TheMue is out for lunch, bbiab [10:58] dimitern: btw, found a nice way to simulate the error, it's pretty simple [11:00] TheMue, yeah? let me know when you're back please === JoshStrobl-AFK is now known as JoshStrobl [12:25] dimitern: sure, "select id, hostname from maasserver_node;" resolves the names, then you do a "select id, name from metadataserver_noderesult where node_id = ..." to find the right gathered datas. [12:25] dimitern: interesting is the one named 00-maas-01-lshw.out [12:25] TheMue, yes, indeed [12:25] TheMue, so you managed to reproduce the bug and the error? [12:26] dimitern: having its ids allows you to do a "update metadataserver_noderesult set data = '' where id = 43;" [12:26] dimitern: in a first round the bootstrap funnily wanted exactly my corrupted node, yes, with a matching error like in lp [12:27] dimitern: I then acquired the node to have a working bootstrap and then released it again to simulate the error during deploying [12:28] dimitern: this currently hangs, so I now want to take a deeper look into the logs and await to find the same error there [12:28] TheMue, btw you can specify which node to pick at bootstrap time with --to node.maas [12:29] dimitern: ah, makes the next round a bit more simple [12:29] TheMue, yeah :) [12:30] dimitern: but so it is fine too, with currently only four nodes here you quickly get the right (corrupted) one [12:30] TheMue, so if the error is reproduced, you should see it just before bootstrap outputs Lauching instance XYZ [12:30] TheMue, and the bootstrap should fail [12:31] TheMue, if it keeps going after the instance is started, then it's not reproduced [12:31] dimitern: yeah, I had the same output like in the lp issue [12:31] dimitern: same error message and bootstrapping failing [12:33] TheMue, nice! [12:34] yeah [12:46] bug 1437038 [12:46] Bug #1437038: 1.23b2 fails to get IP from MAAS for containers, falls back to lxcbr0

[12:48] Bug #1438168 was opened: juju 1.23 doesn't release IP addresses for containers [12:56] * dimitern steps out for a while [14:03] perrito666: if you can tweet you can get on the hangout ;) [14:17] * dimitern is back [14:32] TheMue, thanks for updating bug 1427814, it's really useful to have the info about how to tweak lshw data in the maas db [14:32] Bug #1427814: juju bootstrap fails on maas with when the node has empty lshw output from commissioning

[14:32] dimitern: yw [14:48] cherylj: would you mind updating #1435974 with where things are at with the LICENCE fixes? [14:48] Bug #1435974: Copyright information is not available for some files

[15:03] ericsnow: sure. [15:03] cherylj: thanks :) [15:04] ericsnow: I do have questions about a couple repos. [15:05] I'm guessing we don't need to modify the gojson* repos. [15:05] and the same for syslog [15:06] cherylj: right [15:07] cherylj: (since they are slight adaptations of the Go stdlib code and retain that copyright/license) [15:07] k [15:07] I just need to do the charm branches, and I think I'll be done. [15:07] then on to updating dependencies.tsv [15:08] dimitern, dooferlad: FWIW, "juju space" is pretty ambiguous; any chance we could name it something more clearly related to networking? [15:08] cherylj: awesome! thanks for slogging through all that :) [15:09] ericsnow: np! Although I did run into that unrelated test failure when trying to merge juju/testing 1.22 [15:09] thumper just manually merged it last time [15:09] dimitern: can we add a new upgrade step in a bugfix release? [15:10] ericsnow: That is a discussion which has started several times and is currently on hold. I expect it will come up at the sprint. [15:10] dimitern: as adding the addresser would require upgrading IP addresses to add a Life field [15:10] dimitern: that seems unfortunate [15:10] dooferlad: cool, thanks [15:10] dimitern: I have a bunch of patches but I'm not sure what to do about the upgrade step [15:12] sorry guys - in a call [15:12] dimitern: np [15:13] ericsnow, natefinch: since the license is changing for charm, do I need to update all the branches? Or just 1.22, v4 and v5? [15:15] Bug #1438258 was opened: error while destroying incomplete environment [15:15] cherylj: I'd say consider the branches that will be related to current or future juju releases [15:17] cherylj: don't update old branches (other than 1.22 & v4) [15:17] ericsnow, natefinch: Thanks! [15:17] ericsnow: natefinch: if I want to prepare a binary for someone to try out, know how I should do it? [15:17] we have a script to do source releases but not binaries [15:17] I just need an amd64 build [15:18] can I just tart up juju* from $GOPATH/bin [15:18] voidspace: yeah, you just need juju and jujud [15:18] natefinch: cool, thanks [15:18] voidspace: just "go install ./juju/..." and send them the file out of $GOPATH/bin, right? [15:19] ericsnow: s/file/files/ but yeah - that's what I thought and natefinch seems to be confirming [15:19] voidspace: yep :) [15:19] voidspace: BTW, did you see my first-draft poster yet? :) [15:19] yep... nice thing about go. It's just copy and paste the binaries. for us, we just need the client and the server [15:20] natefinch, voidspace: don't the tools also need to get copied (so they can be uploaded) [15:20] natefinch, voidspace: jujuc, etc. [15:20] ericsnow: natefinch: jujud is "the tools" [15:20] ah, maybe [15:21] voidspace: yeah, the hook tools like set-relation [15:21] jujud is the tools, the upload just symlinks jujud to jujuc [15:21] er vice versa [15:21] natefinch: ah [15:21] ericsnow: all that stuff, symlinks [15:21] natefinch: how funny [15:23] okay, I need reviews for the charm license changes: https://github.com/juju/charm/pull/107 [15:23] https://github.com/juju/charm/pull/108 [15:23] https://github.com/juju/charm/pull/109 [15:23] jam: thanks (for the calender access) :) [15:28] ericsnow: Here's the update list of repos: https://bugs.launchpad.net/juju-core/+bug/1435974/comments/11 [15:28] Bug #1435974: Copyright information is not available for some files

[15:28] Are there any I missed? [15:33] voidspace, hey [15:33] voidspace, so what's the issue with the upgrade step? [15:37] cherylj: LGTM, thanks! [15:37] yay! thanks, ericsnow :) [15:38] dimitern: 1.23 has IP addresses in it [15:38] voidspace, yeah? [15:38] dimitern: so if we backport "release IP addresses" to 1.23, we'll need an upgrade step from beta2 to beta3 [15:38] dimitern: adding the Life field to IP addresses [15:38] dimitern: is that ok? [15:40] voidspace, can we do that? [15:41] voidspace, aren't the steps based on major/minor versions? [15:41] dimitern: well exactly [15:41] dimitern: I don't think we *can* do it [15:42] dimitern: that's exactly what I'm asking though :-) [15:42] voidspace, well [15:42] dimitern: the other way is to backport the *original* approach [15:42] voidspace, I think you could still try to run the step - it shouldn't do any harm if Life is already there [15:43] dimitern: do upgrade steps run at all? I'd have to look through the code to work it out [15:43] dimitern: the original approach was with a Provisioner API instead of a worker [15:43] and no Life field [15:43] I still have that code [15:43] voidspace, hmm [15:43] voidspace, ok, let's think for a minute [15:44] voidspace, what's the worst that can happen if we omit this from 1.23? [15:44] it looks to me like upgrade steps for a version are run / re-run when you upgrade within a version [15:44] voidspace, addresses won't be released [15:44] i.e. 1.23.1 to 1.23.2 will run 1.23 upgrade steps I believe [15:44] voidspace, ah - then it might work? [15:44] dimitern: if I'm right [15:45] PerformUpgrade and runUpgradeSteps in upgrades/upgrade.go [15:45] Bug #1438274 was opened: unit test failures: storeManagerStateSuite.TestStateWatcher

[15:45] voidspace, it seems like a quick experiment is needed to verify this [15:46] yeah [15:46] dimitern: I'll define a new upgrade step that dies horribly and see if it runs :-) [15:46] voidspace, e.g. start from the earliest possible version that has IP addresses collection, which is actually used [15:47] voidspace, which should be the 1.23-beta1 right? [15:47] sounds right [15:48] dimitern: but the upgrade step is only on trunk - so I'll either need to do the full port and test that [15:48] dimitern: or just take trunk and change the version back to 1.23-beta3 or whatever [15:50] voidspace, the latter seems easier [15:51] dimitern: that's what I'm doing [15:51] dimitern: just bootstrapping 1.23 [15:51] voidspace, yeah and deploy some lxc-s [15:51] yep [15:51] dimitern: hmm, there's still an issue. This will become a 1.22 upgrade step instead of 1.23 [15:52] dimitern: so the upgrade step needs to check if the ipaddress collection exists, and bail without error if it doesn't [15:52] and it needs to run if upgrading from 1.22 -> 1.24 [15:52] voidspace, considering the steps are idempotent, it shouldn't be a problem to have the same step for 1.22 and 1.23 I guess [15:53] dimitern: yep, true - that solves the problem of 1.23 beta 1 -> 1.24 [15:53] voidspace, sweet! [15:53] and 1.22 -> 1.24 I think [15:53] in fact 1.23 upgrade steps must run then anyway [15:53] all intermediate steps must run [15:54] exactly [15:54] deploying lxc [16:18] ericsnow, natefinch: Can I get a review on the last license changes? https://github.com/juju/charm/pull/107 [16:18] https://github.com/juju/charm/pull/108 [16:18] https://github.com/juju/charm/pull/109 [16:19] and the utils changes: http://reviews.vapour.ws/r/1329/ http://reviews.vapour.ws/r/1328/ [16:25] dimitern: it looks like 1.22 steps don't get run when going from 1.23beta3 to 1.23beta4 [16:25] which is odd [16:25] I'll try again [16:25] I'm only seeing steps121 in the logs though [16:26] voidspace, I wouldn't expect 1.22 steps to run for 1.23-beta3 -> 1.23-beta4 [16:26] voidspace, but the 1.23 should [16:27] dimitern: well - I'm seeing 1.21 ones run - but neither 1.22 nor 1.23... [16:28] voidspace, how did you prepare your scenario? [16:28] dimitern: http://pastebin.ubuntu.com/10707700/ [16:28] dimitern: juju bootstrap --upload-tools (branch 1.23) [16:28] dimitern: then "juju add-machine lxc:0" and wait for instance id [16:29] dimitern: then checkout master and do the godeps / go install dance [16:29] well, after changing version/version.go to 1.23-beta3 (should have picked beta4 but it shouldn't matter) [16:29] voidspace, and before bootstrap you've rebuilt the juju binaries, right? [16:29] then "juju upgrade-juju --upload" [16:29] dimitern: yep [16:30] then "juju upgrade-juju --upload-tools" [16:30] I mean [16:30] dimitern: http://pastebin.ubuntu.com/10707700/ [16:30] I will try again, with a better new version number and maybe some extra logging [16:30] but first - coffee [16:31] ok [16:35] time to go [19:53] mramm: ping [22:36] +