[00:31] wallyworld, ping [00:32] hazmat: otp, one sec [00:41] time to wrap up warm and take the dog for a walk [00:43] thumper: still snowing ? [00:43] * hazmat wonders if the dog gets wrapped [00:45] http://www.nancysdogpawsboutique.com/images/products/Italian%20Greyhound%20English%20Style%20Fleece%20Dog%20Coat%20brown%20001.jpg [01:04] sinzui: is wolfe-01 you box ? [01:35] hazmat: no, not that cold, dog is fine [01:35] davecheney: not snowing, just cold [01:39] thumper: this one is for you, https://github.com/juju/loggo/pull/3 [01:39] * thumper looks [01:40] davecheney: what is that? [01:40] godoc badge [01:41] just a link to the godoc page for loggo [01:41] ah... [01:41] ok [01:53] waigani: are you working on https://bugs.launchpad.net/juju-core/+bug/1356899 ? [01:53] Bug #1356899: manual provider add-machine fails if no ubuntu user on machine [01:54] davecheney: the revert branch has merged [01:55] waigani: did it? I thought it was rejected [01:56] Pull request successfully merged and closed [01:56] You're all set—the waigani:revert-manual-prov branch can be safely deleted. [01:56] ok, good [01:56] davecheney: ^ from github [01:56] waigani: please assign yourself to the issue and mark it in progress [01:56] * thumper scrolls up [01:56] oh, there it goes [01:56] if it's fixed, mark it as fix-comitted to unblock the build [01:56] actually, mark it fix committed [01:57] done [01:57] waigani: are you going to re-propose without the deleted line? [01:57] thumper: yep [02:09] axw: finished [02:18] coffee needed [02:29] coffee finished [02:31] waigani, btw.. if you want a functional test case for that regression try... using juju digitalocean plugin [02:40] thumper: axw: https://github.com/juju/juju/pull/524 [02:41] hazmat: do DO machines not have the ubuntu user? [02:41] thumper, no ubuntu image thats not cpc cloud does [02:41] EPARSE [02:42] really? [02:42] so, since DO are not CPC, they have no ubuntu user? [02:42] why is that? [02:42] thumper, you've installed ubuntu desktop [02:42] what do they have instead? [02:42] they have a sane version of ubuntu server [02:42] they need some user right? [02:42] thumper, yeah... its called root [02:42] waigani: LGTM, but please test it live [02:43] really? [02:43] ew [02:43] axw: okay, will do now. [02:43] :-) [02:43] thumper, ew.. like a user who can admin stuff without typing sudo.. [02:43] yeah, yeah [02:43] scott/tiger [02:43] * hazmat bows to the oracle [02:44] * thumper missed that reference [02:44] thumper, default/doc'd/example'd password on oracle [02:44] thumper: http://xforce.iss.net/xforce/xfdb/970 [02:44] ?! [02:45] heh [03:06] davecheney: do you have time to cast your eyes over this again? https://github.com/juju/juju/pull/508 [03:10] * davecheney looks [03:21] axw: $ juju add-machine ssh:scott@10.0.3.244 [03:21] scott@10.0.3.244's password: [03:21] [sudo] password for scott: [03:21] install: cannot create directory /home/ubuntu: Permission denied [03:21] ERROR subprocess encountered error code 1 (Connection to 10.0.3.244 closed.) [03:21] axw: scott is in the sudoers file [03:26] axw: I just ran the same with trunk, I get same result [03:59] waigani: sorry was afk having lunch [03:59] axw: np, I've commented on PR [03:59] waigani: can you confirm you can "sudo mkdir /home/ubuntu" on the machine when logged in as scott? [03:59] axw: yep I could [04:00] axw: but it asks for scott's password [04:01] waigani: add-machine should prompt you for that, is it not doing so? [04:01] axw: yep it did [04:02] axw: same behaviour on trunk btw [04:03] waigani: I'll try to repro, sounds odd [04:04] axw: that would be great, thanks [04:11] waigani: I cannot reproduce on master. did you remove /home/ubuntu? userdel does not do that by default [04:13] Here's a patch that fixes the blocking bug if anyone has a minute: https://github.com/juju/juju/pull/525 [04:15] mgz_: you have a minute? [04:15] ericsnow: done [04:16] thumper: thanks [04:37] wallyworld: ping [04:38] ericsnow: hey [04:38] you have a minute to talk about reviewboard [04:39] ericsnow: sure, hangout? [04:39] sure [04:40] ericsnow: https://plus.google.com/hangouts/_/canonical.com/tanzanite-stand [04:50] axw: does it work for you? [04:50] axw> waigani: I cannot reproduce on master. did you remove /home/ubuntu? userdel does not do that by default [04:50] axw: yes [04:50] strange [04:50] *shrug*, just worked for me [04:51] please look into what's going on there, because it doesn't sound right [04:51] yep [04:52] axw: just to confirm, 'scott' just needs to be in the sudoer's file [04:53] waigani: well, I just s/ubuntu/scott/ in the juju file in /etc/sudoers.d [04:54] axw: did you delete or rename home dir? [04:54] waigani: I deleted /home/ubuntu [04:54] axw: I did sudo adduser scott sudo [04:56] waigani: I'll try again with that method. [04:57] axw: and I'll try swith your [04:58] axw: ah [04:59] axw: it's still ubuntu in sudoers.d [04:59] waigani: ? [04:59] waigani: it should be. if you just did "adduser scott sudo" that's just adding it to the group [05:00] axw: right, I thought that was enough [05:00] waigani: what that doesn't do is enable passwordless login [05:00] but if you have a password that should be fine [05:02] axw: scott has a password and I gave it to juju add-machine [05:04] waigani: works for me using that approach too. [05:04] waigani: my /home is 0755 [05:04] axw: and *not* changing /etc/sudoers.d ? [05:05] waigani: correct, I just did "sudo adduser scott sudo" and used the password [05:05] hmph [05:25] axw: I can't reproduce it now, everything seems to be working as expected. [05:25] hrm ok. [05:26] well, I say land it. CI and/or hazmat will complain if it's still broken ;) [05:26] okay, I'll jfdi it [06:14] morning all [06:19] morning voidspace [06:19] o/ [07:16] morning [07:18] TheMue: morning [07:55] Reading the pain points spreadsheet, and I was thinking a manifest file in the charm, listing one or more URIs and the path to install or unpack them, could solve several items. It that the sort of thing already being talked about? [08:08] morning TheMue [08:09] dimitern: heya, seen my mail? [08:14] TheMue, yep, I haven't had time to reply yet, but I'll get back to you soon (I have a call with Mark S in 15m) [08:15] dimitern: ok, np, and thanks already [09:26] Hmmmpf, shitty net, currently it dislikes me. [10:03] * fwereade_ is on a public holiday and not working [10:03] * fwereade_ would still like to express his frustration at not being able to express a `chan foo` as a `chan interface{}` [10:27] TheMue, I think we need a branch to expose all capabilities on the apiserver/agent facade [10:27] dimitern: I found a way [10:27] TheMue, including the new one; no need to expose the environ config, just the part of the interface defined as state.EnvironCapabilities [10:28] dimitern: I’ll paste the part, one moment [10:28] TheMue, the provisioner and upgrader api facades have access to the environ config, but that's a hacky approach [10:29] dimitern: http://paste.ubuntu.com/8052638/ [10:30] dimitern: I’m using the config too [10:31] TheMue, ah, this looks nice [10:31] TheMue, cheers [10:32] dimitern: so right now the providers only return constants, now the second part letting them return real values [10:33] TheMue, what constants? [10:33] dimitern: as I understand the local bootstrap as well as manual and manually provisioned machines on other providers need the safe networker [10:34] TheMue, yes, all the other cases (maas, ec2, etc.) will use the normal networker [10:34] dimitern: true, but nothing I’ll push before the change descibed above. only wanted to see it compiling [10:35] dimitern: do manually provisioned machines exist on maas too? today this isn’t covered [10:36] TheMue, to make it reusable, I'd add a method to the agent that returns state.EnvironCapability, and internally uses the st.Environment().EnvironConfig() + config.New, etc. [10:36] TheMue, manually provisioned machines can exist regardless of the provider the environment uses [10:36] dimitern: ok, sounds good [10:37] TheMue, cheers [10:47] voidspac_, standup? [10:47] dimitern: sorry! [10:47] dimitern: TheMue: omw [11:15] mgz_, hey [11:16] hey [11:16] mgz_, so you're switching to the juju ci team? [11:16] yup [11:16] mgz_, congrats! :) [11:16] mgz_, that was kinda sneaky hehe [12:21] good morning, team :) [12:23] hey katco [12:23] congratulations on the move, mgz_ :) [12:44] katco: hey, i'm back from soccer if you wanted to chat [12:44] wallyworld: you're pretty much up to speed w/ what i'm doing. but we can definitely chat if you'd like :) [12:45] katco: only if you need to, happy to let you keep going... :-) [12:45] wallyworld: $$greatful it landed$$ [12:45] err *grateful [12:45] :-) [12:45] i do have one question: did you win? ;) [12:49] katco: nope, our stikers kept missing the net :-( [12:49] aw bummer [12:49] what position do you play? [12:49] left right out :-) [12:49] centre back [12:49] ah! that was my old position :) sweeper/left fullback [12:50] <3 defense [12:50] i like that position, less running [12:50] haha [12:50] plus i read the game well, so it suits me [12:50] yeah it always struck me as an analytical position [12:51] * wallyworld wonders if fwereade_ is around [12:52] wallyworld: public holiday, but has been around this morning once [12:53] TheMue: tanks [12:53] thanks [12:54] yw === ChanServ changed the topic of #juju-dev to: https://juju.ubuntu.com | On-call reviewer: see calendar | Open critical bugs: None [13:14] fwereade_, I'm working on that api to get a list of metrics if you fancy doing some bikeshedding with me ;) [13:19] who are the eco folks? [13:32] mgz_: ping [13:40] ericsnow: hey [13:41] mgz_: just following up on my reviewboard email :) [13:41] yeah, 's in my queue of many emails to respond to... [13:42] mgz_: no worries (wanted to be sure it didn't get lost in the ether) [13:46] ericsnow: short answer is past the charm and some minor changes, most of the work is in just deploying the thing [13:46] mgz_: awesome! [13:46] so, ther's not *much* to pass over, one thing to sort out is where to host it [13:47] mgz_: I was considering pinging Curtis about that (after I'd gotten info from you) === hatch__ is now known as hatch [17:01] ericsnow, on and ready when you are [17:02] alexisb: brt [17:38] alexisb: hey! [18:45] hi katco ericsnow do either you have a minute to review https://github.com/juju/juju/pull/526 [18:45] * katco looking [18:45] sinzui: sure [18:49] sinzui: looks benign to me, although i'm not sure exactly what [18:49] hangs off of version.Version [18:51] sinzui, katco: yeah, same here [18:52] ericsnow, katco Both version changes define the internal version juju or the installer reports. It is also the version that packaging will adopt [18:52] I make this change after ever release, though since I haven't really released, I am a little disgruntled I need to make this change [18:54] sinzui: why don't we use some tagging mechanism in our versions (e.g. 1.20.0-a1, 1.20.0-rc3)? [18:54] ericsnow: +1 [18:55] don't we do that? we have like 1.20.0-alpha [18:55] ericsnow: i mean presumably versions are free, but there's a cognitive overhead people have when you jump from 1.20.2 to 1.20.4 or 1.20.5 [18:56] natefinch: not if version/version.go has anything to say about it [18:57] natefinch, we cannot go back in time to tell 1.18 not to panic when it sees an alpha in the version [18:57] sinzui: yuck [18:58] sinzui: oh, I see... so for 1.20.x we can't use our fancy new system for alphas, but for the release after, we can, correct? [18:58] sinzui: so if we wanted to make that change, we need a major point release? [18:59] natefinch, yep, all this pain of internal testing wouldn't be happening if we were already on 1.21 [18:59] sinzui: can we just call the next release 1.21? :) [19:00] sinzui: ah, it's just a 1.20 issue [19:00] natefinch, well, we could all agree to abandon 1.20.x dev becomes 1.22-alpha1 and stable becomes 1.21-rc1...because that is how we are actually treating these branches/releases [19:01] sinzui: version numbers aren't important to developers... but I presume there's someone else who would raise a stink if we did that [19:02] natefinch, they are important to debian/ubuntu. Since 1.20.x is not in either, I don't think anyone would notice [19:03] * sinzui is not sure how to announce a major version that doesn't offer new features [19:04] sinzui: "hey guys, this makes my life 100x easier and will mean we get releases out faster" ... I think everyone will understand. [19:04] sinzui: it supports v4 AWS signing i.e. China region :) [19:04] there, see? [19:04] not that i have enough information to have a horse in this race, but there's that [19:05] _although_ we did not have a china instance to test against [19:05] so i would hate to call that out as a major value-add and then have it not work (even though all signs say that it will) [19:05] sinzui: the new feature is that dev revisions are now -alpha :) [19:05] haha [19:06] natefinch, the dev features really are alpha. Given the problems getting them to pass, I think we can say we are a month away from a beta [19:07] i even have plans to release an alpha until we have sorted out the utopic stable release [19:13] sinzui: hey is there a doc that describes our release strategy? versions, process, etc.? [19:13] katco, not versions since we just switched the process. [19:14] sinzui: ah ok. i'd love to grok that at some point (like i don't already have enough to learn) :p [19:15] katco, CI tests the release process so this doc and all the test scripts document what must happen to be a blessed https://docs.google.com/a/canonical.com/document/d/1o0au1ZNPpA8sGBNj2yNgOV3PH7Oxpyqw3hQNuIFaS-w/edit [19:15] katco, This template illustrates the automated and manual steps of a release https://docs.google.com/a/canonical.com/document/d/1aPOHa_JNKjKd5lAcOlFahjbNl8V_KgtMx6rgYdC2f7k/edit [19:16] sinzui: cool, thank you! [19:16] katco, this is the log of the 1.20.1 release which is a little easier to read https://docs.google.com/a/canonical.com/document/d/1BOWKU4Iqy363pp4w8IWBwabDT2TdA9fb4futQVjtuVQ/edit [19:17] katco: odd minor versions used to be dev releases, even minor versions were stable. But luckily we realized that was dumb and non-obvious, so we switched to "anything N.N.N is stable, anything N.N.N- is a dev release" [19:17] natefinch: ah ok [19:18] katco: and stable releases only upgraded to new stable releases, and dev->dev. so 1.20 is the cross-over, where it knows about the new naming scheme, but 1.18. which wants to upgrade to 1.20, doesn't. [19:19] natefinch: gotcha. thus we can't fiddle with versions in 1.20 b/c 1.18 wouldn't know what to do? [19:20] katco: correct [19:20] makes sense [19:25] TheMue, axw https://github.com/juju/juju/pull/527 PTAL :) [20:36] fun friday fact: i just realized i have 3 8 core machines at my disposal here at home. that's quite nice :) [20:37] i wonder if i could set up my own juju ci to continually run tests as i commit so it's not a big bang at the end :p