=== kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === urulama is now known as urulama|lunch [09:28] dimitern, ping [09:28] dimitern, did you see my two branches? ericsnow reviewed them, so I'll try and land them later. [09:28] dimitern, they're only short, so you may, or may not, want to look at them [10:06] dimitern: thanks! [10:16] hey, could a couple of follow-ups on http://reviews.vapour.ws/r/1218, and, if you guys are really in the review spirit: http://reviews.vapour.ws/r/1330/ === urulama|lunch is now known as urulama [10:49] ocr, could you please take a look at http://reviews.vapour.ws/r/1383/ ? thanks! === lp|away is now known as lazyPower [13:30] ericsnow: Can you ping me when you get a minute so we can talk about bug 1439447? [13:30] Bug #1439447: tools download in cloud-init should not go through http[s]_proxy

[14:02] cherylj: OTP === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [14:40] cherylj: hangout? [14:41] ericsnow: sure, let me get my headset on [15:18] aznashwan, are those the centos support patches? [15:32] Bug #1440097 was opened: unittest failure: apiserver/root_test.go:rootSuite.TestPingTimeout

[15:32] just so that folks know all of our scheduled all call reviewers are suppose to be on vacation today [15:33] woot - free for all [15:35] or free for none ;) [15:44] Bug #1440097 changed: unittest failure: apiserver/root_test.go:rootSuite.TestPingTimeout

[15:50] Bug #1440097 was opened: unittest failure: apiserver/root_test.go:rootSuite.TestPingTimeout

[15:54] natefinch: such a downer ;) [15:54] jw4: I try :) [16:01] jw4: hey, wanna rewrite the project in python? [16:01] jw4: i bet we can do it before anyone gets back [16:04] katco: you tested the original v4 code against s3 right? [16:04] I think the rest of aws just does it differently [16:04] mgz: yes, live tests and all [16:05] mgz: except it's passing the amazon test suite just fine [16:06] the region name is different as you pointed out yesterday, and some other stuff [16:06] their test suite is apparently just not as tight as their real auth [16:07] hehe possibly [16:07] I'm looking at boto/auth.py HmacAuthV4Handler vs S3HmacAuthV4Handler [16:10] mgz: wwitzel3 said he had worked on boto quite a bit, he's peeking at the issue as well. what are the main diffs you're seeing? [16:12] it's slightly annoying as the classes are written differently, [16:12] but the headers that get signed is different, the region format, and some of the sequencing [16:13] S3 signs all headers, general only signs Host and X-Amz-* [16:13] katco: lol [16:13] if the headers are present, you must sign them [16:13] katco: we might even get hazmat back if we do [16:13] jw4: hehe [16:14] katco: for s3, everything but the Auth itself, yeah, but not in general [16:14] alexisb: sorry, got a little distracted with work [16:14] alexisb: yes, those are both the CentOS patches [16:14] katco: i'm listening ;-) [16:14] mgz: no, i mean the algorithm requires that. you can't optionally sign headers; i don't think that's a mutable property of the algorithm based on which service [16:14] hazmat: haha o/ :) [16:15] aznashwan, ok, I have had a few folks comment re those patches, I am thinking we may need to do some pair reviewing when we are all in germany [16:15] S3 also doesn't do url path normalisation, whereas the other does === kadams54 is now known as kadams54-away [16:16] mgz: i don't think we're doing that at all [16:17] which can me something as dumb as trailing slash would break it maybe [16:17] alexisb: we have slowly come to terms with the fact that those PR's won't get in until the sprint; but I still like to hope beyond hope :D [16:17] mgz: yeah [16:18] like, our failed one was going a GET with a path [16:19] the one that worked was post to root and form-encode body [16:19] so it could be something as dumb as bad path [16:19] aznashwan, the alternative is to have you break everything down as requested [16:19] katco: but you probably are going to need two versions to keep s3 working at the same time as fixing this [16:23] alexisb: as mentioned, the refactor, added CentOS userdata and package management changes are so inter-dependent that breaking them up would take a tone of re-work, writing temporary code which only gets overwritten in the following PR, and triple/quadruple the pain of maintaining the PR's rebased unto master [16:24] awesome, my robot vacuum cleaner just unplugged my computer [16:25] IoT ftw [16:26] ehehe [16:26] rise of the downtrodden robots! [16:26] i definitely kicked it [16:28] alexisb: that would add an overall delay of at least a month whilst the PR as is now is merge-able with the exception of a handful of tests which are still not CentOS-ready [16:28] this is when you discover the temporary work you were doing in /tmp is less temporary than you would have liked. [16:28] (...which we are working on as we speak) [16:29] gah, stupid f'ing machiner... I don't know how the hell you're supposed to use this thing. It hides everything under too many layers of abstraction. [16:30] aznashwan, then I suggest we plan to have a dynamic review session in person [16:32] alexisb: would not be too bad if it comes to that. I would insist to still get a review here-and-there in the mean time so we get the small stuff ironed out before the sprint so we focus on the bigger picture there. [17:04] mgz: i think i found the problem [17:05] mgz: i vaguely recall removing default paths "/" from the canonical request string for S3 b/c that was causing an issue [17:05] mgz: it looks like adding that in gets me a result [17:06] katco: ha, was paths :) [17:06] mgz: i tip my hat to you, sir [17:06] :D [17:06] mgz: and i then slap aws's collective faces with it. [17:06] mgz: what an undocumented mess [17:09] mgz: https://github.com/go-amz/amz/commit/3431a6c7dea296878e5efa2681e85e0b1a2f5a27#diff-5f72c79adc6ba80e9d4d6cf78433008eL151 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === lazyPower is now known as lp|away [18:08] mgz: feel like a review? https://github.com/go-amz/amz/pull/47 [18:08] katco: surething [18:11] ah foo i broke a signing test. those things are so fragile [18:11] yeah, was wondering if more tests would need changing [18:11] nah it's just a space [18:12] so if it works with s3 and ec2, I think this is basically good to go [18:12] this isn't super-robust, callers can still pass in stuff that will mess this up I'm pretty sure [18:13] mgz: i think we're still broke on relative paths "/../.." or something [18:13] and I don't like the date-magic [18:13] date magic? [18:13] katco: yeah, and probably double slashes as well [18:13] yeah those tests fail [18:13] the pass in a x-amz-timestamp in nay random format and get it back out the right way code [18:14] I'd prefer we just demanded a sensible time object in the interface or something [18:14] well, smallest change possible. [18:14] but that's not something that really needs changing right now [18:14] right [18:14] right, I agree [18:14] repushed with test fix [18:14] just added some spaces [18:16] katco: lgtm [18:17] mgz: do we have this behind a bot yet? [18:19] katco: it's got a funny pre-merge check thing instead, which passed [18:21] mgz: ok, so merge manually? [18:22] yeah, oh, travis runs against both go 1.2 and go 1.4, interesting [18:22] only the unit tests obviously but fine [18:43] natefinch: I heard you had some thoughts about bug 1439447. Don't know if you wanted to weigh in before I make any code changes... [18:43] Bug #1439447: tools download in cloud-init should not go through http[s]_proxy

[18:53] mgz: so i'm trying to test this in 1.23 [18:54] mgz: and when i bootstrap i get a bootstrap failure [18:54] cherylj: my thought was basically that we should have a configuration that lets the user decide whether juju uses the proxy to connect to other machines in the environment or not. [18:54] katco: as in, the machine doesn't come up? [18:54] mgz: "ERROR failed to bootstrap environment: Juju cannot bootstrap because no tools are available for your environment." [18:54] katco: with --upload-tools? [18:54] cherylj: since it seems to me that it's likely that no matter which way we decide to do it, someone's going to want it to work the other way [18:54] natefinch: ha, too true [18:55] mgz: i knew it was something dumb [18:56] cherylj: looking forward to charming at the sprint. it would be nice if i actually got to work WITH juju more, and not just ON it :) [18:56] katco: me too! It should be fun and I think alexisb is working on some actual prizes for the teams :D [18:58] katco: ...that's just begging for a cheesy joke [18:58] why do all the things we work with have silly names [18:58] mgz: please proceed :) [18:59] well, you're clearly charming already? [18:59] LOLOLOLOL [18:59] 10 points england. [19:01] mgz: (et. al.) http://reviews.vapour.ws/r/1386/ [19:02] mgz: and it would be super awesome if you could try china again [19:03] 1 line change btw [19:03] hi, is safe to deploy block-storage-broker to machine0?? [19:03] katco: lgtm, how to the multiple branches with go-amz work? change also needs to land on v4-unstable, or does v3 get merged up periodically? [19:04] create one instance just to manage credential is overkilling [19:04] mgz: i think dimiter merges changes to v3 up to v4 occasionally [19:04] katco: I'll merge that locally and see [19:04] mgz: i don't know why we branch an unstable before we have unstable changes actually [19:04] mgz: seems like it creates a lot of work [19:04] discourage people from landing features on an active verion I'm guessing [19:04] er... s/active/stable/ [19:06] redelmann: that may be a better question for #juju - but gernally feel free to smoosh till something actually breaks [19:06] mgz, thank!! [19:07] mgz: and trunk: http://reviews.vapour.ws/r/1388/ [19:14] whoops, used wrong key [19:14] well, it bootstrapped till the ssh step [19:15] :) [19:27] mgz: can i get a rq lgtm on http://reviews.vapour.ws/r/1388/? [19:27] I was trying to make sure it works first, but this ssh step hates me ;_; [19:27] oh, no worries then [19:27] take your time [19:28] I can ssh into the address fine, but juju refuses to for some reason [19:28] weird... [19:28] but prints stuff out the first time I do on another console? bootstrap is too darn complicated these days [19:29] these days? :) [19:30] get off my lawn! [19:30] haha [19:30] okay, now it's running [19:30] haha [19:30] I have no idea what was happening there [19:31] katco: so, basically, api queries do now work [19:31] and my china bootstrap is happily installing packages currently [19:31] o/ <--- tiny lighter you can't see [19:31] haha... I spent all day trying to debug an api connection problem and somehow forgot to look at the log for machine 0..... where it was evidently filling up with panics :/ [19:31] stamped the trunk dep update too [19:32] joy [19:32] mgz: so to sum things up [19:32] mgz: i finally accomplished the first task i was given when i started canonical. [19:32] mgz: and it only took me about a year. :) [19:33] https://pastebin.canonical.com/128953 [19:34] woohoo! [19:34] that is great to see! :D [19:35] hey, getting it done is getting it done :) [19:40] katco: when the 1.23 branch goes through, I'll take the deb and set up a cloud-health job for the china region, hopefully later tonight [19:40] mgz: :D === kadams54 is now known as kadams54-away [19:41] natefinch: juju running on az-cn is something we can talk about publically right (slightly late, wtv) [19:42] dhah [19:42] ¯\_(ツ)_/¯ [19:43] honestly no clue === tvansteenburgh1 is now known as tvansteenburgh === kadams54-away is now known as kadams54 [19:53] all, I have to go pick up kiddo realy from grandma's today, so I am will be out until james gets back from work [19:53] so everyone have a very happy holiday weekend! [19:58] alexisb: tc! [19:58] alexisb: habe a great weekend === kadams54 is now known as kadams54-away [21:18] Bug #1440199 was opened: api/uniter: Panic: stateSuite.TearDownTest

[21:18] Bug #1440205 was opened: cmd/juju: PANIC: addrelation_test.go:11: SetUpTest.pN39_github_com_juju_juju_cmd_juju.UserSuite

[21:18] Bug #1440209 was opened: juju action do doesn't accept non-string params on command line [21:24] Bug #1440213 was opened: worker/uniter: TestLeadership.pN51_github_com_juju_juju_worker_uniter_test.UniterSuite [21:30] cmars: you have a minute to look over http://reviews.vapour.ws/r/1380/? [21:31] cmars: I've switched it over to YAML and am using the new cmd.FileVar.Open method [21:36] Bug #1440213 changed: worker/uniter: TestLeadership.pN51_github_com_juju_juju_worker_uniter_test.UniterSuite === urulama is now known as urulama__ === tvansteenburgh is now known as tvan-afk [21:48] Bug #1440213 was opened: worker/uniter: TestLeadership.pN51_github_com_juju_juju_worker_uniter_test.UniterSuite [21:48] Bug #1440219 was opened: worker/uniter:TestLeadership.pN51_github_com_juju_juju_worker_uniter_test.UniterSuite === kadams54-away is now known as kadams54