[00:10] thumper, wallyworld (anyone) Tests are passing, ready for another review: https://github.com/juju/juju/pull/8498 Have been doing some manual testing as well [00:44] veebers: awesome, just got out f all my meetings for a bit (until next one), will look [00:45] 181 files! close to a record :-) [00:45] i think babbageclunk has the recent record with his raft one [00:47] veebers: remind me - the utlimate plan is to go to macaroon.v2 rather than unstable right, but we need to do this in stages [00:52] wallyworld: there is no macaroon.v2 [00:52] it is all unstable [00:53] my brain is almost exploding [00:53] threading the presence stuff through hit model migrations [00:53] wallyworld: initially we wanted to up deps but stopped short of macaroon v2-us, then we decided to get in line with what charmstore.v5 had (hence the macaroon-bakery.v2-us parts) [01:18] veebers: review done, a few small things only [01:24] wallyworld: awesome, I'll get them sorted posthaste [01:25] let me know if anything is unclear [01:39] babbageclunk: got a few minutes? [01:41] babbageclunk: are you looking at thumper's comments? [01:41] wallyworld: yup [01:41] also thumper: yup [01:42] thumper: in 1:1? [01:42] ok, i'll look at thumper's PR next [01:42] babbageclunk: ack [01:43] babbageclunk: waiting [01:48] wallyworld: re: the comments, er, comment. Is the expectation something like "RootKey implements bakery.Storage.RootKey" or more: "RootKey returns the rootkey found in the bakery storage, used for making new macaroons." (same for Get) [01:48] if the method comes from an interface we typically use the former [01:49] wallyworld: cool, does it make sense to have the package in there? (i.e. bakery.) or is . the done thing? (i.e. Storage.Get) [01:50] i think we mostly do not include the package name' [02:01] babbageclunk: not too big, just 900 lines or so [02:02] thumper: hey you tricked me! [02:11] thumper: review done [02:13] wallyworld: awesome, thanks [02:28] Is anyone else running the pre-push git hook? [02:33] wallyworld: can you clarify your comment "These need to operate on both the embedded suites"? [02:33] babbageclunk: oh, no I'm not but I should be. Its just a matter of symlinking something isn't it? [02:35] babbageclunk: i normally run it [02:36] veebers: if a suite embeds > 1 other suite, you need to call the SetUp/TearDown funcs on all the embedded suites or else only one will get run [02:36] the other embedded suites will not get setup/torndown [02:38] veebers: yup - see here https://github.com/juju/juju/blob/master/CONTRIBUTING.md#local-clone [02:38] wallyworld: ah I see, I was only matching any setups with a teardown. [02:39] wallyworld: does that mean I need to add a loggingsuite setuptest too? (or are it's setupsuites enough [02:39] veebers: the embedded suites may not explicitly define a SetUp (for example) *today*, but if one were to be added.... [02:39] if loggingsuite doesn't have one at all, then no needed [02:40] but if there's one by virute of an embedded suite in loggingsuite [02:40] wallyworld: loggingsuite has a setuptest, so I'll add it too [02:40] great, yup [02:41] and order of teardown should be opposite of setup [02:41] ack [02:42] wallyworld: does the same count for setup/teardown suite? (i.e. I have the looging one there, should I add the mgo) [02:43] any embedded suite that has a setup/teardown needs to have those called by the outer suite [02:43] otherwise go will be a single, arbitrary one to call and not call the others [02:43] s/be/pick [02:44] ah ok, gotcha [02:44] same even goes for if you add logic to the outer setup and there's only one embedded suite [02:44] still need to then call the embedded suite setup [02:59] thumper: are you happy with https://github.com/juju/juju/pull/8571 or do you want wallyworld to look at it too? [03:00] wallyworld: re: the macaroon.New wrapper, would the function be New or NewMacaroon() to be clearer? [03:01] veebers: NewMacaroon works i think as it will be in a testing package [03:02] sweet, makes sense [04:02] babbageclunk: all good [04:02] thumper: cool, thanks! [04:03] jam: in our 1:1 [04:03] wallyworld: FYI have pushed pr review changes. Also I had a go at getting a failure when using a charm with resources, but couldn't. It's possible I don't fully understand the process required. [04:03] thumper: omw [04:04] I published a charm to the store, I've deployed it (it got me to auth), I upgraded the controller + model to the 2.4-b1 and attached a new resource to teh application but it all worked [04:05] ah man, failling tests. I should have caught that locally :-\ [04:06] ah, I see why. [04:08] babbageclunk: is there a nice way to "go test" but just build the test not run them (i.e. so I can catch compile errors) [04:09] veebers: go test -c [04:09] But I haven't really worked out a nice way to do that across all packages. [04:09] babbageclunk: cool thanks. Huh that succeeds, but there are errors to be found :0\ [04:10] ah, just in juju/juju won't do much will it [04:10] It only builds the tests for the package you're in. [04:10] aye :-) [05:26] sorry, got a quesiton about current supported golang version, `make install-dependencies` will install golang-1.10 from snap but README.md shows it should be 1.9. I am currently on `develop` branch [05:26] seems README needs to be updated. [05:26] kelvinliu: yes [05:26] it was 1.9 until recently [05:27] thx :) [05:27] we should tell folks just to use Go from the snap [05:28] yes, snap is handy rather than installing manually [05:28] it is, and it's easy to try rc of next go version etc [05:28] so we can be sure juju will continue to work when next version is out [05:29] yes, agreed [05:29] wondering if I wanna update the README, how do we manage branchs? [05:30] sure, feel free to propose a change. vino had the same thought but i've given her a bug to work on so you could look at a readme change [05:32] jam: ping? [05:38] night all [05:38] * thumper goes to make dinner and have a glass of wine [05:39] yup thx [05:40] night thumper [05:40] kelvinliu: he never waits for responses. I was going to say it too. :( [05:46] haha [06:28] a very quick and easy review plz: https://github.com/juju/juju/pull/8584 [06:58] anastasiamac: looking [06:59] wallyworld: thnx [06:59] anastasiamac: nice [07:00] m thinking it should mayb (!) say 'odel cannot be upgraded to x.x.x while the controller is y.y.y: upgrade 'controller' model first'... [07:00] does it read better to u^^ [07:01] wallyworld: ^^ [07:01] sure, sounds good to me [07:01] \o/ [07:01] i'll re-phrase in a sec... m just a breath (or a few) away from proposign worker/maniforld for cred check... [07:02] jam: i had to make a fix to a recent juju/description commit which added some stuff for caas support, any chance of a review before you EOD? i need this to be able to add support for model import/export in juju itself https://github.com/juju/description/pull/37 === frankban|afk is now known as frankban [07:27] hi kelvin [07:33] hi vino [07:34] hi.. did u manage to get details abt branches ? i closed freenode. [07:38] yes, https://github.com/juju/juju/blob/develop/CONTRIBUTING.md [07:39] just have a check CONTRIBUTING.md [07:41] jam: thanks for review, i responded to comments if you are interested. i'll test with model export in juju before landing [07:44] kelvinliu: i left a comment - let's change the readme to tell people ot use the snap [07:45] yes, go point, I m updating it now [07:54] snap install of go ? [07:56] yes [07:57] i had issue installing snap version by default installed 1.6 and that had issues with juju installation [08:01] kelvinliu: lgtm [08:01] vino_: it should have installed the latest stable go release, not sure why it installed 1.6 [08:02] juju won't compile with go < 1.10 [08:02] yes. in 16.04 [08:03] hmm, no idea why off hand. i'm running bionic [08:04] i am running bionic as well [08:04] `make install-dependencies` will install all for u, vino [08:04] so i tried upgrading it to 1.10 and eventually it allowed me to do it only with longsleep/golang-backports [08:05] maybe i can recheck [08:05] https://bugs.launchpad.net/juju/+bug/1763201 [08:05] Bug #1763201: Build juju from source documentation requires update [08:05] vino_: you sure you used the snap? [08:05] rather than the one from the archive? [08:05] i'd expect 1.6 to be in the archive [08:05] i did. but i can recheck again. [08:06] the reference to backports implies archive [08:14] i rechecked and snap installs 1.10 - i was wrong here. [08:15] wallyworld : i reachecked and i couldnt reproduce how i got 1.62 where i faced issues to upgrade and referred to link : https://github.com/golang/go/wiki/Ubuntu [08:17] no worries. we should update that wiki page to refer to snap also [08:18] It already does. [08:18] https://github.com/juju/juju/blob/develop/README.md [08:18] snap also or snap only ? [08:21] vino_: i meant this one https://github.com/golang/go/wiki/Ubuntu [08:22] oh never mind [08:22] i didn't scroll down far enough [08:22] it already mentions snaps [08:22] I am not clear. [08:22] apologise [08:23] did u mean that juju README.md should also mention snap ? [08:23] for the juju readme, IMO it should be snap ony [08:23] since people on xenial etc would otherwise get an older version of go that won't ork with juju [08:23] ok snap only is correct. apt install info can be removed [08:23] ok. [08:24] yeah i think so [08:41] wallyworld: if u plan more insomnia or r just looking for bedtime reading, PTAL https://github.com/juju/juju/pull/8586 - cred worker/manifold... :D [08:42] ok, will try and get to it after i finish this current PR [08:42] wallyworld: was kind of hoping u'd look at it at the earliest tomorrow :) [08:42] can do [08:52] wallyworld : i will update the bug i opened this morning https://bugs.launchpad.net/juju/+bug/1763201 [08:52] Bug #1763201: Build juju from source documentation requires update [08:53] but just to update : both snap and apt-get install go install 1.10 and works fine for me to build juju and installing. [08:54] i will detail in the bug report [08:57] snap install go --classic is fine [08:58] only the next option needs to be updated with - sudo apt-get install golang-go [09:02] hi Kelvin.. Can i assign that bug to you as you are updating it ? [09:36] sure, just give it to me. I v already got a PR for it. Vino [09:47] i just saw that. [09:47] i felt u can add apt-get as it works fine for me [09:48] it installs latest go version. [09:49] i have assigned it to u [09:52] get u get mail notification kelvin ? [10:00] https://bugs.launchpad.net/juju/+bug/1763201 [10:01] Bug #1763201: Build juju from source documentation requires update [10:35] manadart: if its possible while you're in the area, bug #1753418 is one reason why our ci run isn't clear. [10:35] Bug #1753418: intermittent failure in kvmProvisionerSuite.TestKVMProvisionerObservesConfigChanges

[10:36] jam: Yep. Will take a look. [10:36] manadart: lower priority than finishing the work, but since you're working with KVM right now [13:47] externalreality, are you ok with doing the change for the runtime panics on that bug to start? [13:47] externalreality, I'll look quickly for ovh account [14:55] ballons, sounds good, will do. === akhavr1 is now known as akhavr === dpb1_ is now known as dpb1 === frankban is now known as frankban|afk [18:13] balloons: subprocess.check_output is working… but have to change a lot of things over to use it… [18:13] unhappy it’s not clear why yet [18:18] OK, so you are unblocked? [18:40] balloons: currently - [20:17] morning team [20:19] thumper: morning and +1000 on the collapse of the juju vs -dev list/channels [20:19] in case you were looking for votes. I've been saving up my paper ballots :P [20:19] balloons: where are we with 1.25.14 build needs - go/1.8 etc [20:20] rick_h_: heh, ack [20:20] good morning thumper [20:24] also status for 2.3.6? [20:25] since we are co-opting the release call for whole team meeting :) [20:25] thumper, yea.. We have 3 releases we need to do. 1.25, 2.3, and 2.4 [20:25] yeah [20:25] 2.3.6 first, then 2.4 then 1.25 [20:26] thumper, 2.3.6 has to have a call on if we wait for the container-image-metadata or not [20:26] we've waited long enough for the 1.25 build, another week or two won't kill them [20:26] I don't think we need to wait [20:26] I'm happy enough that going into 2.4 [20:26] field are working around the issue now [20:26] it isn't critical [20:26] 2.3.6 is [20:26] this needs to land then, and we should be good https://github.com/juju/juju/pull/8589 [20:27] balloons: is mongo 3.6 in bionic yet? [20:27] thumper, it went into the queue a few hours ago. Needs approval, then it will land in proposed [20:29] * thumper nods [20:29] * thumper jumps in call 1 of 3 [20:35] Morning all o/ [20:38] * balloons tips hat [21:31] kelvinliu: if you're online, come to team meeting? [21:40] morning everyone [21:58] veebers: if you would, i updated the pr from yesterday with the change [21:59] hml: mean, you have a link? [21:59] hah I have a Link, I hear him crying right now :-P [21:59] veebers: https://github.com/juju/juju/pull/8582 [22:09] feeling paranoid - i manually did the upgrade series setups on a 2.3.6 config - then did juju upgrade-juju to get to 2.4-beta1 and the unit was still happy. :-) [22:13] hml: what's the difference to what the script used to be doing (with the .tmp that's different to using --output-file) don't both write a 'tmp' file and then mv it in place at the last second? [22:14] veebers: there are shades of “last second” [22:15] veebers: i could have just removed the os.rename and done the mv after ifdown [22:15] veebers: but since we can specify the input file, it’s “nicer” to specify an output file too? [22:16] hml: perhaps I don't fully understand the script . . . ah I see, right *that's* the important part here, not moving the interfaces file to bak before it's time [22:16] I was focusing on the wrong part :-) [22:17] veebers: i changed so it was a copy to .bak - then the replacement was done very last we could [22:17] hml: ack, that makes sense and obvious now that I see it [22:18] hml: LGTM [22:18] veebers: TY! :-) [22:22] thumper: Seems it's really easy to add gometalinter to emacs (using spacemacs too): https://github.com/syl20bnr/spacemacs/tree/master/layers/%2Blang/go#pre-requisites [22:23] veebers: Also really CPU hungry [22:24] (Probably because of the size of the codebase) [22:25] babbageclunk: ah right, I'm not so keen to add more CPU load to my emacs usage ^_^ [22:28] C, c++ [22:28] and python scripting [22:29] in cloud i used JAVA RestApi [22:29] vino: might be wrong channel :-) [22:30] yes yes. [22:30] vino, kelvinliu: Where you making changes to CONTRIBUTING.md? I think I want to make some updates too [22:30] i am still sleepy [22:31] ^_^ [22:31] i already generated a PR [22:31] in the Local Clone section. [22:31] Is master even a branch we really care about now? I think CONTRIBUTING.md should mention develop, not master. [22:32] https://github.com/juju/juju/pull/8588 [22:32] vino: lol that's exactly the change I was going to make [22:33] Oh sorry. [22:33] i missed something and Kelvin did it yesterday in README>md [22:33] vino: hah no it's good, you got to it first. [22:33] the go version [22:34] vino: note, there is a typo in the PR typo, it's a little long too (will be truncated in the git log) [22:38] done. [22:38] corrected it [22:40] vino: you'll need to update the summary part too, it still has "…ection with correct path" as the first line [22:41] good eyes. [22:46] has anyone tried to bootstrap from a mac recently? (i’m using a remote openstack cloud) [22:48] babbageclunk: i might be a minute late for standup, so tell some jokes if i'm not there on time :-) [22:48] wallyworld: I'll do my best man, but if the audience are expecting you they'll still be disappointed [22:57] how do you build agents for a different processor? trying to bootstrap 2.4-beta1 (develop) it complains about not having agent binaries, even if they are built during bootstrap. Wrong OS? and the 2.4beta1 binaries are not online yet [23:03] veebers: review plz? https://github.com/juju/juju/pull/8590 [23:04] babbageclunk: sure thing! [23:04] hml: Don't know, sorry! anastasiamac/balloons/thumper? [23:05] veebers: fanks [23:05] i was able to bootstrap with 2.3.5 at least? [23:07] I think once you have a controller you could build the binaries for the different arch and then use sync-tools to push them up, then upgrade? [23:08] hrm [23:08] get juju to build the binaries? [23:09] babbageclunk: how would you get the controller to build the binaries? [23:10] no, build them locally - I think by setting GOOS and GOARCH? [23:11] hml: Something like this https://dave.cheney.net/2015/08/22/cross-compilation-with-go-1-5 [23:11] haven’t played with those before [23:11] babbageclunk: "Thanks ants, Thants". lgtm [23:12] veebers: Just... look around you! [23:12] hml: with 2.3.5 it would have pulled agent from streams, for 2.4-b1 you'll need to use the testing streams to do that [23:12] babbageclunk: ^_^ that's the one [23:12] veebers: there’s a testing streams? [23:13] hml: for each build we generate a stream entry and put it in s3. [23:13] veebers: though that won’t allow me to test my code easily? [23:13] veebers: how do you use it? [23:14] hml: not if you're needing an agent you've just built :-\ [23:14] veebers: sounds like i need to learn to use GOOS and GOARCH? [23:15] veebers: i was just courious [23:31] hml: if you have a log of the upgrade series test too that would be useful [23:38] veebers: not sure if the logs i have now are useful? [23:38] veebers: i know where they are failing now… [23:38] veebers: if i find a new one, i’ll forward along :-) [23:39] hml: sweet :-) [23:40] veebers: eventually i’d be curious as to what broke which lead me to using subprocess.check_output to run remote commands [23:40] some juju some not [23:40] veebers: it’s not warm fuzzy that this is needed [23:40] but i want it done too :-) [23:41] * hml waves goodnight! [23:41] o/ [23:53] bye hml