[01:12] anastasiamac: here's the huge PR https://github.com/juju/juju/pull/9024 [01:13] * anastasiamac squints at it [01:15] anastasiamac: aw man, swap? [01:15] babbageclunk: nuh [01:15] * babbageclunk sighs [01:15] also babbageclunk, i won that one as already lgtm'ed :) [01:15] * babbageclunk snost and lost [01:16] tra-lah-lah [02:04] wallyworld: reviewed 9006 [02:05] awesome ty [02:05] will look after these cmr bugs [02:09] babbageclunk: https://github.com/juju/juju/pull/9025, I need to go add a few tests to show things hooked up... [02:09] wallyworld: ^^ [02:09] will look soon [02:09] sure, np [02:09] it still needs some changes [02:11] thumper: 1st sentence of the description seems like it's truncated? [02:11] uh... yeah [02:12] probably matches my thought process while typing [02:26] babbageclunk: so writing the extra tests I felt that I really wanted the Reporter interface defined in the worker package, so ...https://github.com/juju/worker/pull/6 [02:48] * thumper wonders if the spam attack on freenode is done yet [02:50] seems not [02:54] babbageclunk: i just glanced at review, ty. quick comment - params.IsCodeNotFound(err) sadly does barf if err is nil :-( [02:55] i found out the hard way [02:56] wallyworld: https://github.com/juju/worker/pull/6 ?? [02:56] it is very small [02:58] thumper: lgtm [03:25] wallyworld: oh stink! Weird, reading it it doesn't seem like it should. Maybe a typed nil is getting in there? [03:26] babbageclunk: wallyworld is not in the channel?... [03:27] ha, I started replying to him before he left [03:27] wallyworld: oh stink! Weird, reading it it doesn't seem like it should. Maybe a typed nil is getting in there? [03:27] :) [03:28] babbageclunk: yeah, i think that's what it is [03:29] wallyworld: oh, probably because it's a *params.Error explicitly on the struct. [03:29] I think this one is ready now... https://github.com/juju/juju/pull/9025 [03:30] looking [03:36] wallyworld: there are even some worker import cleanups for ya [03:36] i know! [03:39] thumper: looks good to me, all seems fairly straight forward [03:39] wallyworld: yeah, it isn't complex [03:39] just joining dots [03:40] yup [04:40] wallyworld: this is needed to move catacomb, dependency, and workertest out of the juju repo into the worker.v1 https://github.com/juju/testing/pull/139 [04:41] ok [04:41] lgtm [04:41] where is the bot on this? [05:01] wallyworld: https://github.com/juju/worker/pull/7/files [05:01] this just moves the catacomb, dependency and workertest packages to worker.v1 [05:01] ok [05:01] +4000!!! [05:03] lol [05:03] thumper: did we really want to add the dep engine to the worker package. i guess it can go there. but i think maybe the intent was to keep the work package purely as a way to manage go routines nicely [05:03] should the dep endgine be a separate repo [05:04] wallyworld: it is in a package [05:04] wallyworld: the engine runs workers [05:04] ok [05:04] it is quite tightly connected [05:04] lgtm [08:45] anastasiamac: I'm working through the tests atm, but I was more wondering if you think there is any potential issues in changing this, that I might want to test ? [08:46] stickupkid: if u look for hardoded test with "xenial" in it, the test names will give u an indication of whether they now should be 'bionic" :) [09:05] hi all [09:06] I have installed MAAS , dose juju charm have supprot for ubuntu 18.04 ? [10:44] does anyone know how the provider tests work around uploading binary versions? bit of background, I'm bumping the LTS from xenial to bionic and it looks like the right version are being uploaded, but i get an error message stating that there are no versions available... [10:45] otherwise i'll start digging to see how the mock layer works... etc - this seems to hit aws-ec2 and joyent [10:57] stickupkid: commented on the PR but the info feels kind of obvious ;) u've got it under control [10:57] anastasiamac: perfect - thank-you [10:58] stickupkid: as for the verion q above, are u sure ur local copy get uploaded? [10:59] https://pastebin.canonical.com/p/cYsSJ2KGHw/ [11:00] just reading the test log it would seem so, but then again, i'm just started digging into how the test is setup. [11:01] stickupkid: ah, ignore my question.. i thought u were testing live... [11:01] live works, tests doesn't :p [11:01] stickupkid: this i can help u dig for :) gimme a sec [11:42] stickupkid: m stil digging but my output from ec2 TestStartStop is different to urs -https://pastebin.ubuntu.com/p/Xqp4k73PmZ/ [11:43] stickupkid: i have stuff related to availability zones and storage before image search is hit... did u change something else on ur system?. [11:47] stickupkid: also, have played with new images coming in into testing but this looks a place to make sure u have bionic equivalents for xenial entries... https://github.com/juju/juju/tree/develop/environs/imagedownloads/testdata [11:51] stickupkid: or mayb not, as they seem to be only used for kvm and wrapped command testing... [12:05] bdx: hi [12:05] bdx: did you deploy cs:~omnivector/sentry recently ? === Cynerva_ is now known as Cynerva [14:58] anybody think I should spend the time on cleaning up LTS - https://github.com/juju/juju/pull/9023#issuecomment-411081657 [15:13] stickupkid: clean up is good, it’s hard to get to later and we’ll have forgotten the pain by the next LTS. my 2 cents. :-) [15:20] rick_h_: i think i found a flaw in the way the EnvVar are used in juju add-cloud, they don’t get validated. :-/ [15:21] hml: aren't those things validated in interactive mode? [15:21] hml: they're not validated once they're loaded as the default values and sent through the schema work? [15:21] stickupkid: not if i stick the env var in as a default prompt value [15:21] hml: is it because default values are not validated at all? [15:22] rick_h_: i think that’s it [15:22] well, file a bug I guess. that's a bummer [15:22] rick_h_: i’m still working on that code so i can fix it. [15:22] just annoying. [15:22] must be something with how default and defaultprompt work. [15:23] hml: ok, if it's easy. otherwise let's try to move on. [15:23] we've got 2.4.2 freeze friday and that's not the most important bug for 2.4.2 [15:43] rick_h_: this code isn’t going into 2.4 branch anyways [15:43] hml: right, I just mean that the hope was to address some issues for 2.4.2 this week. [15:44] rick_h_: ack [20:58] wallyworld: ping [20:59] hey, just about to have meeting, can we catch up in say 30 at release call? [20:59] wallyworld: sounds good [21:13] wallyworld: I built your mariadb charm, but I get the error on deploy: from charms.layer.basic import pod_spec_set. I suspect I need to update that somehow? (I"m using snap installed charm command for that) [21:14] veebers: in a meeting, but if you pull the latest source code it will work [21:14] i updated the charms in my repo last night [21:14] to ue the latest layer stuff [21:15] ack [21:16] ah I see [21:30] hml: rick_h_: i'm still in k8s meeting, will be late [21:48] wallyworld: release call is over, ping me when you’re available pl [21:48] pls [22:00] wallyworld: I have a small review for you as well... https://github.com/juju/juju/pull/9032 [22:00] * wallyworld adds it to queue, still in k8s calls [22:00] 395 files changed, +591 −4,900 [22:01] babbageclunk, veebers, hml, doesn't have to be wallyworld to review [22:01] while large, the branch only touches imports, and only touches worker.v1 imports [22:02] (and updates dependency.tsv for the imports) [22:02] * veebers looks [22:03] thumper: why is this going into 2.4, is it for a bug fix? [22:04] it is just code cleanup, and I prefer to keep things clean??? [22:04] thumper: ack [22:09] thumper: with the worker.v1 dep change, is there much difference with the latest and what we release in 2.4.1? [22:09] looks like there is only a couple hours between shas in that diff, not sure off the top of my head which sha we released with [22:11] veebers: can you join us https://hangouts.google.com/hangouts/_/canonical.com/weekly-caas [22:11] wallyworld: omw [22:13] veebers: it is exactly the same [22:13] veebers: well, I took the packages from 2.4 branch and moved them to worker.v1 [22:13] then updated deps [22:14] no code changes due to this move [22:23] ack [22:24] hml: jump back in release HO? [22:24] wallyworld: omw [22:25] wallyworld: I can't trigger the deps error you saw, I do a godeps in juju develop, go to charmstore-client, make deps and make build and it's all hunkydory [22:25] (this is me saying "builds for me" :-)) [22:26] wallyworld: Found the resource type validation spot. :) [22:26] yay [22:26] cory_fu: nice! [22:27] thumper: hmm, I wonder where those 4000 lines of code went? [22:27] babbageclunk: into the worker.v1 package [22:27] repo [22:27] babbageclunk: as a different pr yesterday [22:27] * babbageclunk forgot sarcasm tag [22:28] the sarcasterix [22:28] sarcasm never works well over writtern communication [22:28] some birds have worked out how to use our nectar feeder now [22:29] hazaah [22:29] I can't believe unicode doesn't have the irony mark in it! https://en.wikipedia.org/wiki/Irony_punctuation [22:30] hmm, hang on, maybe it does [22:32] ⸮ boom ? [22:36] great babbageclunk, you've crashed my irc client. [22:36] thumper: done [22:36] kwmonroe: sorry! [22:36] hahah [22:37] thumper: nice! You expecting a lot of Tui to hit it? I love our neighbors have a couple set up, we get so many Tui around [22:37] * veebers watches his charm push crawl to a halt pushing a docker image [22:37] * babbageclunk was tempted to make that sorry sarcastic but it seemed churlish. [22:38] no worries babbageclunk, i've been meaning to upgrade my quassel for a while. this is great motivation :) i also recently realized i can't send a comment with two hash tags without it hanging.. [22:39] Oh, I was wondering how were still talking after your client crashed, but I guess the quassel intermediary is still running fine? [22:40] gah, menno'd it - ...how you were... [22:41] yeah babbageclunk, the server is on something ridiculous like precise (maybe older). it doesn't care about much. my client is like quassel version 0.pre-alpha.scary on bleeding edge osx and it hangs if i look at it funny. [22:42] the good news (maybe?) is that i can fall back to the android client to say things like "great babbageclunk" when my desktop quassel crashes ;) [22:44] kwmonroe: heh [22:47] cory_fu: would love to compare notes when you've sorted the snap build parts; see if there is room for improvements for how we do things currently [22:48] veebers: Definitely [22:48] wallyworld, veebers: PR for tracking master and fixing the proof errors: https://github.com/juju/charm-tools/pull/434 [22:49] cory_fu: wow, quick, looking in a sec [22:50] wallyworld, veebers: ugh. test is looking for the exact proof error. I'll have to fix that, but please review anyway. [22:51] I'll follow-up on the PR tomorrow, but you're welcome to use that branch for a local build in the meantime. Heading out for now. o/ [22:51] see ya [22:51] cory_fu: have a good one o/ [22:52] kelvinliu: i'm free now for 1:1. man so many meetings this morning. entering hour 3 [22:52] wallyworld, yup, cu in HO [22:57] veebers: hoping to get some tui, we did have one around the other day before we up the feeder up [22:58] thumper: we have so many around here they are almost pests ^_^ There will be 4-8 Tui in the tree beside the driveway. Love it [23:00] wallyworld: do you have your mariadb charm pushed and published to the staging charmstore, I cannot get the image pushed with the charm :-| [23:01] veebers: just in 1:1 with kelvin. but no because i couldn't get charm built [23:01] will try again today [23:02] ok, I'll keep trying. I haven't seen the error you mentioned earlier about blank resources, trying to repro [23:06] With Discourse does anyone else feel less likely to 'like' a post because it's a heart and that seems just a little too much for a post? as in "I like your comment, but I don't like-like it." [23:11] veebers: share the love dude [23:13] rick_h_: ^_^ [23:14] thumper: do you have any utilities for making testing code that uses pubsub a bit more convenient? [23:15] veebers: I believe the proper wording is "I'm not ready for a relationship with your comment" [23:15] wpk: hah ^_^ [23:15] wpk: How goes things ? [23:16] veebers: good, good. Spent some nice time with wallyworld, thumper, and jam in Montreal a few weeks ago :) [23:17] wpk: oh, very cool! [23:18] veebers: coincidentally, IETF102 was at the same time, at the same place, as Canonical management sprint [23:20] wpk: excellent timing [23:25] veebers: it would be even nicer if people from Canonical would come to IETF meetings (people from RH come), but still ;) === blahdeblah_ is now known as blahdeblah [23:30] :-) [23:38] babbageclunk: like what? [23:40] thumper: not sure really - just something to make all the goroutines and channels a bit simpler. I might have to roll my own anyway, since it's more request/responsey. [23:40] babbageclunk: perhaps we could chat after lunch, I found some patterns that might be useful [23:41] ok, that would be cool [23:42] babbageclunk: do you want to sub and test off juju core pubs, or charm pubs? [23:43] kwmonroe: I mean testing code that uses the pubsub hub - https://github.com/juju/pubsub [23:44] not sure about those other pubs. [23:45] kelvinliu: had a quick look, everything seems good with that ci-run change, I've rebuilt that unit test job out of interest, see what it does this time around [23:46] veebers, awesome, thanks [23:46] nw [23:46] babbageclunk: aaaahhh. ok then, go ahead and crach my irc client at will. i thought you were referring to a generic pub/sub testing model. eg, you have a bundle that includes something like nagios (which you don't control), yet you want your bundle re-tested and published any time nagios is updated. [23:47] ^^ that's obviously charm specific and has nothing to do with juju/pubsub. but by golly, i could tell you a lot about it. [23:48] kwmonroe: ah, right! Yeah, I could see that being pretty difficult. [23:54] * thumper merges 2.4 into develop [23:56] wallyworld: I can't imagine this is a good thing to see: WARNING juju.workers.caasunitprovisioner k8s event watcher closed, restarting