[00:32] wallyworld: review done [00:32] thanks menno, will look soon [01:02] wallyworld: I was planning to do a small fix for cinder in goose today, but won't be able to if we can't use the version changes in juju [01:03] wallyworld: you think it's too close to be doing that? [01:21] rick_h_: thanks for all the feedback [01:39] wallyworld: ffs [01:39] wallyworld: I'm trying to get controller details for a non-responsive controller [01:39] but show-controller makes an api call [01:39] so doesn't return [01:49] [01:49] ..aww, I miss Dave [01:55] axw: prometheus review done. good stuff. [01:56] menn0: thanks! [01:56] menn0: develop seems buggered, I got a test failure to do with SCP commands which I went nowhere near [01:57] axw: yep, that's failing each run in CI too [01:57] I'll dig a bit later [01:57] axw: I worked in that area a few months ago so I can try and take a look too. [01:58] axw: i'm bringing up the issue of develop being permanently cursed in the tech board meeting [01:58] menn0: dimiter made some changes last week to do with connecting to multiple addresses, most likely introduced then [01:59] axw: ah right. that seems likely. [01:59] menn0: ok, sounds good. it's been a week since it merged :/ [02:00] axw: actually, I'll take a look now unless you really want it [02:00] menn0: nope, knock yourself out [02:02] * thumper sighs [02:02] menn0: got things to panic [02:02] not sure if this is moving in the right direction [02:03] thumper: this is the cloud credentials handling? [02:03] yeah [02:04] oh ffs [02:05] * thumper wants to slap whoever did the core/description part of cloud-credentials [02:05] menn0: the cloud credentials are never set on the outgoing model [02:05] now that I'm passing them in [02:05] it panics with "" not valid [02:30] thumper: show-controller does make an api call as it needs that info. list-controller doesn't. juju controllers --format yaml is what you want [02:30] thanks [02:35] that is uh... not very intuitive [04:11] wallyworld: just want to check I understand your email to heather. are we not updating dependencies.tsv now, because we're too close to 2.0.1? [04:12] wallyworld: I was hoping to get https://github.com/go-goose/goose/pull/26 in and updated in juju, but can't do it without the api version changes [04:13] axw: i was refering to pull 25 which is the base work to support multi version endpoints. i didn't see the point in landing that just before 2.0.1 if there was no functional change [04:13] but if you have a 26 which actually fixes something.... [04:13] wallyworld: yes, this will fix https://bugs.launchpad.net/juju/+bug/1636648 [04:13] Bug #1636648: cinder fails with badRequest... "Invalid input for field/attribute device" [04:13] then that changes my thnking [04:13] do you see where i was coming from? [04:14] wallyworld: yes [04:14] wallyworld: but I thought this was meant to go through OIL before we cut 2.0.1 [04:14] so not sure on the timing [04:14] yeah, we will need to ensure that [04:14] ok [04:14] let's land yours and we can track progress through OIL [04:15] wallyworld: ok, sounds good. if you have a momen to review that, please do - it's pretty trivial [04:15] * axw lunches [04:15] looking now in fact :-) [04:46] wallyworld: thanks [04:46] np [05:04] axw: oops (one liner) https://github.com/juju/juju/pull/6500 [05:06] wallyworld: LGTM [05:06] ta [05:07] wallyworld: OK for me to land https://github.com/juju/juju/pull/6428 now then? [05:07] then I'll base my update on top [05:07] yep [05:08] if you are happy with it [05:08] looked fine, I'll take another look over [05:11] wallyworld: actually hrm, looks like we're never assigning s.Openstack, so we'll leak HTTP servers [05:11] wallyworld: I'll cherry-pick and fix that, and include in my PR [05:11] sgtm, ta [05:47] wallyworld: bugger, seems there's an issue with keystone metadata source not working on heather's branch [05:48] oh, for the simple streams data [05:48] hmmm [05:48] yo think the issue is on the goose side? [05:49] or in that juju pr [05:49] wallyworld: dunno, looking into it now [05:49] ok [05:55] wallyworld: it looks like the object-store check is not quite sufficient for product-streams [05:55] wallyworld: what do you think about disabling the API version check if the user passes in "" for the version? [05:56] (and just use whatever's in the service URL) [05:56] service catalogue [05:56] i *think* that's ok testing notwithstanding [06:06] wallyworld: https://github.com/go-goose/goose/pull/27 please [06:06] running a test in Juju now, so far so good [06:07] it found the image metadata [06:26] axw: sorry, quassel has stopped beeping at me, missed ping [06:29] axw: lgtm with a suggestion. glad it was a simple fix [06:44] wallyworld: last one I promise: https://github.com/juju/juju/pull/6501 [06:44] no worries [06:45] wallyworld: first commit is heather's, unchanged [06:45] ok [06:47] axw: lgtm, suggested a test for the no device case [07:09] wallyworld or axw: here's the fix for the issue that's causing lots of failures in develop: https://github.com/juju/juju/pull/6502 [07:10] menn0: looking [07:10] and thank you [07:10] just writing up QA steps [07:10] yay [07:15] menn0: LGTM, thanks [07:15] axw: cheers [07:16] wallyworld, axw: there's still other tests that are failing intermittently and need attention, but this gives us a better chance of a bless [07:16] yes, ty menno [07:16] * wallyworld off to soccer, bbiab === frankban|afk is now known as frankban [07:52] macgreagoir: hey there [07:52] dimitern: \o [07:52] macgreagoir: rick_h_ suggested we should pair on the ipv6 setup / stuff [07:52] :) [07:53] macgreagoir: unless you are busy otherwise atm, ofc :) [07:53] dimitern: Can I get back you about that? I'm just trying to finish part of it now. [10:58] hi, lads. Where do I find wishlist for juju2 on launchpad? [10:58] And ladies, hello! [11:01] hm, you mean as in features? [11:07] Is ftp used as a protocol in any operation of juju ? [11:07] nope [11:09] I have enabled TRACE logging to see why controller running inside LXD fails to download image to deploy a charm on a seperate LXD. I get the following http://paste.ubuntu.com/23382970/. Information there doesn't help. What could i further look into? pcap? === Guest59736 is now known as ahasenack === ahasenack is now known as Guest33774 [11:22] deanman: do you actually have routing to streams.canonical.com from that machine? [11:26] The key differentiator is proxy, if not using proxy my setup works just fine, if using proxy the guest Xenial machine is able to download image and boot LXD container but not when deploying a charm. [11:27] mgz: You mean whether this URL is resolvable? [11:27] yeah, in this case with the same proxy settings I guess [11:28] you probably do need some lower level inspection to see whether the correct requests are being sent out [11:28] mgz: I can manually curl that address and see content returned, i guess juju would have nagged anyway if it couldn't reach that URL, could it be that somehow proxy messes with content returned? [11:31] it's possible, but that log does look like juju nagging (at info level only) [11:38] I can deduce from https://github.com/juju/juju/blob/juju-2.0.0/tools/lxdclient/client_image.go that it should have failed earlier if it was a network problem. === petevg is now known as petevg_afk === Guest33774 is now known as ahasenack === ahasenack is now known as Guest66764 [12:56] hey I suddenly have to run to the bank, bbl [13:00] voidspace: do you have a mo to help me work out if the bug I have is in juju or maas? [13:15] mgz: yes [13:15] mgz: hangout or PM? [13:25] voidspace: hangout? [13:25] mgz: cool - I have 5 mins before 1:1 with rick_h_ though. [13:25] lets be fast [13:25] kk [13:25] mgz: see you in core [13:46] mgz: right, 15 minutes until hangout - I'm digging out the maas docs [13:47] * mgz hopes back in core [13:47] hops [13:47] not hopes [13:47] mgz: hopes is better [13:52] odd, somehow my calendar got set to tuesday as the first day of the week [13:53] oh, no, it was on 7 days.. wacky [14:00] katco: voidspace dimitern dooferlad ping for standup [14:02] omw [14:02] mgz: it will be useful to dump the NIC params here e.g.: 2016-10-26 11:31:50 INFO Changed existing interface: node-71f20964-71e9-11e5-8007-525400c43ce5 eth0 [14:02] [14:03] dimitern: yeah, that could do with some more details [14:13] mgz: I'm not sure maas can satisfy *both* a placement with --to and bootstrap spaces constraints [14:14] mgz: if it happens some times but not others, the issue is probably hidden behind the log I pasted above [14:15] I expect --to overrides [14:15] anyway, the --to on bootstrap is fine [14:15] it's the bundle deploy which just uses constraints where things get interesting [14:58] * rick_h_ heads home from coffee shop [15:04] rick_h_: are you in the 2.0 retro? [16:00] balloons: hey i can't chat in bluejeans, but please don't hold back! [16:05] katco, I think the feedback is most useful in the context of how to make it work at Canonical. I am +1 for doing it -- have been for some time [16:09] dooferlad: +1 to all of that [16:09] katco: thanks [16:17] balloons: yes, i agree: it's paramount to confine the discussion to how it should work at canonical. that's a very agile concept anyhow :) === frankban is now known as frankban|afk === Guest66764 is now known as ahasenack === ahasenack is now known as Guest21645 [18:30] rick_h_, ping [18:35] alexisb: pong [18:35] heya rick_h_redir has a request [18:41] rick_h_: re. your request: what priority is this? should i stop working on everything else? [18:41] katco: yes please to unblock release [18:41] rick_h_: ok, starting on this [18:43] k, new isp about to connect me expect me to be unresponsive, apologies [18:49] rick_h_: so wait... we want develop -> 2.0? we're skipping staging? [18:52] katco: yes, staging is a subset of develop at the moment. We need the updates in develop. [18:52] rick_h_: can you hangout rq? [18:53] katco: sure thing, meet you in ?core [19:20] rogpeppe, katco: morning, I've been thinking more about the retry thing, and I'm starting to get your viewpoint around the error handling. As my wife knows, I can be very stubborn but hopefully teachable [19:21] thumper: hiya [19:21] thumper: lol "strongly held beliefs loosely held" (thanks fwereade) is not a fault in my book [19:21] I just wanted to say so "in person" [19:21] thumper: thanks for continuing to think it through :) [19:22] rather than just email [19:22] thumper: that's appreciated [19:22] there is a wonderful saying [19:22] thumper: rogpeppe: and thank you to both of you for putting so much effort into getting this right [19:22] you shouldn't hold on to a wrong idea just because you spent a long time making it [19:22] thumper: yeah i immediately stole that one. not sure where he got it from [19:23] * katco has a moment of silence to miss fwereade's contribution here [19:23] thumper: i like that - that is a good motto. apposite on many occasions i think. [19:23] * thumper nods [19:24] thumper: shall we just go with gopkg.in/retry.v1 then? ISTM that it's something that deserves a larger scope than juju/ [19:25] thumper: (i removed the dependency on utils/clock, so it should be usable by external projects that want to keep their dependencies tight) [19:26] I'd like to bring things together first, then think about where [19:26] I also think that we shouldn't depend on utils/clock [19:27] we have always said that github.com/juju libraries aren't just for juju [19:28] thumper: i agree that those libraries aren't just for juju, but i think the name does imply some juju relatedness to people. [19:28] yeah... [19:28] there is some squatter on the canonical name [19:29] thumper: and some parts are definitely juju-centric - the way that an import of juju/utils mucks with gobal variables in net/http means I'd never recommend that to anyone external, for example. [19:29] what in juju/utils mucks? [19:29] thumper: FWIW i think that snappy is already using the above import. [19:30] thumper: http.go:20 [19:30] ugh [19:31] that seems like a wrong place [19:31] thumper: yes [19:31] libraries that modify globabls of other packages is very antisocial [19:31] thumper: at least it doesn't make all HTTP connections drop after a single request like it did for years [19:31] pfft [19:32] thumper: https://bugs.launchpad.net/juju/+bug/1491608 [19:32] Bug #1491608: importing juju/utils should not side-effect http.DefaultTransport

[19:32] Bug #1636960 opened: Manual provider that spawns LXD containers [19:39] rogpeppe: wow, that's bad [19:52] balloons: ping, can you sanity check? https://github.com/juju/juju/pull/6504 [20:13] It is surprisingly hard to get url.parse to actually reject things [20:26] so i'm on a new machine and having trouble building gh.com/lxc/lxd [20:27] i can't remember if i ever had to set anything up... it looks from the README.md that a go get github.com/lxc/lxd isn't expected to work? [20:27] there's additional libs required? can anyone confirm? [20:29] katco: from the makefile: [20:29] # Must a few times due to go get race [20:29] -go get -t -v -d ./... [20:29] -go get -t -v -d ./... [20:29] -go get -t -v -d ./... [20:30] natefinch: i'm getting errors on includes of .h files [20:30] natefinch: i highly suspect my setup, but i wanted some confirmation [20:30] natefinch: you're able to just do a go get? [20:31] katco: I think I've built it before... possibly did some work to get there, I don't remember. I presume you followed the steps in the readme? [20:31] natefinch: nope, trying to avoid it [20:32] ha [20:32] natefinch: i don't want bleeding-edge lxd on my system, and the ppa doesn't seem to separate out that from the build-deps? [20:33] I dunno.... tych0? [20:35] i htink the ppa bundles everything, and the archive version uses the archive's deps [20:35] what's the problem, exactly? [20:35] tych0: is it possible to use the ppa for building but not for installing a binary? [20:36] katco: sure, just build the binary but don't copy it to /usr/bin [20:36] or did i misunderstand? [20:36] tych0: if i add the ppa, and do apt install lxc, won't it install the binary from the ppa? [20:36] tych0: or upgrade for that matter [20:36] the `lxc` binary comes from lxd-client [20:36] tych0: ah ok different ppa? ppa:ubuntu-lxc/lxd-git-master is safe for building? [20:37] katco: i don't even think you need the PPA to build [20:37] i think the archive versions of golang should be okay for every release [20:46] tych0: you are correct, it was indeed the way i had things set up. thanks for confirmation! and ty natefinch [20:46] katco: sure, np [20:47] balloons: ping? ut? [20:47] katco, yes [20:47] katco, sure let me take a quick look [20:48] balloons: ta [20:48] katco, the buildbot is going to ignore this -- personally I think if we're happy we can just merge it [20:48] it's been tested once, and we'll give it a full run once it's merged [20:49] (it ignores pr's not going into develop) [20:49] balloons: ah ok [20:49] balloons: so i'll just merge and we can start a test run? [20:49] yea [20:50] looks like you have the proper commit [20:50] balloons: cool, merged. [21:43] katco: sent you an email about 2.0.1 [21:43] wallyworld: ok tal [21:43] thanks in advance :-) [21:43] wallyworld: hth, shouldn't take long (WITH THE POWER OF EMACS!) [21:43] lol [21:44] katco: emacs, making development great again [21:44] wallyworld: lol [21:45] * wallyworld prefers vi :-) [21:45] wallyworld: no one is perfect [21:45] funny, funny [21:45] wallyworld: i thought you liked ides? [21:45] yeah i do, but when i need to do quick editing, vi is it [21:45] wallyworld: for me it's emacs -nw -q [21:46] of course it is :-) [21:46] which i have aliased to qmacs for "quick emacs" ;p [21:52] wallyworld: so you can't cherry-pick merge-commits, so i'll have to try and CP all the commits for that merge [21:53] katco: oh joy. i had a thought - just after that PR was another (test only fixes) to solve bad intermittent juju scp test fixes. might be worth taking that one so we get a good/better CI run [21:53] wallyworld: was that the PR from menn0? [21:54] yeah [21:54] next commited PR after the goose one [21:55] wallyworld: the next commit i see is bebd627 which is from dooferlad [21:55] wallyworld: i don't see anything from menn0 [21:56] katco: https://github.com/juju/juju/pull/6502 [21:56] menn0: ta... weird wonder why my local log is displaying something different [21:57] menn0: dang it, last fetch didn't go through [21:57] menn0: i see it now, ty [22:02] katco: cool. sorry was on a call. [22:03] wallyworld: i lied earlier, i can cherry merge commits [22:04] awesome, i'm sure i have done it before so i was a bit confused but in a meeting so didn't really think too hard [22:04] katco: so think you are cherry pick 2 commits right? [22:05] wallyworld: cherry-pick -m 1 [22:05] yep [22:05] that's what i use [22:05] so you are taking the goose commit and menno's [22:07] wallyworld: correcy [22:08] wallyworld: correct [22:08] tyvm [22:16] wallyworld: menn0: sanity check please: https://github.com/juju/juju/pull/6505 [22:16] looking [22:16] katco: looking [22:17] anastasiamac, https://hangouts.google.com/hangouts/_/canonical.com/alexis-bruemmer [22:17] katco: looks good ty [22:17] when you ar eready [22:17] menn0: hit merge conflicts on yours [22:17] menn0: i think i did the correct thing [22:18] katco: LGTM [22:18] balloons: are you up on all this? [22:18] wallyworld: menn0: tyvm [22:18] yes he is :-) [22:19] i raised it inthe release call [22:19] so it's my fault you needed to do this :-D [22:19] wallyworld: ok :) will need a new test run balloons [22:19] just merged [22:19] will will let the current run finish [22:19] and then look to the new one [22:20] wallyworld: it was a hooge deal. terrible. and some commits i assume were good. but i make the best commits, everyone says so. [22:20] indeed :-) [22:20] that was my attempt at trump-speak if that wasn't evident =| [22:43] axw: already up? I have a vsphere thing to ping pong with you [23:08] perrito666: I'm here now [23:11] axw: are you on bluejeans? [23:12] Bug #1636960 changed: Manual provider that spawns LXD containers [23:12] perrito666: I am now [23:16] thumper, menn0, wallyworld ping [23:21] Bug #1636960 opened: Manual provider that spawns LXD containers [23:30] Bug #1636960 changed: Manual provider that spawns LXD containers [23:44] axw: ping after your breakfast etc and we can talk to thumper about cred migrations? [23:44] I'm looking to head up the road to a cafe for a bit [23:45] which isn't good for calls or irc [23:45] ok, can do it later [23:45] or now? [23:45] I really don't think there is a huge amount to talk about [23:45] i think axw needs to do school drop off etc [23:45] thumper: now's fine for me. I just resent an email from june if you want to read that first [23:45] nah, michelle's on leave so she'll take charlotte to school [23:46] what menn0 and I talked about yesterday was pack up the creds with the model, and error out if there are some that are on the controller already but differ [23:47] should we join the standup hangout again? [23:47] ok