=== frankban|afk is now known as frankban [09:02] axw: could you take a look at https://github.com/juju/juju/pull/6354 ? I think I made all the changes suggested, I'd like another +1 before I land it [09:02] frankban: looking [09:02] axw: ty [09:07] frankban: I've just pushed one comment, but I'm still reviewing [09:10] cool [09:16] frankban: done [09:17] axw: thanks, good suggestions [10:04] Bug #1629817 opened: [arm64] bad cpu-cores detection [10:39] mgz: ping [10:40] mgz: hmmmm... unping [10:40] mgz: was going to ask for vsphere creds, but I see them in cloud-city as usual [10:50] anyone know if there's a way to list all the users that been granted permissions on a model? [10:51] Morning all [10:52] perrito666: hiya [10:52] perrito666: do you know the answer to the question i just asked, by any chance? [10:56] Mmm there is iirc wallyworld implemented something like that, can't remember where though [11:04] perrito666: i see "juju users" but that tells me all users for a controller, not a model [11:05] rogpeppe: something with shares possibly? [11:05] perrito666: "juju commands | grep -i share" doesn't come up with anything [11:26] mgz: ping about vsphere [13:01] katco: natefinch: I am likely to be late for standup - daughter has to go to an emergency dentist's appt [13:02] voidspace: yikes. Hope she's ok [13:02] natefinch: just toothache, nothing serious [13:02] natefinch: I'm kind of glad - she still has her baby teeth and it's a good lesson as to why toothbrushing is necessary... [13:03] voidspace: fair enough :) [13:14] rogpeppe: sorry I hung on you, bank made me stash my phone under threat of arrest [13:16] lol argentina [13:16] natefinch: well see who laughs after tue wig driven dude gets command of your country [13:17] perrito666: lol [13:17] perrito666: I was going to say, in my country you can get shot for holding a toy gun in the toy gun aisle of a toy store [13:18] in my country you can get shot for rand.Rand{}.int() [13:18] well, using insecure psuedo-random numbers is a capital offense [13:18] lol [13:19] natefinch: the reason for phones being forbidden in banks here is because there was this mo of burglars that would tip from inside the bank to outsiders when someone whitdrew large sums of cash [13:20] (you can see the level of footshooting put into the solution) [13:20] ahh, so, same reason we take off our shoes at the airport. Because "this one time...." [13:20] yup [13:21] I really hope no one does anything odd with shoes in a bank here, or visiting the banks is going to be very funny [13:21] very japanese [14:00] i need a hand setting a breakpoint on juju. I've got gdb and dlv installed. [14:02] cmd/juju/application/deploy.go deployCharm seems to be what is failing on me [14:02] however gdb and dlv don't seem to know how to break on that function [14:03] gdb is dicey at best, but I've had good experience with delve via editor integration. I haven't used it via the CLI [14:04] natefinch, delve kinda blew me away when i did a funcs command. over 100K lines of function def's [14:04] natefinch, any idea how to point delve at the right breakpoint? [14:07] cholcombe: use editor integration, it makes things 1000x easier. What editor do you use? [14:08] natefinch, well i'm trying to debug the controller running jujud. i haven't tried building juju and running locally yet. maybe that's the way to go? [14:10] cholcombe: ahh... delve does have remote debugging capability, which might still be the best way to go. For delve I'm pretty sure you have to build the binary with delve debugging enabled, but I'm not 100% sure [14:10] gah [14:11] perrito666: np [14:15] cholcombe: I might be wrong about that... the docs only mention that if you build with delve it makes the binary "easier to debug" [14:15] yeah [14:15] i thought gdb could handle this also but it appears to not be able to find anything either [14:15] i suspect the jujud binary was built without debugging flags [14:16] gdb is known to not work well with go [14:16] oh ok [14:16] the problem is i have a charm which juju for some reason refuses to deploy [14:18] cholcombe: the best way to debug these things is to start off with the logs..... what does juju do when you try to deploy it? [14:18] natefinch, https://gist.github.com/cholcombe973/d0ab4865691d6996d0c3b20578f269ca [14:19] charm proof passes and this charm looks like any other charm. i'm not sure what juju's problem is [14:19] i'm running 2.0-rc2 [14:20] it used to deploy but i don't remember which version it worked on. probably after i updated to 2.0 beta's it started failing [14:28] cholcombe: that looks like it might be a bug [14:28] cholcombe: let me try here [14:29] natefinch, yeah none of the beta's or rc's work on this charm. i don't know what's special about it [14:29] natefinch, i also pushed it to the charm store on --channel edge and that also fails [14:30] natefinch, juju deploy cs:~xfactor973/gluster-charm-1 --channel edge [14:33] natefinch, i did a charm grant everyone on that in case you want to try [14:34] cholcombe: crud, I just realized the time, I have to run an errand. Can you try deploying another local charm and see if that works? It seems unlikely that the contents of your charm would give a "not found" style error [14:34] other local charms work perfectly fine [14:34] there's something special about this one [14:34] i've tried deleting everything in the repo that is unique and it didn't help [14:35] wacky, ok [14:35] try removing the dash in the name [14:36] I gotta run, maybe someone else can help. Sorry. [15:25] Can anyone help me with a lxd problem? [15:26] I got bitten by this https://bugs.launchpad.net/ubuntu/+source/lxd/+bug/1629766 while trying to install something unrelated. [15:26] Bug #1629766: Syntax error in /usr/lib/lxd/upgrade-bridge [15:27] So I did my best to totally remove lxd and the half-configured interfaces that the failed upgrade left behind. [15:28] But now I can't bootstrap to lxd - I get a message saying to run `sudo dpkg-reconfigure -p medium lxd`, but running it doesn't bring up the network config questions it used to. [15:31] It just says "Warning: Stopping lxd.service, but it can still be activated by: [15:31] lxd.socket [15:31] " and then completes with returncode=0. [15:47] i've factored out some of the juju/worker code from juju/juju so that we can easily reuse it across projects. i made a few changes as i did so - reviews welcome: https://github.com/juju/worker/pull/1 [15:52] natefinch, i found the problem. when i deleted the actions.yaml file it started deploying again. [15:52] must be bad syntax or something. juju really needs a better error message for that [15:56] Bug #1629912 opened: the provision network IP addresses have to be smaller than the ones on the public network === frankban is now known as frankban|afk [16:20] Bug #1629919 opened: destroy-controller fails and a kill-controller is required. [16:39] dooferlad, ping [16:39] or voidspace [16:50] dooferlad or voidspace ping [17:05] alexisb_: pong [17:06] heya voidspace [17:06] just quick q for bug triaging [17:06] https://bugs.launchpad.net/juju-core/+bug/1629912 [17:06] Bug #1629912: the provision network IP addresses have to be smaller than the ones on the public network [17:06] ^^^ looks like an ip ordering issue which is a known bug? [17:07] alexisb_: that's what it sounds like, although it's unclear [17:07] alexisb_: I didn't know we had an IP ordering bug! [17:07] hmm, ok :) [17:08] I will give ti to rick for now then [17:24] hows it going all? [17:25] is there a way to specify what subnet juju bootstraps to? [17:26] Bug #1629912 changed: the provision network IP addresses have to be smaller than the ones on the public network [17:28] is there a way to get the juju controller and model uuid from the cli? [17:30] juju show-model [17:30] juju show-controller [17:30] is that what you were asking hatch? [17:31] alexisb_: how embarassing [17:31] * hatch slowly backs away and closes the door on his way out [17:31] :D [17:32] :) [17:34] hatch: isn't this where your fellow contrymen say their most iconinc catchphrase? [17:34] perrito666: lol [17:50] Bug #1629951 opened: cannot specify subnet to create controller in on bootstrap [18:32] new to Go.. how can I build/install juju 1.25? the instructions seem geared towards trunk... [18:36] gQuigs: after getting the code, git checkout 1.25 (or juju-1.25.6 for the latest 1.25 release) [18:36] gQuigs: you can download/install 1.25 on many machines... you probably don't need to build it unless you want to hack on it [18:37] natefinch: I need to test a 1.25.6 fix that might break stuff [18:37] gQuigs: valid :) [18:37] a binary build of 1.25.6 would be fine, but I haven't found one [18:38] gQuigs: https://jujucharms.com/docs/1.25/reference-releases [18:39] gQuigs: it's hard to find if you don't know how to get there (and even if you do it's like 4 clicks) [18:39] gQuigs: scroll down to "Proposed" [18:40] natefinch: I have actually found that, but the Proposed branch hasn't moved to include the fix I'm looking for (at least according to the PPA) - https://launchpad.net/~juju/+archive/ubuntu/proposed?field.series_filter=xenial [18:40] gQuigs: ahh, bummer [18:41] but I never tried the centos (tar.tz) builds.. could they be newer? [18:42] I'm sure they're all built off the same codebase [18:44] yup, confirmed that, all the same older version :( [18:45] gQuigs: building juju really isn't hard: http://pastebin.ubuntu.com/23271348/ [18:46] gQuigs: lines 4-8 are just setup... after that, rebuilding just requires go install ./... (which means build everything in this directory and all subdirectories) [18:46] ah, looks like I got most of that, except for the go get launchpad.net/godeps [18:47] let's see [18:47] ahh yeah, you need godeps... that sets all the repos juju depends on to the correct revisions [18:48] If you want to run the tests, you'll need to install mongo (yes, it's terrible, I'm sorry) ... but if you just want to build and run, you don't need it. [18:54] well, it worked, but maybe I'm just blind, but how do I actually run it? [18:54] haha [18:54] so the binaries get put in $GOPATH/bin [18:54] juju and jujud [18:55] yea, they didn't end up there... [18:55] oh uh hmm [18:55] maybe I should try it again.. I already had already tried to get juju installed... [18:56] alternatively, from the root of the repo, you can do go build ./cmd/juju and go build ./cmd/jujud and it'll put the binaries in the current directory [19:15] natefinch: could this be a difference between go on trusty vs xenial? (sorry forgot to mention I'm on trusty) [19:26] ahh.. blah.. do I have to manually install go 1.6? [19:29] gQuigs: what does go version say? [19:30] go version go1.2.1 linux/amd64 [19:30] oh yeah, that won't work [19:30] I mean, it might work in 1.25 [19:30] but it really should be 1.6 [19:30] yea, I just installed go1.6 manaully [19:30] will try again [20:11] morning all [20:11] morning thumper [20:27] thanks a billion for your help nate! [20:27] (it's working now) [20:28] thumper: morning tim. here, start the day with a fresh cup of easy review: https://github.com/juju/charm/pull/224 [20:29] katco: shipit [20:29] thumper: ta [20:30] thumper: will have a series of these soon... maybe only 1 more based off the pr you have already reviewed [20:32] gah, is there no way to juju login without it being interactive? [20:40] thumper: do you know about login/logout? This code is a mystery to me... [20:40] as in the CLI? [20:40] no [20:40] from the apiserver, yes [20:41] thumper: yeah. cli.... there's a bug, but I can't even really find where we're prompting for a password... [20:56] heh.. agent-streams: local works :) === natefinch is now known as natefinch-afk [21:04] thumper, I assume you are ditching the bug scrub call? [21:04] yeah [21:04] which is fine by me, means i can find some lunch [21:04] ok [22:01] * thumper recalls a bug about what he is looking at [22:10] Bug #1630029 changed: models should inherit vpc-id from controller [22:14] wallyworld, you around? [22:14] alexisb_: yes, otp [22:14] wallyworld: ohai... thought you were on national holiday. we can still do our 1:1 if you like [22:14] one sec [22:15] wallyworld: welcome back [22:15] katco: ok, one sec [22:18] alexisb_: sorry, free now, did you have a question? [22:19] Bug #1630029 opened: models should inherit vpc-id from controller [22:19] wallyworld, yes, thumper and I do: https://bluejeans.com/5036865018 [22:22] wallyworld: https://bugs.launchpad.net/juju/+bug/1629089 [22:22] Bug #1629089: API Login ACL response values differ from CLI [23:22] axw: lemme know if https://github.com/juju/juju/pull/6321#pullrequestreview-2234618 suits what you asked so I can merge this [23:39] perrito666: LGTM, thanks [23:39] tx axw