[00:18] veebers: what do we need to do to add another repo into the bot landing? [00:24] thumper: as in a dep? [00:25] veebers: no, as in github.com/juju/description [00:26] thumper: I would have to find out [00:29] thumper: I believe we would need a new jenkins job and to update some scripts/cron for that. I would have to dig further to get a full answer [00:29] thumper: Re: lp:1669046 - we have tickets in already to update all of our standalone controllers to 2.1.0. [00:29] I'll create one shortly to get our HA controller env upgraded, but it may take a little longer - are there any special precautions/procedures for an HA upgrade? [00:35] thumper: Also, if you need anything more on that 1.x to 2.x upgrade preparation, a reply to the ticket I created would be appreciated; gives me a place to log time worked on it. [00:40] blahdeblah: ack [00:50] hmm... [00:50] sublime not doing well with this global rename [00:51] doing save all on 5000 different go files causes it's plugins to work overtime [00:51] * thumper twiddles fingers waiting [00:53] oh bollocks [00:54] babbageclunk: i just noticed we have remoteApplicationsC in the ignored collections in migration_internal_test.go [00:54] managed to somehow end up with submodules [00:55] wallyworld: oh, whoops - I'll fix that [00:56] babbageclunk: no wuckers. i noticed because i'm about to delete an entire collection and associated data model we won't be using anymore [00:57] no rush on it - get the spaces worker stuff done so i can test it out :-) [00:57] wallyworld: ok - still trying to track down what I've broken. [00:57] joy [00:58] 11000 changes across 2800 files [00:58] yeah... do it [01:00] wallyworld: I mean, I've worked out the line that's done it, but I don't understand why. Trying it again now [01:00] let me know if you need a 2nd set of eyes [01:01] wallyworld: will do, thanks [01:06] thumper: So, re: the precautions/procedures for HA upgrade question? [01:10] wallyworld: sorry, https://github.com/juju/juju/pull/7047 should have been closed. I have another branch which supersedes it [01:11] no worries, i thought that was the case but then i second guessed myself and reviewed anyway. but yeah, it didn't seem to match the latest discussion [01:11] wallyworld: ok, worked it out - the connection's dropping because of a panic - nil pointer dereference. :( fixing! [01:11] i hate those [01:11] babbageclunk: what that in a unit test? [01:12] that's why i added the stacktrace to catacomg errors [01:12] cause i had a similar issue that was hard to diagnse where it was happening [01:12] wallyworld: well actually it's the same branch but updated. pushing now, will address comments and ping you later [01:12] np [01:16] bleh nope, I'm wrong. closing and will create a new one [01:22] blahdeblah: shouldn't be any different [01:23] thumper: Cool - so we don't need to upgrade each node separately or anything like that? [01:24] nope [01:26] sweet [01:26] thanks [01:39] wallyworld: https://github.com/juju/juju/pull/7066. I'm running QA now [01:39] ok, otp with tim [01:54] wallyworld: Oops, ListSubnets doesn't like blank space names either - I think there are going to be a few of these. [01:54] yeah [02:11] did something in 2.2 change for Azure? I see a _bunch_ of test failures in azure with errors like "'Failed': Code="AuthenticationFailed" Message="Authentication failed."" [02:15] veebers: Sounds like authentication failed. [02:15] :) [02:16] babbageclunk: ah an interesting take on an error message :-\ [02:16] veebers: In non-Friday-afternoon-japes mode, not that I know of. [02:17] babbageclunk: thing is this is happening during a bootstrap [02:17] babbageclunk: ack. I still need to pack for the sprint, leaving the house early tomorrow :-*( [02:18] veebers: ouch, I'm not leaving until Sunday afternoon. [02:19] babbageclunk: ah, and arriving Sun evening right? I'm hoping that arriving Sat evening I lessen jetlag and hit the ground running on Monday [02:19] veebers: yeah, shake off some of that jetlag on Bourbon St, I hear you. [02:19] ;) [02:20] babbageclunk: ^_^ [02:22] wallyworld: ugh, segfault whackamole [02:24] wallyworld: actually, this one isn't a segfault, just the subnets command (calling ListSubnets which I just fixed) now complains about blank spaces. [02:33] wallyworld: rather than giving subnets a blank space, maybe we should be giving them a special space name? I feel like we're going to encounter knock-on effects from this. [02:34] wallyworld: Although I'm not sure about what the special space name would be. [02:34] babbageclunk: sorry have been otp till now [02:35] wallyworld: no worries, mostly thinking out loud. [02:35] babbageclunk: IIANM it is the intent to use "" as the default space rather than a const "default" or whatever [02:35] we will just have to solve the whack a mole sadly [02:36] babbageclunk: i need food, will be biab [02:36] wallyworld: ok, lucky I hate those pesky moles. [02:47] wallyworld: given that ListSubnets might now return a blank space tag, should I make a new facade version? [03:09] wallyworld: review plz? https://github.com/juju/juju/pull/7067 [03:54] babbageclunk: reviewed [04:06] axw: reviewed - IMO we need to include Type in list storage output [04:07] wallyworld: should be obvious from the pool? [04:07] and/or provider ID? [04:08] perhaps, but not guaranteed to be obvious? the pool name might not reflect the type so well? same with provider id which could well be opaque [04:08] pool names can be defined by the user [04:08] wallyworld: true, but I would think the operator would know what they've set up [04:08] but it is not the operator who always runs list-storage [04:09] wallyworld: I'm reluctant because it's going to add a fair bit to the width [04:09] think of a support person or other user who did not do the original deploy [04:09] the width of the table? [04:09] wallyworld: yeah [04:09] juju status is not exactly un-wide [04:09] wallyworld: indeed, but one of the things we did in 2.0 was to cut it down [04:10] cut down size of instance IDs, etc. [04:10] to avoid overflowing terminal width by default [04:10] sure. but "volume" or "fiesystem" is not that much extra [04:11] wallyworld: actually you can't tell from the pool alone anyway, since you couldh ave a filesystem on EBS or whatever [04:11] true [04:11] I'll look at adding it in [04:11] I do think we need it TBH. ty [04:12] we can always remove if people complain, but i honestly don't think they will [04:51] wallyworld: https://github.com/juju/juju/pull/7066/commits/fa75bb44923f8aafaf70a8c7615d64dafb494ea0 [04:51] looking [04:52] axw: awesome, thanks [04:52] np [06:12] Anyone who's worked on the juju2 user permissions subsystem around? Just want to ask some questions before I log some seemingly rather silly bugs. [06:13] Asking anyway... :-) [06:13] blahdeblah: 2.0.x or 2.1.x? [06:13] there were some changes... [06:13] anastasiamac: 2.1.0 stable! [06:14] blahdeblah: ask away.. there may not b answers that make ur day :) [06:14] So, Q1: is it expected that superusers are exempt from being disabled? (Because in my testing, they are.) [06:14] To me this is rather unexpected. [06:16] blahdeblah: yes. i believe it's so tat someone can destroy and kill [06:16] blahdeblah: maybe that logic needs to be adjusted to not disable if it's the last one [06:17] blahdeblah: we have a discussion at sprint next week about user mngmt... so i can bring it up, especially if it's bugged up :) [06:18] OK, so then the bug I will logging is that when you disable a superuser, it gives no indication that your command did absolutely nothing. [06:18] (although it doesn't do nothing; it does actually mark the user as disabled, and it shows this way in list-users --all) [06:23] anastasiamac: Q2: Is it expected that non-superusers which are disabled and then re-enabled lose all of their grants to various models? To me this is also rather unexpected. [06:24] When I disable a user and then re-enable that same user immediately (say, because I typoed the user name and disabled the wrong one), I don't want to have to work out which models were affected. [06:25] blahdeblah: i *think* this is expected as you want to b explcit about ur grants when re-enabling a user [06:25] blahdeblah: but again, it something to discuss [06:26] No, I don't. I want them to go back exactly as they were. :-) [06:26] k \o/ file it and i'll see what we do [06:26] OK - thanks [06:57] wallyworld: do you recall if CI/builders are using go 1.7 to build already? or still on 1.6? [07:01] axw: still 1.6 AFAIK [07:01] bugger [07:03] i could be wrong but i think it's 1.6 [07:03] is there a 1.7 feature you want to use? [07:03] wallyworld: yeah, http request context [07:03] wallyworld: 1.8 would be better, since that has graceful shutdown/close methods on server [07:04] is 1.8 officially out yet? last i heard it was still rc? [07:05] wallyworld: https://blog.golang.org/go1.8 [07:05] i/m 2 weeks out of date [07:05] wallyworld: also I suspect sort.Slice could be useful in a few parts of the codebase... [07:05] indeed [07:06] we'll have the right people necxt week to discuss [07:06] i'll add it to the agenda [07:07] wallyworld: cool, thanks [07:07] IIANM we can use whatever we like now, and are not dependent on archives [07:09] yeah, i think so too === frankban|afk is now known as frankban [10:02] Bug #1669729 opened: Cannot update machine data

[10:36] any idea why i can't switch to a model that i have never connected to using a new user in juju? [10:36] i can see the model in `juju models` [10:37] but `juju switch aws:default` just spits out 'ERROR "aws:default" is not the name of a model or controller' [10:38] looks like there is some problem at the controller level even [10:38] SimonKLB: is your controller called aws? [10:39] babbageclunk: yea [10:39] added a new user, got the register base64 string, added it to the new user and named the controller aws [10:39] SimonKLB: maybe something's getting confused between that and the aws provider [10:40] ill retry with a more unique name [10:40] SimonKLB: It's a guess, sorry. [10:41] SimonKLB: Could also try passing --debug to `juju switch` [10:41] babbageclunk: do you have to re-do the add-user process, because now the second time i use the register string i get 'ERROR secret key for user "X" not found' [10:42] or can i genenrate the string again without having to do add-user? [10:43] SimonKLB: I don't think you can reuse the register string (I'm not certain of that though) - try adding another user? [10:43] yea will do [10:45] babbageclunk: same thing using "mycontroller" as the name [10:45] nothing stands out in the --debug output [10:46] SimonKLB: you can see the model in `juju models`? [10:46] SimonKLB: did you grant access to the user? [10:47] babbageclunk: yea i granted access [10:47] http://paste.ubuntu.com/24101356/ [10:48] SimonKLB: Try `juju switch mycontroller:admin/default`? [10:48] doh... [10:48] thanks man haha [10:49] SimonKLB: hmm, I don't think the admin/ should be necessary, since just default is not ambiguous. [10:49] any idea why its named admin/default there and not elsewhere? [10:49] well it worked using admin/ but not without :O [10:50] SimonKLB: It's admin/default because it's owned by admin. [10:50] aaah [10:50] make sense [10:51] SimonKLB: It's still pretty confusing though - can you file a bug? Or I will. [10:51] launchpad or github? [10:51] morning [10:54] nvm launchpad it is! [10:54] SimonKLB: yes launchpad please! [10:55] morning perrito666! [10:55] babbageclunk: coming to the sprint? [10:55] perrito666: yup yup, assuming I get into the country [10:55] babbageclunk: yes, we are all under the same assumption, I dont thing the place for the sprint was very wisely chosen :p [10:56] we go to uk after the brexit and now to the US, who is pickig the destinatios? [10:57] perrito666: political storm-chasing [10:57] heh [11:06] babbageclunk: https://bugs.launchpad.net/juju/+bug/1669745 [11:06] Bug #1669745: Model name different for model owner and non-owner might confuse new users [11:07] SimonKLB: Nice write-up, thanks! I've tagged it with usability. === frankban is now known as frankban|afk === marcoceppi_ is now known as marcoceppi === frankban|afk is now known as frankban [17:15] if anyone cares to take a look, i've just put a PR up for review of a change to gopkg.in/juju/worker.v1. my aim is to use it to replace a bunch of the state/workers.RestartWorker stuff: https://github.com/juju/worker/pull/2 === rogpeppe1 is now known as rogpeppe === ejat_ is now known as ejat === Dmitrii-Sh_ is now known as Dmitrii-Sh === frankban is now known as frankban|afk [21:20] i saw perrito666 new ride today: https://goo.gl/NDtqUN