[00:28] menn0: do you have a link to the release notes? [00:29] wallyworld: link is in canonical #juju channel [00:29] ta [00:35] * babbageclunk lunches [00:35] * babbageclunk runs actually [01:41] * redir eod [01:54] wallyworld: mind if we hold off on 1:1 till quarter past? I'd like to fix this LXD bug ASAP [01:55] yep no problem, just ping when ready, no rush [02:12] wallyworld: would you please review https://github.com/juju/juju/pull/6912 [02:12] sure [02:16] axw: in TestFinalizeCredentialLocalAddCertAlreadyThere, i can't see where "out" is used [02:17] could we check its contents like in the similar test [02:17] wallyworld: yeah, I deleted it after running the test.. heh [02:17] wallyworld: ok [02:17] i'd prefer we keep in and just do a check [02:18] wallyworld: yep, will do [02:18] lgtm aprt from that [02:18] wallyworld: pushed [02:18] wallyworld: thanks [02:19] axw: a message to balloons to hold the release would be useful. he may even see this ping [02:20] I see jam is landing things too [02:20] so yea :-) [02:20] balloons: well, i merged for him :-) [02:20] ahh [02:20] go for it [02:20] be busy, and make beta5 as shiny as it can be' [02:20] great, agreed :-) [02:20] thanks [02:23] wallyworld: 1:1 now? [02:23] sure [02:36] wallyworld, axw, anastasiamac: I forgot to mention in the meeting, it's Waitangi Day here on Monday, so the kiwis won't be around. [02:36] babbageclunk: ok, thanks [02:36] babbageclunk: bloody kiwi bludgers [02:38] wallyworld: axw: shall we skip standup on monday then? [02:39] anastasiamac: don't you want to talkto me and andrew? :-( [02:39] wallyworld: m always happy to tlk to andrew \o/ [02:40] :-( :-( :-( :-( :-( [02:41] wallyworld: that's a lot of faces u r wearing.. masks? :D [02:41] it's me crying [02:42] u and me both [02:46] anastasiamac: lol, I think it's because you only want to talk to Andrew not Ian ^_^ [02:48] * wallyworld sobs quietly [02:50] * anastasiamac hands tissues virtually [02:58] redir: ping? [02:59] redir: Entertainingly a bit of live testing shows that the PUT version doesn't do what I want (partial update), while the POST does. [03:00] redir: directly counter to the docs. [03:33] wallyworld, axw: my connection just dropped, so not sure whether anyone saw my last message: could someone review https://github.com/go-goose/goose/pull/39 [03:34] Please? [03:34] yeah, connection dropped [03:34] babbageclunk: will do [03:34] Thanks! [03:35] I need to duck out soon to take my sister to the airport but I'll be around later on. [03:39] babbageclunk: what openstacks have you QAd on? I'm curious about the discrepancy between docs and reality [03:40] axw: yeah, I thought that was weird too. I've only tried on canonistack. [03:40] babbageclunk: ok. would be good to test on rackspace also [03:41] axw: good call, I'll do that now [03:46] axw: rackspace behaves the same, although my test fails because they put some other stuff into the metadata. Fixing that now. [03:47] wallyworld: do you think it would be reasonable to add a new "introspection" access level? to enable prometheus, you would add a user for it with introspection access; and then other users could be granted that access also, and they can hit up the pprof endpoints without having full access [03:47] babbageclunk: okey dokey [03:47] wallyworld: otherwise we'll need to add separate auth for introspection, with a special user/pass [03:48] axw: so introspection < read < write < admin? [03:48] wallyworld: there's no read on controller [03:48] wallyworld: login < introspection < superuser [03:48] ah controller, right [03:48] wallyworld: hmm, but add-model ... [03:48] sigh [03:48] single dimension access does not work [03:49] yeah [03:49] we could chamge that [03:49] any of login, add-model could have introspection added [03:49] superuser implies it [03:49] wallyworld: or we could just have introspection imply add-model. you still have to have creds [03:50] wallyworld: so, login < add-model < introspection < superuser [03:50] not a fan that [03:50] because [03:50] why would you *need* to allow someone to add-model just because you want them to be able to introspect [03:51] wallyworld: the permissions model just does not allow for multiple dimensions. what's the alternative? [03:51] wallyworld: (I agree that it *shouldn't imply add-model) [03:56] wallyworld: I guess another option is that you have *either* introspection *or* add-model [03:56] wallyworld: kinda crappy, means you can't grant introspection to someone who can add models [03:56] wallyworld: but maybe the least crappy option for now? [04:02] wallyworld: bleh, but of course everything relies on them being in a single dimension for comparison [04:06] wallyworld: new idea: support adding access for application tags, and introduce a conceptual "juju-controller" application. having "read" access on that would give you the ability to introspect it [04:07] wallyworld: that goes in the direction of modelling juju in juju [04:15] Is thumper not around today? [04:15] blahdeblah: on holidays [04:15] OK - ta [04:16] blahdeblah: back next week, but maybe not monday (NZ public holiday?) [04:27] I've got a reoccurrence of a bug he was helping me with, but the bug as described doesn't match what the symptoms were when we looked at it. [04:28] :( [04:28] The bug is https://bugs.launchpad.net/juju-core/+bug/1484105, but the problem we saw at the time was that there was mongo corruption which prevented charms from reporting status correctly. [04:28] Bug #1484105: juju upgrade-charm returns ERROR state changing too quickly; try again soon

[04:29] So what happens is the charm gets upgraded on disk, but the results don't show up in juju status as upgraded. [04:30] So I'm not sure whether I should log a new bug about this, or just update that one, even though the symptoms don't match. :-\ [04:31] blahdeblah: maybe a new bug? a link the old one to it to describe how u think they r related?... [04:33] Is there much point opening a new bug on 1.x? It makes juju fairly broken, but we have a reasonably clear procedure for fixing it, so it's not likely to get much traction... [04:35] blahdeblah: true. add new info to existing bug... m not even sure if we have capacity to dig into any 1.x at this stage [04:36] blahdeblah: which 1.x is it? [04:36] 1.25.9 [05:27] axw: thanks for review, was good outcome [05:27] wallyworld: np [05:27] wallyworld: re my earlier spam. I think I've settled on supporting user being superuser on controller, or having read access to the controller model [05:27] wallyworld: sound reasonable? [05:28] yeah, i think so. if you can read model you should be ok to introspect [05:28] we can ask for "guidance" [05:28] but keep it simple firs tup [05:28] wallyworld: I'll just email the list once as a heads up, guidance can be given later [05:29] yep, that's what i meant sorry [05:29] cool [06:23] wallyworld: if you have time, please review https://github.com/juju/juju/pull/6913 [06:23] sure [06:24] wallyworld: do you know if it's intentional that when you use "controller-config" or "model-config" with a single key, that multi-line strings get formatted as YAML ? [06:24] um, i *think* so, I *think* tim did it? [06:24] recently? [06:27] axw: just started looking - don't we already have a base auth http handler? [06:30] wallyworld: I don't think so? everything I've looked at uses httpContext [06:32] wallyworld: re controller-config, seems a bit crappy that "juju controller-config ca-cert" doesn't just give me the ca-cert unadorned. it comes out with YAML formatting [06:32] hmmm, we used to have an authContext of some sort. but now we have httpContext with func (ctxt *httpContext) stateForRequestAuthenticated(r *http.Request) [06:33] agree about the cert config, should ask tim next week [06:33] wallyworld: Server has an authContext, httpContext has a Server [06:34] ok, np. seems there's been a bit of refactoring since i last looked at that code [06:35] i have to do school pickup, bbiab [06:36] axw: wallyworld: single-key feels like we should be doing a raw-dump of the output, *not* formatted [06:36] its intended that you can do "FOO=`juju config foo`" [06:36] well probably FOO=$(juju config foo) [06:37] jam: yeah, that's what I thought. was surprised when it didn't work [06:38] axw: sounds like a bug to me [06:38] jam: yeah, I'll file one [07:26] axw: that PR LGTM. Will be a nice enahncement [07:27] wallyworld: thanks === frankban|afk is now known as frankban [11:30] Bug #1661576 opened: show-status-log formatting contains unwanted pad lines === mskalka|afk is now known as mskalka [14:42] Bug #1661624 opened: show-status-log bogus timeline [16:52] Bug #1661681 opened: Broken agent complaints about tomb: dying [16:58] morning juju-dev [16:58] redir: morning redir [16:58] I thought you where out today [16:59] fo a bit in a couple hours but not all day perrito666 [17:42] redir, you about? [17:42] yup [17:42] balloons: [17:42] redir, I kind of want to land this for beta5 to fix the packaging issues: https://github.com/juju/juju/pull/6915 [17:43] look at whose being crazy now [17:43] redir, the only real change is to allow 2.* and not 2.0. The rest is just renames for sanity === frankban is now known as frankban|afk [17:50] balloons: shipit [17:52] redir, :-) [18:27] redir, while you're at it, bump the version PR: https://github.com/juju/juju/pull/6914 [18:28] balloons: looking [18:28] balloons: lgtm [20:34] k EOD, ill be online if annyone needs me === frankban is now known as frankban|afk