[01:05] * redir goes eod === redir is now known as redir_afk [02:30] davecheney: you around? [02:54] Bug #1583422 changed: status-history is broken [03:06] Bug #1583422 opened: status-history is broken [03:08] natefinch: ac [03:08] ack [03:08] davecheney: do you think this bug requires maas? https://bugs.launchpad.net/juju-core/+bug/1537585 [03:08] * natefinch waits for a slow mup [03:09] davecheney: sounds like "we're deploying a lot of machines and containers, and some contention/race in the code is blowing up if we go too fast" [03:15] Bug #1583422 changed: status-history is broken [03:18] natefinch: no idea, never done any maas [03:18] why do we have a concept of connecting to the API without logging in ? [03:18] what operations are permitted without authentication ? [03:19] davecheney: not many, but a few [03:20] that blows [03:20] that means any api client method like ModelUUID has to return an error [03:20] because it plays double duty as a "hey, you aren't logged in yet" [03:20] signal [03:22] davecheney: it's possible we've cleaned everything up since I last looked, but I don't know for sure [03:23] the problem i'm trying to fix is api/service/client.go which reutrns "" if there is an error retrievin the model uuid [03:23] and I cannot change the api like I did for https://github.com/juju/juju/commit/682e48db28d27b8730fce3aab1a63d5260787caf [03:23] because of the horrid romulus mess [03:24] so I tried to store the model tag earlier, but apparently there is logic that expects you to be connected, but not have a model [03:24] ug [03:24] technical debt -- fuck yeah! === meetingology` is now known as meetingology [04:27] Bug #1558734 changed: POWER8 agent stacktraces and refuses to boot [05:03] Bug #1583893 opened: 1.25.5: goroutine panic launching container on xenial

=== mup_ is now known as mup === mup_ is now known as mup [07:40] Bug #1583934 opened: revision specified in the charm, how? === akhavr1 is now known as akhavr [11:18] hmm, do we have terminology to distinguish concisely between "agents running in a hosted model" vs "agents running in a controller model"? e.g. "controller agents" won't really work because not all agents in controller models actually run controller tasks or need controller-level privileges [11:20] the biggest distinction I'm aware of is whether they're migratable -- stuff running in a controller environment isn't -- but it doesn't seem quite fundamental enough [11:27] "hosted" vs "fixed"? [hosted unit agent, hosted machine agent] and [fixed unit agent, fixed machine agent, (implicitly fixed) controller machine agent] [11:27] ? [11:29] it's sort of nice to emphasise immobility of agents in controller models, because they can't even be rehosted... but it would be wrong to imply *mobility* of agents in hosted models, because it's not the *agents* that move, it's that they can switch to being managed by a different host [11:31] english is hard :/ [11:32] if anyone sees the scrollback and has thoughts, they would be appreciated [11:42] wwitzel3: ping [13:59] Bug #1584059 opened: Deployment of swift-storage charms fails with Juju 2.0 - swift-storage-relation-joined KeyError: 'JUJU_ENV_UUID'

[13:59] good morning juju [14:00] katco: good morning katco [14:01] perrito666: feeling any better? [14:02] katco: yep, I transitioned into the bearable but persistent cold status [14:02] which is.... well bearable [14:02] perrito666: well at least that's progress [14:03] yep, I can swallow [14:03] and no longer shivering [14:04] ugh being sick is the worst [14:04] yes it is === mup_ is now known as mup [14:15] fwereade: yt? [14:16] fwereade: also wondering [14:17] you're in demand [14:17] morning katco [14:17] redir: i wonder if he's thread safe [14:17] redir: morning [14:18] :) [14:18] deadlocked [14:18] redir: hopefully livelocked... [14:18] or that [14:18] redir, heyhey! [14:18] katco, redir: I've got him right now :) [14:18] katco, also :) [14:18] doh [14:18] ericsnow: snooooooow! [14:18] redir: we've been talking about your patch :) [14:18] * redir backs away slowly [14:19] haha [14:19] in moonstone? [14:19] redir, no worries, sorry I wasn't available before [14:19] np [14:19] fwereade: you have time to chat with redir and me right now? [14:19] redir, ericsnow, sure, I'll join you there [14:20] gimme 2 minutes [14:20] moonstone? [14:50] redir: ericsnow: natefinch: we can defer the standup. this conversation with william is more important for now [14:50] katco: ack [14:52] ok [14:59] natefinch: we're in moonstone if you're interested. basically pair programming [15:01] katco: cool, I'll pop in [15:55] dooferlad: ping? [15:57] katco: he emailed this morning saying he was still ill [15:57] voidspace: ah sorry to hear that :( do you think it's OK if natefinch picks up bug 1537585 ? [15:57] Bug #1537585: machine agent failed to register IP addresses, borks agent <2.0-count>

[15:57] wow, there is some virus affecting canonicalers [15:58] katco: I don't believe he's started on it, so it should be fine [15:59] voidspace: ok ta [16:00] natefinch: can you ping someone in landscape to understand where they need the fix first? 1.25 or 2.0-beta8? [16:01] katco: yep [16:03] natefinch: ta [16:06] fwereade: so you wanted two more tests there? [16:06] in aggregate [16:12] redir, I think we want tests that: [16:12] redir, one request gets sent after suitable delay (done) [16:13] redir, several requests in a short space of time get batched (done) [16:13] fwereade, that advancing delay-time.Nanosecond and then killing causes all pending reqs to fail [16:13] redir, sorry^ [16:13] brb [16:13] np [16:16] redir, and, that having sent/advanced/received one batch, you can send/advance/receive again and that works too [16:17] redir, (and, in all above cases, checking that you've made the ListInstances calls you expect, once you've killed the worker) [16:17] redir, also nice to do a test where ListInstances errors, and one where it returns some results + ErrPartialInstances [16:18] fwereade: ty for taking the time to help redir [16:18] katco, always a pleasure, seriously [16:18] katco, it's one of the most useful things I can do [16:18] katco, complain about concurrency horrors to me early, I will enjoy myself :) [16:18] gtg [16:32] tx fwereade [16:32] * fwereade back briefly [16:32] redir, yw [16:33] redir, (would you also separate the configs please? and test those too :)) [16:33] fwereade: affirmative [16:33] redir, cheers [17:34] bbiab [17:36] Bbl [17:56] ericsnow: hey got a bit to chat? [17:56] katco: sure [18:45] Bug #1584170 opened: RFE - juju status should show all services inside it [19:03] oh, yeah. back. [20:03] Bug #1584193 opened: juju deploy is in a different from than jujucharms.com [20:19] sinzui: I'm trying to repo a bug, which might require maas. Does QA have a maas I could borrow? [20:20] natefinch: you probably want the 1.9 which you cna use, but is often is heavy use [20:21] sinzui: yeah, 1.9 is what it was reported on [20:21] natefinch: I will add your keys to the munna host [20:21] sinzui: I just need like 2-3 machines that I can spam a bunch of containers onto [20:21] natefinch: how many machines do you need? [20:22] * natefinch points up one line ;) [20:22] natefinch: sorry get with bac and or jrwren for guimaas access and see if it works for you [20:23] rick_h_: awesome, thanks :) [20:25] sinzui: I may already be on munna [20:27] natefinch: you weren't, but you are now. I am sending you some ssh comfig for munna. maybe you used cloud-city in the past. that will still work [20:28] sinzui: yes, I think so. Thanks. [20:30] redir: FYI, I've got to take off a bit early [20:30] natefinch: check your email [20:36] Bug #1584212 opened: juju 2.0 doesn't allow to remove a cloud (juju remove-cloud) [20:48] no sweat ericsnow [22:26] if anyone's around, http://reviews.vapour.ws/r/4873/ is biggish but trivial (just moves + one test fix) and I'd love to get it on its way into master this weekend [22:27] (probably better reviewed on github given RB's usual confusion re moves) ^^