[00:53] review would be appreciated. small apiserver addition https://github.com/juju/juju/pull/909 === psivaa_ is now known as psivaa === meetingology` is now known as meetingology === Guest97591 is now known as Guest20083 === allenap_ is now known as allenap [01:35] axw: o/ [01:36] thumper: \o [01:36] glad to see you're alive after friday night [01:36] heh [01:36] I should learn to pace myself better [01:37] but I do like mojitos [01:37] :) they weren't bad at all [01:38] $ juju user list --show-disabled [01:38] NAME DISPLAY NAME DATE CREATED LAST CONNECTION [01:38] admin admin 33 minutes ago just now [01:38] foo Foo Bar 20 seconds ago not connected yet (disabled) [01:38] thumper Tim Penhey 31 minutes ago not connected yet [01:38] getting there... [01:39] axw: I don't suppose you could look at https://bugs.launchpad.net/juju-core/+bug/1380337 could you? [01:39] Bug #1380337: adding machine after destroying another fails

[01:39] looking === alpacaherder is now known as skellat === psivaa_ is now known as psivaa [01:51] thumper: axw: i've already looked into that bug - it only affects 1.20, and it a dup [01:51] wallyworld_: oh hai [01:52] wallyworld_: okey dokey. do you want me to continue looking, or you know the root cause already? [01:52] o/ [01:52] axw: i have an idea, just needs to be fixed [01:53] wallyworld_: oh, is this the tools bug you mentioned on the hangout? [01:53] yeah [01:53] ok [01:53] axw: what's the status of bug 1360605 which is marked as In Progress [01:53] Bug #1360605: support maas zones for automatic az placement

[01:54] wallyworld_: needs review of a gomaasapi branch [01:54] then it should be good to merge [01:54] oh yeah, that's right [01:54] i'll do that [01:54] well that's confusing [01:55] time.Time.Add takes a duration and returns a time [01:55] time.Time.Sub takes a time and returns a duration [01:55] awesome [01:58] WT actual F [01:58] ./user_info_test.go:144: constant -2.1 truncated to integer [01:58] I have to put a f there? [02:01] stabby stabby [02:01] -2.1e1 is fine [02:01] -2.1e0 gets turned into an int [02:01] wtf [02:03] ah... [02:03] used with int64 and * [02:05] FFS!!!! [02:05] debug-log is broken with the local provider again since all-machines.log doesn't exist === Guest20083 is now known as Ursinha [02:09] thumper: -2.1e1 is an integer, isn't it? given that constant calculations have infinite precision [02:10] mwhudson: according to the language spec it is a float constant, [02:10] mwhudson: the issue here is that it was being used in a compile time multiplication of two constants [02:10] so the float was being co-erced into an int === pjdc_ is now known as pjdc [02:26] axw: i think bug 1356886 can be closed now, right? [02:26] Bug #1356886: failed add-machine ssh: leaves behind garbage in state <14.10>

[02:27] wallyworld_: kinda. it will clean up eventually now. it would be nice if we removed it immediately, but it's not that simple [02:27] I think we can close it [02:27] wallyworld_: what was the lxc issue? [02:28] thumper: looks like destroy-machine removes stuff under /storage which includes tools required to start new machines [02:29] thumper: that debug-log issue sounds like a stop the line regression doesn't it [02:29] wallyworld_: hmm (re: lxc), and I'd say so w.r.t. debug-log [02:30] wallyworld_: not sure when it stopped working [02:30] been a while since I've had a local environment running [02:30] where I cared about the lgs [02:30] logs [02:30] wayne needs to fix it, since i think he may have broken it [02:30] we can stop the line later when he comes on :-) [02:56] haha [03:14] wallyworld_: PTAL https://code.launchpad.net/~axwalk/gomaasapi/maas-testserver-zones/+merge/236447 [03:14] sure [03:18] axw: what's the go with the agent_name handling? [03:18] wallyworld_: replied to your comment inline [03:18] hmmm, didn't see the reply, let me look again [03:19] hrm, doesn't show up after I pushed again.. one sec [03:19] wallyworld_: Correct. From https://maas.ubuntu.com/docs1.5/api.html#nodes: [03:19] "param agent_name: [03:19] An optional agent name to attach to the acquired node." [03:19] (under the "acquire" operation notes; for "list" it's a filter value) [03:19] (in reply to your question about agent_name being a side effect) [03:20] ah, there's a link you need to click to show the diff comments [03:20] where's that? [03:20] ah, I see [03:20] hmm, not sure i like that api design, but it is what it is [03:22] axw: +1, land away [03:22] wallyworld_: thanks [03:22] np [03:44] * thumper sighs deeply [03:44] wallyworld_: take a look in #juju [03:44] ok [03:44] wallyworld_: you'd think this was tested right? [03:51] thumper: i would have thought so, but i haven't been involved with HA so much so am not sure [03:52] HA has been problematic at times though I seem to recall === uru_ is now known as urulama [06:35] morning all [06:35] morning dimitern [06:36] hey, jam! [07:08] wallyworld_, you still around? [07:08] alexisb: hi [07:09] hey there wallyworld_ [07:09] having fun? [07:10] wallyworld_, totally! [07:10] ;) [07:39] fwereade: hey you about? [07:39] waigani, heyhey [07:40] fwereade: hello, could I talk something over with you: watchers + localID/docID [07:40] waigani, sure, can I have a quick ciggie while you start a hangout please? [07:40] waigani, or I could do irc if you'd rather [07:41] fwereade: for sure, I'll start a hangout [07:41] fwereade: those things will kill you btw [07:44] fwereade: https://plus.google.com/hangouts/_/g2voow4a3pvpvu6jwsnyrqlwv4a?hl=en [07:48] fwereade: my computer is spazzing out, hang on [07:49] waigani, np, I'll stay in that one, just say if you need to start a new one [07:49] fwereade: https://plus.google.com/hangouts/_/g2wu75d5xxebxhb3pw6jwusynya?hl=en [08:21] morning all [08:48] fwereade: can we have another quick chat please? [08:49] fwereade: https://plus.google.com/hangouts/_/gqitbaigxm45nkusctyi7lh5uya?hl=en [08:55] waigani, sorry, there now [08:55] waigani, shall I start a new one? [08:55] * fwereade does [09:00] * fwereade bbs === uru_ is now known as urulama [10:34] ok, off to a baby scan - hopefully on my return we know if the baby is a boy or a girl [10:34] biab [10:39] enable-os-upgrade: false is one of the greatest kept secrets of 1.21 === luca__ is now known as luca [12:56] Can I beg someone to give some love^Wreviewtime to https://github.com/juju/juju/pull/689 ? [13:03] hmmm.... screwed my network onfiguration :-) [13:03] *configuration [13:04] so, on my laptop for the moment... [13:04] the joys of developing networking for juju? [13:04] w7z: yep :-) [13:04] w7z: created a bridge for MaaS and it worked fine until I rebooted and lost all network [13:04] w7z: I thought it was going a bit too well... [13:05] >_< [13:07] and back... [13:11] wesleymason: the branch looks fine, I'd nitpick the error message but don't want to delay you further [13:11] alexisb: yep [13:12] mfoord: a result of the scan? [13:16] TheMue: boy :-) [13:18] mfoord: yay, grats. am I right, your second one, isn't it? [13:18] TheMue: that's right - first one is a girl [13:18] so we'll have one of each colour :-) [13:19] mfoord: hehe, yeah, well done [14:32] dimitern: take a look here: http://paste.ubuntu.com/8565153/ [14:32] dimitern: the action guys get this error [14:33] it's when running ./worker/firewaller tests on master [14:39] TheMue, looking [14:40] dimitern: thanks [14:41] TheMue, gah... it seems like a timeout issue - i.e. the worker stops *before* entering the loop() and returning the error.. I'll have a look and propose a fix [14:41] jcw4, is it readily reproducible? [14:41] dimitern: for me it seems to be every time [14:42] jcw4, sorry about the trouble, but that's actually great to test the fix :) [14:42] dimitern: I can pull a branch from you to test any fixes if you like [14:42] jcw4, will you be around for ~1h more? [14:42] maybe less even [14:42] yep; beginning of my day [14:43] jcw4, ah, sweet, will ping you when I propose the branch [14:43] cool === tasdomas` is now known as tasdomas === jcw4 is now known as jcw4|afk [15:20] menn0: ha, we just finished the same review [15:20] menn0: poor thumper now has 37 comments on his PR :p [15:20] katco: whoops [15:20] katco: hopefully we didn't overlap too much [15:21] menn0: hehe i'm sure we did on things like docstrings [15:21] menn0: well that'll teach him :) [15:21] katco: I'm currently reviewing bogdanteleaga's PRs (on Github) so let's not both do those [15:21] katco: we should probably co-ordinate for the rest of the day as well :) [15:22] menn0: lol ok [15:22] katco: it's not normally a problem because our time zones don't overlap much but I'm working from the London office today [15:22] menn0: it looks like review board is taken care of [15:22] menn0: ah i see [15:23] menn0: my bad. I should have thought about this. [15:23] menn0: it's not a big deal. i'm familiarizing myself with code as well, so it's not wasted effort. [15:39] TheMue, jcw4|afk - the fix is ready for review - http://reviews.vapour.ws/r/172/ -- if you can have a look and approve it (other reviewers are welcome as well), I'll merge it tomorrow morning [15:40] * dimitern reached eod [15:40] dimitern: looking === jcw4|afk is now known as jcw4 [16:32] calling it a day everyone, take care all [16:38] can I haz digital ocean provider? https://www.digitalocean.com/community/tutorials/an-introduction-to-droplet-metadata [16:40] there's an unofficial one, fwiw [16:41] https://github.com/kapilt/juju-digitalocean [16:41] uses manual provider sadly === ChanServ changed the topic of #juju-dev to: https://juju.ubuntu.com | On-call reviewer: see calendar | Open critical bugs: 1381626 1381632 [17:11] katco, there are two kinds of unit-tests failures introduced in the last few days, bug 1381626 and bug 1381632. Can you help get developers to look into them? [17:11] Bug #1381626: TestActionFail fails on ppc64el

[17:11] Bug #1381632: firewaller_test.go TestStopsAfterGettingMode fails on utopic

[17:13] sinzui, katco fwiw I'm investigating that too [17:13] jcw4, thank you for helping [17:13] All three of the common errors are related to a permission error [17:14] somewhere between lines 1409 and 1416 of apiserver/uniter/uniter_base.go [17:14] I'm suspecting a Tag issue [17:15] sinzui: well, I feel on the hook since my commit was cited [17:15] :) [17:16] katco: to be clear I'm investigating the first one (ActionFail) right now; I haven't looked at the Firewaller one, but I'm 90% sure that dimitern has a PR up that will fix that one [17:16] jcw4, I don't believe permissions issue are arch related...I don't know why a single arch would start failing. I am very puzzled [17:16] sinzui: agreed [17:18] katco: (sinzui) this was the PR that I think will fix bug 1381632 http://reviews.vapour.ws/r/172/ [17:18] Bug #1381632: firewaller_test.go TestStopsAfterGettingMode fails on utopic

[17:18] I was getting the same error and that PR branch fixed it for me [17:25] TheMue, sinzui since http://reviews.vapour.ws/r/172/ is approved to :shipit: can we $$merge$$ it now, or wait til dimitern does it tomorrow? [17:27] jcw4, the two regressions will block merges until they are fixed [17:28] sinzui: I think this PR *is* a fix for one of those regressions [17:28] so maybe I should $$fixes-LP-1381632$$ it [17:29] jcw4, oh, then anyone add fixes-1381632 ot the merge. You don't need "$$" [17:30] cool [17:54] sorry was eating lunch [17:54] sinzui: it looks like I triggered that merge prematurely - I'm still confident it's the right fix, but dimitern hadn't finished the full test suite check on that branch [17:55] katco: 'sok I'm getting excited about my lunch pretty soon here too [17:55] :) [17:55] lol [17:56] sinzui: just reading through the backlog... it looks like you could use help investigating 1381626? [17:56] katco: yeah - that's the one with my merge implicated, but I don't see a direct correlation [17:57] jcw4: ok i'll start investigating [17:57] ta [18:01] katco, I suspect ppc is vulnerable to failures in that part of the tests, something made the tests go from intermittent to always fail. As this is trusty, I cannot see a correlation with "permission denied" [18:02] sinzui: interesting that it would begin breaking all of the sudden [18:02] sinzui: it must be one of the changes, unless something has changed with the scripts you guys use [18:02] my change that triggered it only renamed the params.ActionItem to params.Action struct [18:02] the structure of the struct is identical, just the name changed [18:03] katco, we have not changed the tests since their last pass. [18:03] sinzui: ok, well i'll keep digging in [18:04] katco, We see warning that imply the machine's clock is behind? [18:04] tar: juju-core_1.21-alpha2/src/gopkg.in/yaml.v1/README.md: time stamp 2014-10-15 17:50:59 is 44.413476891 s in the future [18:05] sinzui: interesting, but i don't know why that would cause an issue as long as the relative time remains stable during testing [18:06] katco, yep [18:06] katco: the source of the error is here: https://github.com/juju/juju/blob/master/apiserver/uniter/uniter_base.go#L1410-L1415 [18:06] I wish I'd made two slightly different errors, or else added logging here [18:07] all three of the Action related test failures come back to that block [18:08] jcw4: thank you. has anyone run a git bisect? [18:08] katco: I can't even repro the failure locally [18:08] jcw4: ah ok. i might have some tricks. please hold (elevator music) [18:08] haha [18:13] katco: one of those tricks doesn't include access to a ppc64 machine does it? [18:14] jcw4: lol no, although i keep wondering if that might be handy since we seem to see these with some regularity [18:14] yeah [18:16] jcw4: fails for me on tip: go test -compiler=gccgo github.com/juju/juju/api/uniter/... [18:16] jcw4: but looks like it might be failing in a different spot [18:16] bugger [18:16] jcw4, thank you for finding the PR, it indeed fixed the utopic tests. [18:16] yay! [18:17] about the last comment, not katco's comment: bugger to that too [18:17] Now I see a compilation error in ppc64el tests. [18:18] :( [18:18] sinzui: do you have handy a commit point where this was working? === ChanServ changed the topic of #juju-dev to: https://juju.ubuntu.com | On-call reviewer: see calendar | Open critical bugs: 1381626 1381671 [18:19] katco, commit 1a7538d6 was the last passed [18:20] sinzui: ty [18:29] nice; I repro the same error with --compiler=gccgo [18:30] * jcw4 goes for long anticipated lunch === jcw4 is now known as jcw4|nomnom [19:20] sinzui: something to think about while i investigate: do we run tests with gccgo to gate merges? === BradCrittenden is now known as bac [19:21] katco, no, but we agreed to do it this cycle: https://docs.google.com/a/canonical.com/document/d/1WMeul2xZNOE1vxjj5Vb8Itl-LsGNZeBQurnv76kseGE/edit#heading=h.vxbalgtua4zd [19:22] ah cool === BradCrittenden is now known as bac [20:53] jcw4|nomnom: hey when you get back, i could use some help debugging your change. i'm not that familiar with facades yet, but it looks like tag parsing might be to blame? not sure. [22:03] wallyworld: are you there yet? [22:03] katco: hi, finishing coffee :-) [22:03] wallyworld: yum :) [22:03] wallyworld: i need to EOD, but didn't want to wait until stand-up to hand this off: https://bugs.launchpad.net/juju-core/+bug/1381626 [22:03] Bug #1381626: TestActionFail fails on ppc64el

[22:03] ok, let me look [22:04] wallyworld: i'm pretty sure someone who knows facades better will be able to hop right to the issue since i've done the leg-work [22:05] katco: ty, will look into it [22:05] wallyworld: hopefully you have it solved b/f standup... i'd love to know what it is [22:06] wallyworld: but anyway, time to make dinner :) cya in a bit [22:06] ok [22:24] katco: sorry === jcw4|nomnom is now known as jcw4 [22:25] I'll keep digging too [22:27] jcw4: thx check out the lp bug, i added what info i had there [22:27] katco: yep, just reading that now. I'll update if I find anything more [22:38] katco: you're probably gone but the issue was actually in the names package [22:59] anyone around to give me a code review on https://github.com/juju/names/pull/29 ? [22:59] its needed to fix a blocking bug in CI [23:26] wallyworld: I figured out that bug that katco handed off to you. Its a serialization issue [23:27] wallyworld: I have a PR thats the first part of the fix for that https://github.com/juju/names/pull/29 [23:27] wallyworld: are you able to review that for me? [23:27] jcw4: great thanks, will do, just otp, will look real soon [23:27] cool, ta [23:54] jcw4: reviewed. i'm surprised we haven't run across this before. i would have thought there would be other cases where we compose from unexported structs. maybe not [23:54] wallyworld: I know [23:55] wallyworld: also other Tags in the names package have unexported fields [23:55] wallyworld: so my guess is they're not being serialized yet, even though we've decided to use Tags in the API instead of strings [23:55] jcw4: fields but not embedded structs [23:56] wallyworld: right. [23:56] stupid gccgo i guess [23:56] hehe [23:56] jcw4: maybe you can add a comment to the code to explain why you needed to export it [23:56] wallyworld: will do