[00:00] ericsnow: thnx for shipit :) [00:00] anastasiamac: I figured it would avoid any confusion :) [00:01] anastasiamac: did you get your dependencies.tsv stuff sorted out? [00:01] ericsnow: it's gr8ly appreciated! i was worried how it horrible it'll look to merge without shiptit! but m loving the echo ;) [00:01] anastasiamac: :) [00:01] ericsnow: yes! ur advice was brilliant [00:02] anastasiamac: good; I was just echoing what others had shown me :) [00:03] ericsnow: :) [00:07] thumper ping [01:01] axw: good point about _ubuntu* : I advocated for changing the name from _nonlinux, but your point about conflating usage and what is actually compiled is good. [01:01] axw: would you keep the name _nonlinux? [01:02] jw4: *shrug* I don't see a reason to change it. why did you want to change it away from that? [01:02] axw: to me that was confusing because it looked like _linux, _darwin, _windows etc. and made me question whether the _non... was a valid build constraint [01:02] I see [01:03] axw: I'm fine with leaving the name, but I would love to see a less ambiguous name [01:03] I think that if someone uses that incorrectly it'll become obvious pretty quick, since having multiple build tags almost always involves having a common entry point [01:04] which would cause an error if the tags actually resolve [01:04] kk, that's reasonable to me [01:06] axw, jw4: thanks for taking a look [01:07] nps [01:07] axw, jw4: that change doesn't pay enough so I've dropped it :) [01:07] ericsnow: cool - thanks for considering it :) [01:08] jw4: the comment in there should suffice [01:09] yep [01:16] wallyworld: I've lost my webcam, trying to fix it now... [01:16] np [01:31] the HUD woke up to tell me I had 10% battery life, then to prove it's point went to 100% cpu and drained my battery [04:17] wallyworld, quick question about juju/names? [04:22] sure [04:28] mattyw: ? [04:28] wallyworld, I reckon UnitService (https://github.com/juju/names/blob/master/unit.go#L54) should return an error rather than panic [04:29] wallyworld, it appears there are number of places in core that don't check IsValidUnit before making this call [04:29] wallyworld, I was going to make the change so that UnitService returned (string, error) but thought I'd check with someone first [04:35] mattyw: sorry, my connection is a bit flaky. what was your question? [04:36] wallyworld, does that sound like a reasonable change - UnitService not panicing - but returning an error if the unit name is invalid [04:39] mattyw: i'd be reluctant to change just that one method when all others panic if the name is invalid. i think the expecation is that names are validated before calling into the package === kadams54_ is now known as kadams54 === kadams54 is now known as kadams54-away [05:23] axw_: got time for a hangout? [05:25] wallyworld: sure, I need to duck down in 10m to get something out of the oven tho [05:25] wallyworld: see you in 1:1 [05:25] ok, let's make it quick [05:38] wallyworld: need to write a bunch of tests, but I should be able to propose a branch soon that connects "juju machine add --disks" to the machine provisioner [05:38] I just got ec2 to provision a machine with extra volumes once again === axw_ is now known as axw [05:56] axw: that is \o/ [06:28] wallyworld: thnx for shipit! [06:28] np :-) [06:28] wallyworld: it's my fastest yet :) [06:29] indeed === liam_ is now known as Guest14772 [11:46] niemeyer, hey, if you have some time I'd appreciate a review on https://github.com/go-amz/amz/pull/12 please [12:21] voidspace: ping [12:21] anastasiamac: poing [12:21] *pong even [12:21] morning everyone [12:21] perrito666: morning [12:21] perrito666: morning :) [12:21] ah, I'm OCR today [12:21] forgot [12:21] voidspace: r u ocr? [12:21] *damn* :-) [12:21] anastasiamac: I believe so... [12:21] voidspace: thnx ;) [12:21] voidspace: could u plz have a look at http://reviews.vapour.ws/r/701/ [12:22] voidspace: it's tiny and i'd love to land it in 1.22 [12:22] voidspace: well, for 1.22 [12:22] anastasiamac: looking now [12:23] voidspace: greatly appreciated!! [12:40] anastasiamac: looks like a nice straightforward improvement [12:40] anastasiamac: LGTM [12:40] when creating a new facade version for uniter, should the new one embed the previous one? [12:41] voidspace: thnx :) [12:45] dimitern: I guess landing an mp for gomaasapi is a manual merge [12:45] voidspace, it is I think [12:46] done [12:50] voidspace, \o/ [12:55] dimitern: as far as I can tell it's not possible to go from node -> nodegroup via the maas api [12:56] dimitern: I'm just double checking the details api to see if it's in that [12:57] node id -> nodegroup uuid [13:02] and details is xml, yummy [13:03] and doesn't include nodegroup [13:04] voidspace, right, that sounds like another bug worth reporting :) [13:04] yeah [13:04] I think they're getting sick of me... [13:05] :D [13:05] it's unavoidable though.. [13:32] voidspace: 've changed the code after ur lgtm [13:32] voidspace: could u plz cast ur eyes over it again? [13:33] voidspace: changed the facade and its test to reflect the fact the each entity in the list could have its annotations... [13:33] voidspace: should be a tivial modiifcation too [13:33] voidspace: modification even :) === kadams54 is now known as kadams54-away [14:18] dimitern: Review is up [14:19] niemeyer, thanks! I'm making the suggested changes now [14:20] dimitern: np! [14:29] niemeyer, how do you feel about renaming v2-dev to v2? [14:29] so we can start using it in juju [14:29] sinzui: o/ [14:30] hi anastasiamac [14:30] sinzui: m trying to land something for 1.22 but keep getting "No space left on device" [14:30] sinzui: what shall i do? [14:30] I will invesitigate [14:31] sinzui: thnx :) [14:32] sinzui: also it's past midnite here but I really want it to land... when r u plannning to take 1.22? (in how many hrs?) :P [14:32] the juju-core-slave has 23G of space. I think the ec2 instance is low on space [14:33] anastasiamac, we cannot take 1.22 until 1.21.0 is created. the feature freeze is artifical [14:33] anastasiamac: sure [14:34] anastasiamac, the ec2 image had 6.5G of free space when the test started [14:35] anastasiamac, is this failure repeatable? [14:36] voidspace: thnx ;-) [14:37] sinzui: i've tried to merge twice and got it both times ;) [14:37] anastasiamac: looking at the changes [14:37] mgz, have you seen cases where the ec2 instance starts with 7.75G free, then the tests fail with no space left on device [14:38] anastasiamac: those changes look good to me [14:38] updated ShipIt [14:38] voidspace: tyvm :) [14:38] sinzui: when will the 1.22 branch be created? [14:40] TheMue, I may create it at my EOD or tomorrow because of CI test lag. I will select the last passing rev [14:41] sinzui: ok, we hope to get one more action PR into it until then, thx [14:51] ericsnow: ping [14:57] wwitzel3: pong [14:57] ericsnow: hi :) [14:59] #join python-uk [14:59] #/oin python-uk [14:59] whoops [15:00] wwitzel3: you saw my commits (formatting and tests)? [15:00] ericsnow: I did [15:01] CI says "Can't take a write lock while out of disk space" [15:01] resolution = ? [15:01] sinzui: tried to merge 3rd time, got the same "No space left on device" [15:01] anastasiamac, :/ [15:01] anastasiamac, as this is ec2 I feel pretty helpless [15:01] sinzui: it's getting l8 for me. I'll try to land it in the morning - about 4 or 5 hrs from now :) [15:02] ericsnow: yeah, looks good [15:02] wwitzel3, perrito666: standup? [15:02] wwitzel3: oh, good :) [15:03] anastasiamac, I have a clue [15:04] sinzui: k :) [15:04] sinzui: m mean I m sure u do :) [15:04] I think the git-merge-juju job explicitly creates a 2G tmpfs [15:04] CI doesn't use that feature, the only this job does [15:05] I will ask if I can increase the artificial constraint [15:05] sinzui: it'd b amazing! [15:26] sinzui, nice catch [15:27] anastasiamac, I am trying your PR again === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [15:45] bodie_, resubmit. anastasiamac passed after the size incread [15:45] increase === rogpeppe2 is now known as rogpeppe [15:48] hazmat: you have any thoughts on what a plugin provider should look like? [15:48] hazmat: I have some and am writing up a spec [15:48] hazmat: I recall that you had done something like this already and wanted to get your thoughts [15:49] thanks sinzui [16:03] ericsnow: providers don't have state do they, so my thinking was that we just take the existing providers and make them into an executable - with the existing api becoming commands [16:03] ericsnow: marshall data in and out [16:04] voidspace: right [16:04] ericsnow: you could prototype it over the weekend... ;-) [16:04] voidspace: haha [16:23] hello, I'm trying to play a bit with juju actions (trunk), but can't fully figure out how to get something working, anyone available to help? basically I always get "failed" as action status, but nothing I can see in the logs [16:30] rogpeppe: ping [16:32] voidspace: pong [16:33] rogpeppe: I've updated a dependency - specifically gomaasapi [16:33] rogpeppe: and I want to use the new version [16:33] rogpeppe: dependencies.tsv for a bzr branch has revision number and some combination of email/date/hash [16:33] where does the weird hash thing come from, seemingly not from bzr [16:34] CONTRIBUTING.md suggests to run godeps -t $(go list github.com/juju/juju/...) > dependencies.tsv [16:34] which changes a lot of things... [16:34] but not the one I want [16:36] voidspace: one mo, i'm in a call [16:36] rogpeppe: no problem [16:36] rogpeppe: I think I've solved it - or at least got godeps to generate the line and I'll revert the other changes [16:37] ericsnow, well done, thank you very very much! [16:37] I will send out a summary before my eod to you and perrito666 [16:37] alexisb: is there a mailing-list for the team leads? [16:37] alexisb: glad to do it [16:38] :) [16:38] ericsnow, no, just my local one :) [16:38] :) [16:38] alexisb, ericsnow tx [16:38] I will send you the address [16:38] alexisb: k [16:39] alexisb: I expect the lib will be fine (looks to be officially supported and uses the apache v2 license) [16:39] voidspace: the approach i usually use is: godeps -u dependencies.tsv [16:39] voidspace: then update the dependency you want [16:40] voidspace: then godeps -t ./... > dependencies.tsv [16:40] voidspace: except that currently doesn't work (annoyingly) in juju because of platform-specific deps [16:40] GOOS=windows godeps -t ./... > dependencies.tsv [16:40] and make sure your godeps is up to date :) [16:40] ericsnow: does that get linux-only deps? [16:40] ericsnow: (perhaps there aren't any...) [16:40] rogpeppe: I don't know [16:41] rogpeppe: I believe we only have windows-only deps [16:41] i really need to sort out that godeps issue, but i haven't any allocated time for it... [16:41] nate would know better [16:41] rogpeppe: right, thanks [16:42] voidspace: i quite often do the cut&paste thing too, just to update a single line of dependency [16:42] rogpeppe: with newer godeps does that work right (due to the timestamp)? [16:42] ericsnow: it should be fine [16:43] rogpeppe: good to know [16:43] ericsnow: the time stamp is a deterministic function of the dependency === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [17:10] does somebody know how action parameters are supposed to be accessed by the executable performing the action? some via juju env command? [17:13] wwitzel3: ready? [17:24] ericsnow: yep [17:26] wwitzel3: I'm in moonstone [17:29] katco: ping [17:29] cherylj: howdy! [17:29] katco: do you have a few minutes? [17:29] cherylj: sure, let me just grab a drink rq [17:31] dimitern: hah, we had a maas provider test depending on the bug in the TestServer that I fixed [17:31] dimitern: testing the fallback when it can't query architectures [17:31] voidspace, really? :) [17:31] dimitern: it relied on getNodegroups erroring rather than returning an empty list [17:32] voidspace, good catch! [17:32] dimitern: so I've changed the production code to treat an empty list of boot images to be the same as an error [17:32] and use the fallback for supported architectures [17:32] dimitern: I don't think it matters in practise [17:33] voidspace, I'd have a look when you propose it [17:33] dimitern: sure [17:33] dimitern: it's easy to see what I mean [17:33] dimitern: nearly ready for proposal [17:33] cherylj: ok sorry, what's up? [17:33] voidspace, cheers [17:33] katco, I had some questions for you. Want to do a hangout? [17:34] cherylj: sure === kadams54 is now known as kadams54-away === urulama is now known as urulama_ [18:18] hey can someone add cherylj to the juju team on github? [18:19] alexisb: ^ [18:19] katco, let me see if I have admin to do that one sec [18:19] alexisb: i think you are in the owners group, so i think you can :) [18:24] cherylj, what is you gitbub user name? [18:24] cherylj [18:24] alexisb: cherylj [18:26] cherylj, katco sent an invite let me know if that is what you need [18:28] alexisb, got it! Seems to have worked [18:29] coolness :) [18:29] thanks for forcing me to learn something new [18:29] :) [18:38] thanks again, katco for helping me make my first (trivial) change! [18:38] cherylj: congrats :D [18:39] cherylj: you're now an official juju hacker :) [18:39] yay! :D [18:39] cherylj: oh i forgot to mention: the review board ticket will now be closed as well. ericsnow's integration does that for us as well. [18:39] cherylj: you can always get back to it by looking at your "Outgoing > All" header there [18:39] cherylj: \o/ [18:39] katco: cool, thanks! [18:40] cherylj: well, it won't be closed until the juju bot closes your PR [18:40] ok, time for nom nom nom... bbiab [18:41] right [18:41] happy weekend everyone [18:42] EOW === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === urulama_ is now known as urulama [19:58] What kicks off the mongodb process for the local provider? is this an upstart job that i'm failing to find or is this controlled by the juju bootstrap sequence? [19:59] sinzui: hey this looks very strange to me: http://juju-ci.vapour.ws:8080/job/github-merge-juju/1767/consoleFull [19:59] sinzui: is something the matter with the environment maybe? [20:00] lazyPower: i think it's the bootstrap process? i don't think it's configured as a service [20:01] katco: 10-4, we're running down the issues with kicking off a juju env in a docker container - and this is where we're stuck [20:01] mongod isn't starting - so now we're looking for the mongo logs to figure out why thats the case - the $JUJU_HOME/local isn't being created [20:01] thanks for the feedback tho - *hattip* [20:02] katco: that one looks like the normal mongo falling over with bad record MAC then everything else being seriously ill afterwards [20:02] lazyPower: ah i think i ran into this like 7mo ago, and docker containers didn't have the infrastructure to start... services. so mayb it is a service? [20:02] mgz: so flakey test? i hadn't seen this one yet (i don't think...) [20:02] lazyPower: running local env inside docker seems... perverse [20:03] katco: yeah, tell her to retry it [20:03] lazyPower: i just remember it having to do with how docker containers interacted with linux's init system (upstart or possibly something in /dev) [20:03] mgz: but it has its merits in terms of handing someone a docker container to get them looking at juju. [20:03] mgz: will do, thank you (and happy new year btw!) [20:03] :) [20:04] mgz: when we get this figured out, we'd love your feedback on the project [20:04] +1 or -1 alike - we're in this for the community [20:04] lazyPower: and ideally our local setup would be less of a special unicorn and not have a magic machine 0, but that's where we're at currently [20:05] mgz: i knowwwwww [20:05] mgz: we want a container for bootstrap so bad here in ecosystems, but well be patient [20:05] *we'll [20:05] we know there are bigger fish to fry [20:06] lazyPower: fwiw i think core wants that pretty badly too [20:06] but as you said it's not high on the list [20:09] yeah, I try not to fishbowl too much [20:09] but i'm guilty of it [20:27] katco, lazyPower : $JUJU_HOME/local is created. it just gets cleaned up when the bootstrap process fails [20:27] ahhh [20:27] good 2 know, is there a way to freeze it from getting reaped? [20:27] s/reaped/cleaned up/ [20:28] what we need is a list of dependencies that jujud, mongod and friends expect for local bootstrap [20:28] lazyPower, --keep-broken [20:28] niiiiice [20:29] cherylj: resubmit again [20:29] katco: okay :) [20:30] * perrito666 changes locations to get freshness [20:45] ok, apparently 37ºC is not something that can be dodged by going to the 0th floor [20:49] katco: Success! Huzzah! [20:50] cherylj: lol yaaay [20:50] 3rd time's a charm === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === ChanServ changed the topic of #juju-dev to: https://juju.ubuntu.com | On-call reviewer: see calendar | Open critical bugs: 1409141 [22:01] anastasiamac, I believe a test introduced in one of your revisions is bad, not your code, but your test. I reported bug 1409141 about ppc test failure [22:01] Bug #1409141: TestSetEntitiesAnnotation fails on ppc64el

[22:13] sinzui: m looking. thnx! [22:13] sinzui: :( [22:14] sinzui: btw, thnx for landing my branch! [22:14] anastasiamac, I just sent an email about the bug to the list. Since CI has a queue to test, I am going to wait until tomorrow to forking 1.22. [22:15] sinzui: sorry :( [22:15] anastasiamac, I hope this gives you time to adjust the test [22:15] anastasiamac: \o/ congratulations your first CI breakage, it took too long [22:15] anastasiamac, you didn't create the queue@ [22:15] sinzui: i'll look at it a bit later than - kids r going nuts :( must be saturday [22:16] anastasiamac: is it already saturday for you? [22:16] perrito666: :( yes [22:16] anastasiamac: wow [22:16] perrito666: i grow older faster here [22:16] anastasiamac: lol [22:19] sinzui: i believe the problem is ordering - I say that I expect thing in one collection but the ordeer returned is reversed. [22:19] sinzui: i don't care about the order so will try to fix now ;) === urulama is now known as urulama__ [22:38] sinzui: proposed fix is on reviewboard http://reviews.vapour.ws/r/706/ [22:39] * katco looking [22:47] katco: thnx :) [22:49] anastasiamac: a few comments as i was trying to understand the code, and 1 spelling correction [22:49] anastasiamac: other than that, i'll lgtm [22:51] katco: about setting collection inline in other collections instead of declaring them as separate variables first [22:51] katco: i find it hard to read and maintain if struct changes [22:52] anastasiamac: it's just a map within a map right? [22:52] katco: which was the case for me until the format of SetAnnotations was decide [22:53] katco: yes it is now but went thru couple of changes [22:53] anastasiamac: ah ok [22:53] anastasiamac: well, whatever works for you :) it's certainly fine as is [22:53] anastasiamac: the function declaration is actually confusing for me, because there's no change in indentation from the declaration and its body [22:54] anastasiamac: nor whitespace [22:57] katco: I'll add whitespace [22:58] katco: had to params on separate linesotherwise signature is too long [22:58] anastasiamac: consider breaking after last param. indentation looks like a struct and is more readable [22:58] katco: thnx ;) that's what I meant [22:58] anastasiamac: also, i have not seen anyone leave the first line of the func declaration blank like that "func(\n\n..." [23:00] katco: oh k - i would have thought it's logical to move each param on its own line :P [23:00] katco: i'll move one to func( [23:00] anastasiamac: ty! [23:01] katco: thnx for looking! updated review :) [23:02] katco: can u lgtm? [23:02] anastasiamac: did you forget the newline before ) error {, or was that intentional? [23:02] L37 [23:04] katco: actually, my IDE tells me that it's not right to have return on a separate line. so intentional [23:04] anastasiamac: you just need an additional comma after the last param [23:05] ericsnow: almost back :) [23:05] wwitzel3: k === kadams54 is now known as kadams54-away [23:06] katco: i c. in that case, why would I want a newline after declaration? The indentation looks k now ;) let me commit for ur perusal [23:06] katco: committed [23:06] anastasiamac: yeah with the indentation for the return type, no need for newline after declaration [23:07] wwitzel3: I'm in moonstone when you're ready [23:07] anastasiamac: that looks much better... did that run through gofmt? the indentation looks strange in the declaration as well [23:07] anastasiamac: like the parameters are further in than the 1st param on my screen === kadams54-away is now known as kadams54 [23:10] katco: i believe it did. i have to go in a few min. can i get a lgtm? [23:10] anastasiamac: yup, thanks for humoring me [23:10] katco: anytime :) [23:11] anastasiamac: fwiw, play.golang.org also keeps that indentation style. that seems weird to me... [23:11] katco: m relying on my IDE to guide me as far as formatting (since it's all hooked in) [23:11] you know, if we change lgtms for amen we could get funny sentences more often [23:11] haha [23:12] perrito666: check out https://github.com/anastasiamac/juju/blob/set-annotations-Bug1409141/api/annotations/client_test.go [23:12] katco: m not usually too fussed about formatting (coming from Java and all) but by all means pull me up on it when needed [23:12] doesn't L34-37 looks weird indentation wise? [23:13] anastasiamac: i usually try to ignore formatting in reviews, but that one just got me. every time i tried to scan for the function i lost track of where it was [23:13] it does yet I know why that happened [23:13] perrito666: please enlighten me, amen. [23:13] go fmt sucks at fixing multiline function declarations [23:13] haha [23:14] it will do the right thing with the bracket [23:14] perrito666: :) [23:14] which will screw the rest [23:14] well that is rather frustrating [23:14] or the parens I cant recall [23:14] its like one of the rules breaks the rest [23:14] I usually make declarations one line [23:15] to avoid these things [23:15] ericsnow: sorry for the delay, it was late enough I ended up having to make dinner instead of just lunch for myself :/ [23:15] wwitzel3: no worries :) [23:16] perrito666: anastasiamac: http://play.golang.org/p/jKrQB2HQbg [23:16] looks a little better [23:17] if you drop the func onto a line by itself, that's when the formatting goes wacky: http://play.golang.org/p/INbYmRMf3- [23:17] katco: yes when each param is on its own line [23:17] anastasiamac: the difference being the "func(" isn't on it's own line [23:18] actually its when there is ( and ) in different lines [23:18] indentation of ) will make all look weird [23:25] perrito666: katco: thnx for ur help! something new every day :P [23:25] anastasiamac: thank you for taking the time on your sat. to fix that! [23:25] katco: i broke it :) [23:26] katco: besides it ain't fixed until it's merged [23:26] anastasiamac: isn't that the hard truth :p [23:26] anastasiamac: don't punish yourself, breaking the build happens, a lot, to everyone [23:27] katco: thnx for being my ocr on this ;) [23:27] anastasiamac: i am biased towards tanzanite members ;) [23:27] perrito666: m not punishing myself :) just taking ownership :) [23:27] katco: i knowthe feeling! [23:28] * perrito666 notices the matrix 2 is playing in the background... man those are cheap cgis [23:29] I wonder how much time has the TV been on [23:31] merge failed: "upgrade in progress - Juju functionality is limited" [23:31] * anastasiamac will re-try [23:31] anastasiamac: yup, that is a race [23:31] perrito666: who is winning? [23:32] perrito666: did i create t or shall I just re-try? [23:32] anastasiamac: nop, always been there [23:32] just retry [23:32] perrito666: thnx :0 will do [23:51] fix merged. hopefully this will unblock trunk eventually :) [23:52] anastasiamac: we will fix it if it doesn't please go enjoy your weekend [23:52] btw, what time on saturday is? [23:53] perrito666: now almost 10am :) [23:53] perrito666: what time is it for u? [23:53] did you pull an allnighter? [23:53] 20:53 on friday [23:53] perrito666: no ;-) my son is helping with Asian Cup (soccer) and ws too existed to sleep so we've been up since 5am [23:54] perrito666: and m more of an owl than an early bird - i work beta in the evening so I usually stay up til at leats 1am :) [23:54]