=== natefinch-afk is now known as natefinch [00:31] wwitzel3, ericsnow: btw, I made a change to the juju status --processes output to 1.) be valid yaml (list of hashtables) and present valid looking data. Would appreciate some eyes on it: https://docs.google.com/document/d/1PcRQXaerlsACro4y1y5LWD-uvhfHya2CkOcoljyFyCU/edit#heading=h.3ypclmmspp3g [01:12] sinzui: is there any way I can watch the current CI run? [01:13] thumper: login to http://juju-ci.vapour.ws:8080/view/Juju%20Revisions/, sort the columns on last success [01:14] sinzui: this doesn't tell me which are voting etc [01:14] I want to get a jump on any potential failure... [01:14] thumper: the ones in read at the bottom [01:15] also, how can I tell which rev from which branch is being checked? [01:15] I know I can click through [01:15] but wondering if there was a summary page [01:15] for the current one [01:15] perhaps a nice future feature :) [01:15] thumper: voting isn't Jenkins, it is a ci-directory feature. Jenkins also corrupts the description so it is hard to read the rules for each job [01:16] * thumper nods [01:17] thumper: I might break that view. I am trying to add columns that give more incite [01:17] i'm linda sad that vivid and wily unit test runs have failures [01:17] damn left hand typing [01:18] thumper: They are close ot passing though. I think wily is fixable in a few hours [01:18] eta on current ci run? [01:19] thumper: I cannot make description viewable, but...The weather icon is a condicator cloudy means there is a real chance of failure and black cloud often mean released revisions with failures [01:20] * thumper nods [01:20] thumper: about 2 hours. I removed all the utopic jobs last hour. I am trying to make the openstack on maas job run faster [01:23] something bad just happened revision results is queued. I think the won tests failued for 1.24 [01:24] wat? [01:26] I think the machine needs to be restarted. I am doing that now an requeuing the tests. the failurea re not consistent [02:24] thumper: FYI, looks like your most recent change in the cmd repo breaks a few tests in juju core [02:24] what are you referring to? [02:25] thumper: https://github.com/juju/cmd/pull/19 [02:25] ericsnow: it all works in jes-cli :) [02:25] so I know I've fixed it where I'm bringing it in [02:26] thumper: k [02:26] ericsnow: do you need the new rev? [02:27] thumper: just need to know how you fixed jes-cli [02:27] with magic and fairie dust [02:27] :) [02:31] thumper: looks like you took care of it in https://github.com/juju/juju/pull/2540 [02:32] * thumper nods [02:32] see... fairie dust [02:32] sprinkled over 37 files [02:33] lol [02:53] thumper: what's the plan for hosted env destruction, in light of fwereade's reply? [03:02] axw: I'm thinking we'll need to do something new [03:02] but I'm thinking through what [03:03] * thumper afk taking kids to jiu-jitsu [03:03] thumper: okey dokey. I think I'll go ahead and clean up storage in destroy-env --force for now, but leave the apiserver-side destruction. we can use what I'm doing if needed though [03:03] later [05:56] urulama: hey [06:00] hey, thumper [06:00] urulama: when is good for you? [06:00] if you need an hour or so [06:01] I can come back [06:01] tomorrow same time is ok for you? [06:01] yeah... [06:01] I guess [06:01] my mother got sick so kids stayed at home an i need to figure out what to do [06:01] :( [06:01] kk [06:01] np [06:01] chat tomorrow [06:01] thanks, thumper [06:50] jam: saw your email, shared doc with you as i could access [06:50] thx [06:51] got it [07:16] axw: if you get a chance before eod, could you look at http://reviews.vapour.ws/r/2224/ for me? [07:17] wallyworld_: ok [07:17] ta [07:33] sigh, yet another UPS that is not U [07:40] wallyworld_: LGTM [09:32] dooferlad: regarding our talk about postgresql see here https://twitter.com/rpy/status/622291301002670080/photo/1 [09:32] TheMue: arg! [09:32] dooferlad: hehe [09:33] dooferlad: dunno how to ever have this idea [09:48] does ec2 support agent-metadata-url? === frobware is now known as mcd === mcd is now known as frobware [10:07] bogdanteleaga: yes, it does [10:07] mgz: I'm trying to set agent-metadata-url after the secret key [10:07] mgz: and it still goes to streams.canonical.com [10:08] mgz: am I missing something? [10:10] bogdanteleaga: welll... in practice our ci envs use tools-metadata-url but that's just for compat with older versions [10:10] mgz: I've tried that too, it gave a few warnings, but it still didn't work [10:10] mgz: I've been doing this with maas for a while now, but it just doesn't seem to pick it up here [10:11] bogdanteleaga: did you bootstrap with --debug? paste the log? [10:13] http://paste.ubuntu.com/11913639/ [10:14] http://paste.ubuntu.com/11913647/ [10:23] bogdanteleaga: nothing jumps out at me [10:24] I'll have to enable try debug with one of our scripts and compare the simplestreams bits more closely [10:24] bogdanteleaga: have you done this before for earlier juju versions? [10:25] mgz: nope, I've always did it for maas, but I haven't tried the other providers [10:25] wait, why does it say "trying keystone catalog" [10:25] that's not ec2 code [10:26] heh [10:27] bogdanteleaga: have you tried `juju metadata validate-tools`? [10:28] http://paste.ubuntu.com/11913726/ [10:29] I'm still not sure wtf is happening with the keystone thing [10:30] bogdanteleaga: that was with -e amazon? [10:30] nope but it's set as default [10:30] and you can see it's trying amazon in the logs [10:30] the first ones [10:39] bogdanteleaga: http://paste.ubuntu.com/11913764 [10:40] looks like if there are no matches you'll only see the last fallback, which is streams [10:40] so I think I'd start by trying to get validate-tools to see the matching tools [10:45] I gave you a paste of validate-tools [10:45] it still doesn't try the ones specified in the url [10:45] mgz ^^ [10:46] bogdanteleaga: compare with my paste [10:46] mgz: I should have a tools-url? [10:47] when I tried a version that wasn't in the streams I had, it doesn't show my different agent url [10:47] bogdanteleaga: no, that's spurious junk in the environments.yaml file I'm using there [10:48] bogdanteleaga: so, it may just be that your streams data is somehow now matching when you think it should [10:48] I might have an idea [10:48] I'm using devel tools and I don't think it's looking after those by default [10:50] right, yeah, you should add agent-stream: devel [10:50] I'll try to upload them again [10:51] no, leaving as devel in the streams data is fine [10:51] you should just add that in your environments.yaml [10:52] oh yeah, I meant I'll upload them again just in case [10:52] then run the validate-tools again and see if it helps [10:52] the issue with doing devel in the environments.yaml [10:52] is that the bootstrap does not fail [10:52] and it uses local tools in case it doesn't find them on simplestreams [10:52] which doesn't work on windows [10:53] nah, somethings badly broken [10:53] it shouldn't output anything about keystone [10:53] I think I'm missing something, I'll try it from the top [10:58] morning [11:02] hey horacio! [11:02] bogdanteleaga: http://paste.ubuntu.com/11913877 [11:02] mgz: I managed to get it working, it was a problem with how I uploaded them. It still says crap about keystone though [11:03] with debug, you can see it's trying the config location then falling back to the main [11:03] mgz: that's probably gonna be helpful thanks [11:12] jam: not sure if you have a minute to clarify the hash as part of the naming for resources? [11:12] wallyworld: /wave [11:12] hey [11:14] jam: might be easier to have a hangout if you had time. if not now, then i can hang around [11:16] wallyworld: a hangout is probably warranted, I'm currently in another conversation, so I don't want to be too distracted to talk both yet [11:17] jam: no problem, just ping me if you are free later [11:37] sinzui: isn t https://bugs.launchpad.net/juju-core/+bug/1475724 fixed? [11:37] Bug #1475724: UniterSuite.TestUniterRelations

[12:05] perrito666: probably not, but it's not a completely consistent failure so it's hard to day [12:08] mgz: I retracted update-status patch [12:08] so at least That should be fixed (not the underlying problem) [12:09] well... maybe we should mark it as fixed and see [12:10] the batch of failures we have on it from the 16th rev 5b05541 are probably due to that branch landing [12:10] at least the trusty one I just looked at was a failure due to having one rather than two update-status calls [12:12] mgz: this is master? [12:12] we currently hve an updated jes-cli being tested that has got passes on tests that were failing pretty reliably [12:12] perrito666: yup, and feature branches from master [12:12] mgz: odd, this landed last night [12:13] perrito666: I'm looking at the history of failures since the 16th [12:13] not a CI run with last night's change [12:13] master hasn't been run with your revert in yet it seems [12:13] * perrito666 is a human popsicle [12:13] you're melting in the sun? [12:14] falling off your stick? [12:17] mgz: argentinian constructions are old, completely made of brick and certainly not prepared for cold [12:18] ah yes, you are in the other hemisphere [12:19] I guess you don't want to hear about how sunny and hot it is here then? :) [12:28] mgz: you live in brittain, you probably call "sunny and hot" the same thing I am calling freezing cold [12:28] :P [12:29] mgz: its 4C which must be a typicall hot summer morning for uk, right? [12:32] doesn't go support a way for us to have blobs of text in something other than a const? [12:33] I am reviewing a go file that contains 2 lines of go and a lot of powershell [12:35] perrito666: I am sure that the right way of having lots of powershell is to read in a powershell file [12:36] dooferlad: I am sure of it, the problem is how do you distribute that? [12:37] perrito666: other than distributing the .ps1? [12:37] dooferlad: this is something I wish could be embedded as a resource on compile time [12:37] dooferlad: yes [12:38] dooferlad: the thing is, our software should not depend on packaging particularities [12:38] perrito666: can we really not just distribute the file? That seems kind of wrong. [12:39] dooferlad: dunno, I would say not as we have the same thing for bash at some points [12:40] perrito666: sounds like something worth starting a conversation about. I would hope that after we move to go 1.4 we could use go generate (http://blog.golang.org/generate) to pull in external files into consts [12:45] dooferlad: take a look http://reviews.vapour.ws/r/2199/ I accept suggestions [13:20] bogdanteleaga: see last comment in bug 1287949 [13:20] Bug #1287949: {image,tools}-metadata-url not usable w/ ec2 provider

[13:33] bogdanteleaga: poké for hangout [14:14] ericsnow: please get the blessing from fwereade too on my review [14:14] perrito666: k [14:18] perrito666: and thanks :) [14:32] * fwereade just wrote some disgustingly smug test code, and can't figure out if it's good or bad: http://paste.ubuntu.com/11914787/ [14:40] fwereade: cuteness depends on map ordering? :) [14:40] mgz, the cuteness depends on the order I wrote them in, but not on any runtime properties AFAIAA [14:41] fwereade: cute to read, might be confusing to debug [14:42] it's always what the failures would look like that I think about [14:45] mgz, I'm not sure the word choice helps or hinders much -- it's deliberately a tangled racy write because that's what I'm testing [14:46] can anybody check this out? http://paste.ubuntu.com/11914363/ [14:47] I'm trying to get aws with a custom image up [14:47] it seems to detect the custom image metadata and then throw it away and resort to the official one again [14:47] fwereade: to be clear, it's fine as written, I'm just chatting about paint [14:48] mgz, cool -- I'm still on the fence for some reason [14:48] bogdanteleaga: did you add in the unsigned selection hack? [14:48] mgz, mainly because I look at it and I can see me admiring my own cleverness and that's a bit unattractive ;) [14:49] mgz: nope, because from what I gathered from the bug it wouldn't work for either tools or images [14:49] mgz: and tools work for me [14:51] then I think you need to add in more logging in the simplestreams code till you understand why it's not selecting an image from the first stream [14:52] it does seem to select it at first [14:53] sure, but it's not matching an image, need to work out why [14:55] if you look at line 7 it does match it, it just fails after that and I think after the failure it tries again with the official sources [14:55] can't seem to understand why it fails though [15:05] bogdanteleaga: the 1.24.3 - 1.24 tip upgrade does seem to be broken by the registry thing [15:07] bogdanteleaga: http://data.vapour.ws/juju-ci/products/version-2906/maas-1_8-upgrade-win2012hvr2-amd64/build-62/machine-1.log.gz [15:21] katco, ericsnow: ping if you need me, I think this --extend bug is the last hurdle for me *fingers crossed* [15:21] wwitzel3: cool [15:22] wwitzel3: ty [15:43] bogdanteleaga_: ah, you dropped [15:44] Bug #1473517 changed: juju environment not usable after the upgrade

=== bogdanteleaga_ is now known as bogdanteleaga [15:55] bogdanteleaga: http://data.vapour.ws/juju-ci/products/version-2906/maas-1_8-upgrade-win2012hvr2-amd64/build-62/machine-1.log.gz [15:58] sinzui, ocr: http://reviews.vapour.ws/r/2228 === dames is now known as thedac [16:01] mgz: youve got to be kidding me right? [16:01] that is 7 pages of reviews [16:01] perrito666: you are not intended to review-review it [16:01] just check I haven't screwed up the merge [16:02] the reviews happened in the landings onto the jes-cli branch [16:02] mgz: rb is failing for some files [16:02] and ci has done a full run on the branch [16:02] perrito666: you probably need to use github, rb is bad with moves [16:02] perrito666: I added a comment about the testing done [16:03] perrito666: http://reports.vapour.ws/releases/2906 is blessed and confirms https://bugs.launchpad.net/juju-core/+bug/1475724 is fixed [16:03] Bug #1475724: UniterSuite.TestUniterRelations

[16:05] how wonderful it would be that the sign on would redirect me to the link I was going [16:06] sinzui: mgz evidence seems to tell it is all ok, but I dont think I could notice a merge misshap in 139 files [16:06] :/ [16:10] sinzui: I think we have been careful enough to be able to say this is ok to merge [16:11] bbl, lunch [16:38] Bug #1300692 changed: relation-get fails with return code 2

[16:41] katco, if you're OCR, can I point you at http://reviews.vapour.ws/r/2229/ and http://reviews.vapour.ws/r/2230/ ? [16:41] katco, 2229 is a minor interface change, already reviewed elsewhere, but maybe of interest to you [16:41] katco, 2230 actually uses it [16:44] Bug #1300692 opened: relation-get fails with return code 2

[16:50] Bug #1300692 changed: relation-get fails with return code 2

[16:51] fwereade: not ocr today... if i have time i'll give it a look later :) [16:51] katco, ...hmm, apparently it's not wednesday [16:51] who knew? [16:51] ahem [16:52] haha [16:52] perrito666, ^^ [16:52] ericsnow: so I've narrowed it down to command.go not properly applying --extend [16:52] ericsnow: which is highly probably since we don't have tests for that [16:52] wwitzel3: k [16:54] ericsnow: so I'm added some tests and hopefully that will point me to the problem :) [16:57] katco: wwitzel3, ericsnow: back [16:57] wwitzel3: cool [17:02] ericsnow: as I'm attempting to make tests for this, I think I am getting the sense why it wasn't testing :P [17:02] wwitzel3: :) [17:14] wwitzel3: hey where is that charm at? === tvansteenburgh1 is now known as tvansteenburgh [17:27] charmers, core: how do access predefined hooks like relation-changed or relation joined hooks in python when writing a charm? Must I import those charmtools or charmhelpers modules? [17:27] bdx: I'm not sure I understand what you're asking [17:29] marcoceppi: for example, in the tutorial https://jujucharms.com/docs/stable/authors-charm-writing [17:29] hooks are referenced like "db_host=`relation-get private-address`" [17:44] marcoceppi: how can I access similar hooks in python? [17:46] bdx: I see, those aren't hooks, those are hook-tools [17:46] hooks aer the relation-joined, relation-changed, config-changed, etc [17:46] bdx: there's a charm-helpers libaray, charmhelpers.core.hookenv which has all those available as methods [17:47] bdx: https://pythonhosted.org/charmhelpers/api/charmhelpers.core.hookenv.html [17:47] bdx: relation_get will be what you're looking for [17:58] marcoceppi: ok, that sweet..... I think I was just a little confused using hooks between bash and python [17:59] wwitzel3: natefinch: ericsnow: i'm back. what can i help with? [18:00] katco: not sure [18:09] wwitzel3: please lmk when the charm is ready to go. i think we're blocked on that atm [18:13] katco: I was thinking that ComponentStatus should be a map[string]interface{} that way a component can just dump a struct into it, and have strongly typed values that automatically get converted in and out. [18:14] natefinch: ok, noted. let's discuss more after the demo [18:14] katco: ok [18:14] natefinch: can you still manage your piece w/o doing that? [18:15] katco: yeah [18:20] Bug # changed: 1370896, 1430245, 1461354, 1463480, 1465115, 1468653, 1469130, 1469199, 1471332, 1472014, 1472729, 1475163 [18:34] Can I just make a simple request? I'm not sure who thought juju scp would be better over rsync, but it's not, it should be juju rsync ;) === Viperz28_ is now known as Viperz28 [18:57] sinzui: Can we delay for 1/2 hr and meet at 3:30? [18:58] xwwt_: sure === xwwt_ is now known as xwwt [19:01] btw, I am back [19:08] wwitzel3: ping? [19:40] katco: yep, I'm still trying to figure out why the --extend args are not being properly populated [19:41] wwitzel3: need any help? running out of time [19:42] katco: other people exploring it would be fine, I'm just walking through and re-compiling different logging outputs right now trying to find out how far it is actually getting [19:43] btw, I'm trying to test the status output, just whipping up a test charm that uses the docker plugin [19:43] katco: right now I'm poking around in process/context in the launch command and launch_test [20:15] ericsnow: natefinch: wwitzel3: let's plan on touching base in moonstone in 15m or so? [20:15] katco: k [20:21] katco: k [20:22] ditto [20:52] OMG, is that a bless on jes-cli I see? [20:58] thumper: yeah! [21:04] * fwereade cheers at thumper, wonders if he'd like to talk environment lifecycles shortly? am approaching sleepiness but also sort of want to stay up and talk to wallyworld [21:05] fwereade: on calls for a while now [21:05] fwereade: but yes, at some stage [21:05] probably not right now [21:06] thumper, sgtm, I'm not *especially* in the mood myself right now :) [21:20] fwereade: i have a meeting in 10, should be short hopefully [21:25] wallyworld, cool, ping me when you're done, I may or may not still be both present and coherent [21:25] ok [21:26] or we can talk at yuor SOD [21:26] we have one scheduled [21:32] sinzui: ping [21:59] fwereade: I'm forward porting this https://github.com/juju/juju/pull/2801 but of course everything has changed in master [22:00] fwereade: I'm going to remove the insertWithoutEnvironment collection attribute. yell at me if you don't think that's right. [22:00] menn0, that's exactly right, and it's what I did when I backported all that stuff to 1.24 [22:01] cool [22:01] menn0, I think if you check 1.24 the multiEnvRunner already does the right thing [22:02] fwereade: well yes, I removed the always-assert-on-env-life feature from there (that's what 2801 is about) [22:04] menn0, I think I had a point but it escapes me right now [22:05] fwereade: np :) [22:42] why is 1475724 fix committed yet still blocks? [22:49] <_thumper_> perrito666: possibly because the blocker tag is still there and the revision hasn't passed the relevant CI test? === _thumper_ is now known as thumper [22:49] thumper: ah makes sense [22:50] sinzui, ^^^ [22:50] * perrito666 was falling asleep on the keyboard, decided to cook to wake up a bit, now is falling asleep and laptop has garlic smell [22:51] perrito666: fix commits is not CI's proof that the fix works and doesn't not introduce other regressions. [22:52] perrito666: alexisb thumper : CI is almost done proving that the fix work and will open master with jes-cli merged [22:53] Joyent is very ill. [22:55] wallyworld: is http://reviews.vapour.ws/r/2225/ a back/forward port? [22:56] menn0: forward port, already landed in 1.24, waiting for master to unblock [22:56] wallyworld: cool, i'll ignore it unless you say otherwise [22:56] np [22:59] Who wants to be first to break master [23:00] memememme [23:00] perrito666: thumper alexisb wallyworld : Ci blessed master (wiith jes-cli merged) and it closed the blocking bug. We can merge now [23:00] sinzui: \o/ [23:01] mail client with notification ftw, instant bless notice [23:01] :p [23:01] Don't panic (as I did) about the email with all the failed joyent jobs. Joyent is/was very ill. Machines were not being provisions in our region [23:01] yay [23:09] Bug #1475724 changed: UniterSuite.TestUniterRelations

[23:33] sinzui: is that rev going to be 1.25-alpha1? [23:47] * wallyworld off to pick up newly repaired coffee machine :-D bbiab