[00:00] arguably this function is serving two masters [00:00] so there will be some uglyness [01:14] thumper: tests pass in raring now :-D https://code.launchpad.net/~wallyworld/juju-core/fix-raring-tests/+merge/153488 [01:14] wallyworld_: seriously? [01:14] you're my hero if it works [01:14] * thumper grabs the branch [01:15] thumper: works for me, ymmv [01:16] i had the same issues as you, so easy to see it was fixed or not [01:18] * thumper wonders if his raring branches for goamz and goose are recent enough [01:19] I'd love not to have to work in that VM [01:19] it is a PITA [01:19] thumper: i have goose trunk (tip) and whatever goamz i got a while ago and it's all good for me [01:19] wallyworld_: thumper i fixed all the issues I could find in atlanta [01:19] nobody can replicate your git problem [01:19] which is le shit [01:19] * wallyworld_ can [01:19] ORLY [01:20] davecheney: wallyworld_ has [01:20] davecheney: https://code.launchpad.net/~wallyworld/juju-core/fix-raring-tests/+merge/153488 [01:20] ther was the one with the message difference [01:20] fixed :-) [01:20] fixed too [01:20] * wallyworld_ now has a clean test run under raring [01:21] davecheney: I think https://codereview.appspot.com/7809043/ is ready now [01:21] thumper: ta [01:25] wallyworld_: umm... your tests have hung [01:25] some do take a while for me [01:25] go test ./... is hanging [01:25] but they always have [01:25] the uniter ones take ages [01:25] this has been over 5 minutes [01:25] and stuck in one place [01:25] normally faster than that for me [01:25] which place? [01:26] whatever comes after cmd/charmload [01:26] which may not be where it is stuck [01:26] the changes are/should be very unobtrusive [01:26] as it is using buffered i/o [01:26] just setting some env vars [01:26] mine has always "stalled" after charmload [01:26] even before the changes [01:29] thumper: these ones for me always take a while [01:29] ok launchpad.net/juju-core/cmd/juju 65.754s [01:29] ok launchpad.net/juju-core/cmd/jujud 38.378s [01:29] they are logged always just after charmload [01:30] also these [01:30] ok launchpad.net/juju-core/state 29.673s [01:30] ? launchpad.net/juju-core/state/api [no test files] [01:30] ok launchpad.net/juju-core/state/api/params 0.006s [01:30] ok launchpad.net/juju-core/state/apiserver 59.152s [01:30] ok launchpad.net/juju-core/worker/uniter 67.571s [01:31] I'm pulling latest bits, and will try again [01:31] your branch doesn't look wrong [01:31] must be something else [01:31] did you try rerunning? [01:32] thumper: funny, i just ran the juju tests with gocheck.vv and it took ~40 seconds even with all the extra trace, as opposed to 65s above [01:33] * thumper runs with gocheck.vv [01:33] you can't to that at the top level and recurse afaik [01:33] .vv == vroom vroom ? [01:33] theoretically .vv means unbuffered I/O [01:33] but I think it lies [01:33] very verbose :) [01:33] yeah [01:33] seems to be stuck in the same place [01:33] normally the command tests use all my cores [01:34] but running at virtually 5% [01:34] blocked on i/o i guess [01:34] no, just not doing anything [01:34] *cough* sleep *cough* [01:34] wallyworld_: have you tested in raring? [01:35] back in lisbon a lot of test failures were hit with the sleep hammer [01:35] because this is broken for me [01:35] thumper: i am running raring [01:35] hmm... [01:35] and i could reproduce you git failures [01:35] wondering if I've hit a different race condition [01:35] * thumper has a fast machine [01:35] * wallyworld_ has a slower machine and a smaller penis [01:36] thumper: maybe try and underclock your cpus and see what happens? [01:36] wallyworld_: how would I do that? [01:37] power settings - battery mode? [01:37] or cpu greq select? [01:37] freq [01:37] hmm... [01:37] poos [01:37] still stuck [01:37] killed it [01:38] at 5 minutes of wall time [01:38] I wish there was a way to send it a signal and get it to dump where it was at [01:38] yeah [01:38] maybe paste the vv output? [01:38] nothing to see [01:39] ? launchpad.net/juju-core/cmd/charmload [no test files] [01:39] that was the last output [01:39] davecheney: should I have any mongod process running after the tests have finished? [01:39] I have one [01:39] should I kill it? [01:40] thumper: you need to cd to just the juju tests [01:40] to use gocheck.vv [01:40] since ./... and gocheck args don't work together [01:40] seriously? [01:40] * thumper sighs [01:40] which ones were failing before/ [01:40] ? [01:41] thumper: the git tests [01:41] are the easiest to see [01:41] thumper: no, but they do leak [01:41] also the uniter ones [01:41] and deployer ones [01:41] i go thought and clean them up [01:41] or learn not to ^C the tests [01:42] davecheney: question, this had be stumped - cmd/jujud/unit_test.go:20: s.agentSuite.SetUpSuite(c) <--- i had to add this and didn't know why the call just didn't drop through [01:42] why does aliasing the import fix it? [01:43] * davecheney scrolls back to find the CL [01:43] wallyworld_: it's not aliasing [01:44] hang on [01:44] which file are we talking about ? [01:44] unit_test.go [01:44] i misread the diff a bit [01:44] wallyworld_: stuck here http://pastebin.ubuntu.com/5615399/ [01:45] davecheney: you wanted testing.GitSuite renamed to GitSuite [01:45] wallyworld_: right, sorry was looking at wrong CL [01:45] it's not renaming, it is giving it a name [01:45] at the moment both fields in the struct are anonymous [01:45] which embeds them [01:45] but i don't understand why that is needed and why i had to add the explicit call throughs [01:46] however that creates and ambiguity on who defines the embedded UnitSuite.SetUpSuite [01:46] thumper: you have thr wrong mongo [01:46] ah... wat? [01:46] thumper: you need to install the one from the tarball and set your path to that one [01:46] thumper: since ssl is not supported in the packaged version [01:46] thumper: yes, don't use mongo 2.0 from raring [01:47] Installed: 1:2.2.3-0ubuntu1 [01:47] and something about lawyers says we can't add ssl [01:47] thumper: bzzt, that one does not have ssl enabled [01:47] wallyworld_: I think the lawyers are still arguing [01:47] clucking bell [01:47] juju dev sucks the big kumara [01:47] yeah, sucks balls [01:47] didn't bigjools have a ppa somewhere [01:47] this is out of our control sadly [01:47] I don't like the tar ball answer [01:48] yes, believe so [01:48] wallyworld_: by giving testing.GitSuite a name, you break the ambiguity [01:48] I do [01:48] wallyworld_: nothing is out of our control [01:48] we control EVERYTHING [01:48] bigjools: you may need to rebuild [01:48] bigjools: plzfix [01:48] * davecheney contols both the horizontal and the vertical [01:48] davecheney: ah thanks, i 'll re-read the code with that though in mind. i need to think about it so i fully grasp it [01:48] thumper: hmmm? [01:48] thumper: that osn't bigjools' version, it's coming from raring [01:48] wallyworld_: no worries [01:49] mine has SSL [01:49] yeah.. I know [01:49] bigjools: needs to make a new one [01:49] pebkac [01:49] the other part of the solution is, when you embed a structure, its field name is the name of the structure [01:49] thumper: what's up with it? [01:49] hmm... [01:49] thumper: so now you can +1 my mp when i fix davecheney's issues [01:49] so by naming testing.GitSuite, GitSuite, none of the rest of the code notices, as that was what it called it originally [01:49] :-) [01:49] perhaps the ppa has been disabled [01:50] thumper: bigjools packaged 2.2.2 [01:50] apt-cache policy .... tells you all [01:50] raring offers 2.2.3 [01:50] ^ sorta guess [01:50] and the ppa was disabled on upgrade [01:50] bigjools: make a new one and I'll buy you a beer [01:50] * wallyworld_ is glad he just used the barball [01:50] tarball [01:50] arf [01:50] wallyworld_: right first time [01:51] hah [01:51] but I don't want to use the dumb tarball [01:51] my favorite mongo feature is the 60hz timer it sets up [01:51] wallyworld_: where are the instructions? [01:51] specifically to waste your battery [01:51] thumper: they _should_ be in the README [01:51] * davecheney scratched head [01:51] thumper: not sure, i just figured it out - untar to /opt (say) and update yuor path [01:52] it just feels SO wrong, that I'm resisting [01:52] thumper: they would ahve to be in the README (checking) [01:52] until we get a better solution [01:52] we onboarded so many juju devs in Atlanta [01:52] * bigjools agrees with thumper [01:52] * wallyworld_ agrees too [01:52] but we encountered some resistence trying to change it [01:53] * davecheney also agrees, but points the phalanax of lawyers that lie between us and our desired solution [01:53] * thumper gathers all his toys in one corner [01:53] yeah I only built for quantal [01:53] bigjools: that would be it [01:53] I'll do another [01:53] bigjools: awesome, ta [01:54] davecheney: who can I poke to make this go faster? [01:54] there is that long email thread [01:54] but I'm prepared to go higher [01:54] * thumper is sick of it [01:54] thumper: antonio and jamespage have identified themselves as the owner of the issue [01:55] go thumper, go thumper, go thumper.... oooo yeaahhhh go girl [01:55] the last status update I got said they were waiting on 10gen to do something for them [01:55] * thumper goes to look them up on the direcdtory [01:55] so, we're waiting on the good graces of the 10gen lawyers [01:55] GLWT [01:55] davecheney: in which cases they'll be waiting forever [01:55] exactly [01:55] ... [01:56] thumper: are you cc'ed on that email thread [01:56] davecheney: yeah, and I'm about to reply and start kicking tires [01:56] go thumper go [02:01] ok, it seems that james did create a version for raring, but it has been superseded by the raring release [02:02] * thumper emails everyone [02:02] thumper: an ssl version? [02:02] yeah [02:02] according to the email [02:02] I'll build anyway [02:05] hmmm [02:06] bigjools: how long will it take? [02:07] thumper: iirc about 30-60 mins [02:07] ok... I'll wait [02:07] and use your one [02:07] * bigjools can bump ppa priority :) [02:07] \o/ [02:07] in fact there's a little know changelog trick to get more priority too [02:07] known* [02:19] davecheney: thanks for the review - with the empty SetupSuite() methods, I cargo culted those from elsewhere. So I guess those other places should be fixed too at some point [02:20] wallyworld_: stick a card in leankit to refactor that shit [02:21] will do [02:21] the contract for gocehck.Suite() takes an interface{} [02:21] so there is no requirement for every Suite to have those methods [02:21] they are only needed if used [02:21] and empty ones dont' really help [02:21] apart from that, get thumper to LGTM [02:21] yeah, i thought so too, but figured it must have been needed for some reason/convention [02:21] then you'll be in sweet sweet raring heaven [02:22] oh yeah :-) [02:22] wallyworld_: i think you were right wen you said cargo cult [02:22] it happens if one is a bit unsure when implementing something and one assumes that what was done before was correct :-) [02:23] and yet we have two +1's required [02:23] so it seems stuff still slips through [02:25] wallyworld_: i don't think you want to argue for quorum on a change request [02:25] :) [02:25] sure :-) but my point is with two reviewers, perhaps "obvious" stuff could be caught [02:26] buddha says "when arguing with loved ones, don't bring up the past" [02:27] not meant to be negative, just an observation [02:27] an observation well made [02:37] davecheney: i only found one place to fix as it turns out - LoggingSuite has empty Setup/TearDownSuite() and a few calls to those - i'll just fix it up as a drive by [02:37] +1 [02:37] thumper: as soon as you +1 my mp, i can land the raring test fixes :-) [02:38] link? [02:39] thumper: it's building https://launchpad.net/~julian-edwards/+archive/mongodb/+packages [02:41] thumper: https://codereview.appspot.com/7677044/ [02:42] wallyworld_: acked [02:42] do it [02:42] land it [02:42] now [02:42] \o/ [02:43] thumper: just doing a quick driveby to remove some redundant code, will be done in a few minutes [02:48] * davecheney away -- lunch [02:49] davecheney: i've removed the empty setupsuite stuff, but am thinking - perhaps it was done that way so that stuff could be added if required without needed to then go and add setupsuite calls to all the structs which embedd the logging test suite [02:51] wallyworld_: maybe leave the drivebys for another day [02:51] yeah, i just reverted them :-) [02:51] i'll just leave this here http://www.youtube.com/watch?v=Ktbhw0v186Q [02:51] afk for reals this time [02:52] you are a sadistic man [02:53] thought it was going to be a rickroll [02:53] or worse [02:59] thumper: can you feel it? [02:59] can I feel what? [02:59] haha [03:00] that's not the first time someone has said that to me [03:00] * bigjools lunches [03:00] thumper: if you give it a oull, it will come [03:00] pull [03:00] ... [03:00] the raring test fixes i mean [03:00] oh [03:00] not his lunch? === jtv1 is now known as jtv [03:00] you all have dirty minds [03:01] wallyworld_: I need bigjools's mongo branch first [03:01] well given who was saying it ... [03:01] and I'm kinda busy actually working [03:01] sure, just letting you know :-) [03:01] wallyworld_: except Julian, who has dirty hands. [03:01] lol [03:01] * bigjools wipes [03:01] ... [03:01] bigjools: on the curtains? [03:02] That's it. No food for me this lunchtime. [03:02] thumper: at least 30mins left to build [03:02] it takes an hour [03:02] * bigjools heads off [03:02] Meanwhile, a question for the juju experts: we need to implement EnvironProvider.InstanceId() for the maas provider, but the EnvironProvider has no idea what maas it's supposed to talk to. [03:04] It's not like the ec2 provider where you have a fixed IP address for the metadata service. [03:04] jtv: in the vm world, it is the machine id on which the provider is running [03:04] So... the bootstrap node? [03:04] yeah [03:05] and others too i think [03:05] :( [03:05] the id is used to update the state [03:05] so the agent can see that the machine has been provisioned [03:05] i think [03:05] The fact that the method lives on EnvironProvider suggests that no provider needs to be running. [03:05] Otherwise it'd be on Environ, where we have a real chance of getting the information required. [03:05] yes [03:06] this is done before that happens [03:06] Damn. [03:06] on ec2 and openstack, there's the metadata service [03:06] which provides this [03:06] On MAAS too, but at this point we don't know its address. [03:06] :-( [03:07] i *think* the update of the state can be delayed [03:07] till the provider starts [03:07] but you'd need to talk to william [03:07] pleasesaygrantpleasesaygrantpleasesaygrant [03:08] * jtv is in a timezone far, far from the UK [03:08] readereadereadereade [03:08] buggerbuggerbuggerbugger [03:08] the main other person who would know is also in the uk [03:08] haha [03:08] uk... ISO country code for: [03:08] Ukraine [03:08] That helps [03:08] no it isn't [03:09] ? [03:09] i meant england [03:09] gb == united kingdom of great britain & northern ireland [03:09] wallyworld_: I figured. Just doing some wishful thinking [03:10] jtv: right now, i can't see why it has to be the environprovider which provides that info - i think some refactoring could fix the problem [03:11] but i don't fully appreciate the finer details of the workflow to offer practical advice and understanding of the consequences [03:11] That would mean the world to us at this stage. We're currently stuck on this. [03:11] jtv: maybe send an email to the list - william is good at responding [03:12] Yes, he is. Thanks, I'll do that. [03:12] hopefully can be sorted for next week [03:14] We have until... This evening. :/ [03:14] I'm putting together the email now. [03:23] jtv: iso country codes are fucked [03:24] * bigjools relurks [04:13] jtv: UKRAINE A2 = UA A3 = UKR [04:14] jtv: however UK is not a 2 character country code for anything [04:14] ! [04:14] Damn. [04:14] Well, I apologize. [04:14] * thumper was bored so looked it up [04:15] given that domains for the uk finish with .uk [04:15] you'd think it was the iso code [04:15] but you were right on the .gb [04:15] which I don't think anyone sues [04:15] uses [04:15] jtv: perhaps ukraine and UK got together and agreed that neither should have 'UK' [04:15] just to piss everyone off [04:16] thumper: you asked for it, http://www.youtube.com/watch?v=rNu8XDBSn10 [04:16] davecheney: that's ok [04:16] davecheney: I did think for a minute that you were rick rollingme [04:16] didn't the ukraine used to be called something else ? [04:16] davecheney: like the USSR? [04:17] thats the one [04:17] before it got broken into little pieces [04:17] Well they were a republic within the USSR. [04:17] Not an "independent republic" — those were parts of Russia and did not have the theoretical right of secession. [04:17] * davecheney does not want to give the northern territory it's own top level domain [04:18] .aunt? [04:18] next thing you know we'd give QLD .xxxx [04:18] ? [04:19] http://www.xxxx.com.au/ [04:20] Phew. I was taking a risk by loading up that one... [04:20] A neighbour had a visit from the MIB about having looked at Wikileaks. [04:20] But this ain't so bad. :) [04:21] which leads me to the joke: why do australians put four X's on their beer? [04:21] jtv: GTFO [04:21] a) because they can't spell beer [04:21] Sounds good. Go on. [04:21] ? there is a second answer ? [04:21] jtv: seriously, MIB? [04:21] Well no, I don't know what they were actually wearing. But they were from the government and they were not friendly. [04:21] I heard that amazon have bought up .book .author and something else [04:22] wow [04:22] .cloud ? [04:22] don't recall [04:22] bigjools: built? [04:22] I always wanted to have to gov't agencies visi and ostro here: ostro.go.th & visi.go.th. [04:22] i remember the freeforall back in '98 when you could buy a $.com without having to own the trademark [04:22] * bigjools points thumper to the fine url from earlier [04:23] * thumper scrolls back looking for the url [04:25] published an hour ago [04:25] mmm lunchtime dip in the pool was very nice [04:29] hmm... [04:29] doesn't fix the problem [04:29] tests still sit doing nothing [04:30] thumper: error message? [04:30] unknown option sslOnNormalPorts [04:30] you sure the new one is being used? [04:30] that implies it's still using the old one [04:30] Installed: 1:2.2.3-0ubuntu2+ssl [04:31] hmmm [04:31] * wallyworld_ has no idea [04:31] hmm, just installed mongodb, think i need the other three package [04:31] I thought it would do that [04:31] i just untarred the tarball for mine - perhaps everything was included [04:32] but i could have sworn there was just a single mongodb dir in there [04:33] jtv: you sent your email to the list? [04:34] wallyworld_: you didn't see it? I may have had the wrong email account active. Let me retry. [04:34] wallyworld_: working now [04:34] thumper: \o/ [04:34] what did you change, just install the extra packages [04:35] jtv: yeah, didn't see it so thought i'd ask in case you mis sent it [04:35] yeah, just also installed mongodb-clients mongodb-dev mongodb-server [04:35] it that all? [04:35] I thought that installing mongodb would bring in the others [04:35] yes [04:35] i was trying to be sarcastic :-) [04:35] wallyworld_: I do get some failures though [04:35] :-( [04:35] but I think they are timing related [04:36] so re running, different results? [04:36] cmd/jujud OOPS: 29 passed, 1 skipped, 2 FAILED, 3 FIXTURE-PANICKED, 14 MISSED [04:36] 5 failures, not good [04:37] Thanks for checking, wallyworld_ [04:37] np [04:37] wallyworld_: is it in trunk now? [04:37] I'll get trunk and try again [04:37] yah [04:39] nope, still failing tests for me [04:40] * thumper runs all again and pastebins... [04:40] * thumper takes kids to get fud [04:44] pastebin? [05:36] * wallyworld_ off to parent teacher interview, bbiab [10:23] fwereade: you've got a couple of reviews [10:23] rogpeppe, lovely, tyvm [10:24] fwereade: i've got the next stage in the allWatcher out for review, if you fancy a look. it's all easy after that one. https://codereview.appspot.com/7594048/ [10:57] rogpeppe, I don't think I'll manage to do that properly before this afternoon, I'm a little bit involved with environs at the moment [10:57] fwereade: np [10:57]