=== jillr_ is now known as jillr [00:26] thumper: // InitializeFromConfig needs to be called once after the environment [00:26] func InitializeFromConfig(config PreferIPv6Getter) { globalPreferIPv6 = config.PreferIPv6() logger.Infof("setting prefer-ipv6 to %v", globalPreferIPv6) [00:26] } [00:26] urgh [00:26] http://paste.ubuntu.com/13356006/ [00:26] this method does not do what it's comment says [00:26] wa? [00:26] it initalises nothing [00:26] unless it's a side effect of calling config.PreferIPv6 [00:27] davecheney: it sets the global variable [00:27] globalPreferIPv6 = config.PreferIPv6() [00:28] oh fuck [00:28] welp, there's another data race [00:30] really ? [00:35] davecheney: can we call you "the prolific dave cheney" from now on? [00:36] http://features.slashdot.org/story/15/11/18/1748247/interviews-alan-donovan-and-brian-kernighan-answer-your-questions <-- for those who have not seen it, dave is featured in a Q&A by D&K [00:40] \o/ [00:40] i have to be honest [00:40] i never read that post [00:40] it was full of toxic slashdotters [00:40] maybe I should try again [00:42] Nothing new but you are named as a proposer of the solution to version handling in go pks [00:43] orly [00:43] The rest of the questions are same old ones [00:43] i should *definitely* read that then :) [00:44] how often does the upgrade watcher cycle? [00:52] davecheney: I want to change s/Opener/APIOpener/ back, as envcmd.Opener doesn't really describe it, but envcmd.APIOpener does a bit more [00:52] davecheney: can you live with that? [00:53] thumper: sure [00:54] * thumper move back to sitting deck [01:12] wallyworld_: I'm back, free to chat whenever you're ready [01:13] axw: ok, i have another meeting in 15 and need to do some other stuff also so best to wait to our schedule time [01:13] wallyworld_: nps [02:36] thumper: menn0 https://github.com/juju/juju/pull/3785 [02:45] davecheney: shipit [02:46] ta [03:06] hey menn0, are you going to make the change to TestFilesystemInfo? or are you still waiting on input from mwhudson? [03:18] thumper: menn0 http://juju-ci.vapour.ws:8080/job/github-merge-juju/5520/console [03:18] here is a great example of why always using etsting.BaseSuite is a bad idea [03:19] in this case the test has failed because testing.Base suite also sets up some networking crap [03:19] even though the test is just checking did something get written to the log at trace level === natefinch-afk is now known as natefinch [03:20] it boggles my mind that we have tests that depend on the log output. [03:20] ...? [03:21] it makes sense to tese your logging isn't crap [03:21] it doesn't make sense to do that at a functional test [03:24] there are some very specific circumstances where you would want to unit test some logging, I can understand that. But it should be the exception rather than the rule.... and we have a ton of tests that'll fail if you disable logging (I know, I tried it back when I tried to make our failed test output not so horrific) [03:26] completely bonkers [03:27] we have a bad habit in juju tests of testing crap we really don't actually care about, so our tests are brittle, and break for ridiculous reasons, like fixing typos [03:28] for example [03:28] we have a method to set a global singleton to true, or false [03:28] and a third method that set it to false, but doens't log that's done anything [03:28] da fuq [03:29] and tests that rely on this behaviour [03:32] while i'm having a rant [03:32] i'm really pissed that all the test break if you don't have LXD installed [03:32] and not in a [03:32] c.FailNow("You don't have LCD installed") [03:32] * natefinch hangs head. [03:33] they fail because something in the guts of the code the test is calling is shelling out to /usr/bin/lxd [03:33] and trying to parse the output [03:33] really shit [03:33] Sorry, yeah, we should fix that. [03:34] I'll file a bug for it. Just didn't think of it, since we'd all installed LXD in order to be hacking on it. [03:34] thanks [03:35] i already logged a bug about the lxd provisioner failing horribly whne lxd is not installed [03:35] natefinch: open question: how did this pass in CI ? [03:35] / +build go1.3 [03:35] :p [03:36] the lxd library uses some tsl function that's only in go 1.3+ [03:36] tls [03:36] so the answer is... the tests don't get run in go 1.2... I don't know what happens on the wily build machine [03:37] eric and wayne were handling most of that [03:37] (and wily is still non-voting due to go 1.5 problems) [03:38] natefinch: that means the tests are not being run by the landing bot [03:39] davecheney: yes [03:39] bummer [03:39] indeed. We all wear these now: https://cdn.shopify.com/s/files/1/0287/4374/products/DSC_8422_1024x1024.jpg?v=1411004337 [03:40] natefinch: should we buy those for the sprint? [03:40] i'll have one in large thanks [03:40] make mine extra black [03:40] like my soul [03:40] lol [03:40] davecheney is so emo [03:40] pain, etc [03:42] also, this makes me so very sad: [03:42] /home/nate/src/github.com/juju/juju$ grep -r "ErrorMatches" . | wc -l [03:42] 2848 [03:43] are there 2800 places where we care what the (more or less) exact error message is? I doubt it. [03:44] make that two jumpers [03:49] does anyone even look at the bugs filed on charms? [03:51] the wordpress charm is fundamentally broken for people who actually want to like, you know, use wordpress (not just show the login screen). I reported a bug about it 2 months ago and the bug hasn't even been triaged. [03:52] https://bugs.launchpad.net/charms/+source/wordpress/+bug/1491995 [03:52] Bug #1491995: Can't upload files to wordpress [05:23] anybody? [07:07] Bug #1488523 changed: Azure provider attempts to reuse dying vnet [07:07] Bug #1494002 changed: azure deployment failure with mem constraints === Michael_temp is now known as private01 [10:01] dimitern: stdup? [10:02] voidspace, omw, 2m [10:30] frobware: https://github.com/juju/juju/pull/3788/files [10:30] dimitern: https://github.com/juju/juju/pull/3788/files [10:42] frobware: I do have test failures on that rebase branch but they're "weird" [10:43] frobware: can you run tests on that branch please [10:43] voidspace, yep [10:54] frobware: all tests are passing so far on maas-spaces (as is) for me, so new failures are probably genuine :-/ [10:54] might be worth you re-trying the rebase locally and seeing if you get the same result [10:59] voidspace, that's what I was initially doing. [11:02] maas-spaces is all pass for me [11:03] trying master now [11:12] voidspace, I rebased maas-spaces against master; make check works for me too [11:14] frobware: I'm getting test failures on master - at least some lxd related [11:14] so that maybe the problem [11:15] voidspace, scrolling back through the chat there is mention of tests failing & (lack of) lxd installation [11:16] voidspace, but given our rebase attempt they should fail in the maas-spaces branch as well... ? [11:16] container_initialisation_test.go fails on master for me too [11:16] as far as I can see the tests that fail on the rebase fail on master for me too [11:17] so should be ok [11:17] other than tests failing on master being a problem! [11:17] I'm installing lxd... [11:25] voidspace, master is OK for me. I don't have lxd installed. [11:49] frobware: I'm on wily, I wonder if that's it [11:49] installing lxd doesn't make a difference [11:56] frobware: so do tests pass for you on my rebase branch? [11:56] if so I'll merge it [11:57] voidspace, let me try your branch [12:01] voidspace, ah. so wily will be go-1.5. [12:02] frobware: no, I'm using go 1.3.3 [12:02] I thought lxd provider required go 1.5 [12:02] hmmm, it used to - maybe that changed [12:02] it all builds ok without 1.5 [12:03] I have a wily plaground and `apt-get install golang-go' gives me 1.5. Are you building or using your own go build? [12:04] frobware: yes :-) [12:12] dimitern: frobware: so some of my failing tests on *my WIP branch* (not the rebase branch) were because with the addition of Spaces to NetworkingEnviron the dummy provider was no longer "supporting networking" [12:12] dimitern: frobware: so that's an easy fix [12:12] thank goodness [12:12] voidspace, alright! [12:13] voidspace, nice :) [13:41] goddammit just as I am properly getting into something I have to go out for a while -- back this evening [13:41] gluck [14:39] does anyone know if a maas can be done whith storage from a nas instead of having each machine have storage? [14:56] I am maas newbie, but I guess you would need to create a custom image that use the NAS as remote filesystem === BradCrittenden is now known as Guest42648 [15:14] Bug #1518347 opened: LXD tests require LXD, but there's no friendly "you don't have LXD" error message [15:20] Bug #1518347 changed: LXD tests require LXD, but there's no friendly "you don't have LXD" error message [15:23] Bug #1518347 opened: LXD tests require LXD, but there's no friendly "you don't have LXD" error message [15:26] Bug #1518347 changed: LXD tests require LXD, but there's no friendly "you don't have LXD" error message [15:43] I have lxd installed, but get lots of test failures "can't connect to the local LXD server" [15:43] (permission denied) [15:52] voidspace: weird. we're in a meeting now. what series are you on? [16:03] voidspace, you still about? did we rebase? [16:37] o/ - can someone lend a hand in clueing me in as to the purpose/case for JUJU_ACTION_TAG is? [16:37] it appers to be a copy of JUJU_ACTION_UUID with the word "action" prepended, and thats the only discernable difference [17:57] lazypower, um, don't use it [17:57] lazypower, we probably never should have exposed it [17:57] lazypower, tags are meant to be homogeneous ids for the api layer [18:02] natefinch: I'm on wily [18:03] frobware: I didn't merge - I don't get a clean test run on master so it's not easy to tell if the merge works [18:03] voidspace, ack [18:06] voidspace, upstream/maas-spaces rebased seem to be OK though. [18:12] frobware: did you try my branch? [18:12] frobware: if tests pass on that I'll merge it [18:12] frobware: https://github.com/juju/juju/pull/3788 [18:16] voidspace, well they did about 10am... :) [18:16] voidspace, I need to EOD in a bit. If you don't need this until Monday I could double-check again and then merge. [18:17] frobware: hah, I was waiting for you to tell me you'd tried it... [18:17] :-p [18:17] voidspace, I did at 10am. [18:17] frobware: no hurry, Monday is ok [18:17] frobware: although I haven't touched that branch since, so if they passed then they'll pass now [18:18] voidspace, I subsequently noticed later in the day I was using go 1.5 for that build so I should try again with 1.2. [18:18] frobware: I think the CI bot will try with 1.2 [18:18] frobware: I've hit $$merge$$ anyway [18:18] good point [18:18] natefinch: I'm using wily but using go 1.3 === akhavr1 is now known as akhavr [18:40] voidspace, lxd needs at least 1.3, for the lxd provider juju builds with what is in wily 1.5 [18:41] voidspace, not sure if that helps [18:47] go 1.3 should work fine. I usually build with 1.4.2, just because it's the newest non-1.5 build (and the 1.5 builds are significantly slower). [18:47] wonder if it's a wily thing [18:51] alexisb: thanks, I have 1.3 [18:51] it builds fine but I have test failures [18:52] voidspace, related to this: https://bugs.launchpad.net/juju-core/+bug/1517748 [18:52] ?? [18:52] Bug #1517748: provider/lxd: test suite panics if lxd not installed [18:52] alexisb: see my comment on that bug [18:53] aah I hadn't refreshed [18:53] frobware: test failures for that merge, hard to see if they're the same or related or spurious or what [18:53] frobware: http://juju-ci.vapour.ws:8080/job/github-merge-juju/5525/console [18:53] voidspace: we will be addressing that next week. are you sure you're on tip of master? [18:53] katco: I was a couple of hours ago... [18:54] katco: I have lxd installed and still have test failures though [18:54] katco: so not *entirely* sure it's the same bug [18:54] voidspace: is your user in the lxd group? [18:55] katco: yep [18:55] I gotta go, 7pm here and guests arriving [18:55] EOW, sorry [18:55] voidspace: np ty for input [18:55] voidspace: we'll get it fixed enxt week [18:56] katco: cool, thanks [18:56] cherylj, I pretty sure I understand why the bond and juju-br0 don't work. Need to futz with our script a bit - [18:57] frobware: ah, good to hear! [18:57] * frobware EOF -- that's End of Friday. :) [19:09] frobware: have fun! [19:18] natefinch: hey you're going to make sure the min-version branch has a bless before you land it into master, correct? [19:19] katco: yeah, good point. I'll move it into a feature branch so it can get CI'd. [19:19] natefinch: k [19:20] katco: it's really not a hugely sweeping change by any stretch of the imagination, but I think it pays to be overly careful given the problems of the recent past. [19:20] natefinch: not even a question of impact, it's just policy [19:23] katco: fair enough [19:36] natefinch: talk to mgz or sinzui to see if you can get the next ci run to hit it [19:37] natefinch: I can push your branch into the juju namespace if you need that [19:46] mgz: that would be very helpful [19:47] mgz: it's the minver branch on github.com/natefinch/juju [19:50] natefinch: doing it [20:06] natefinch: you are in and up next [20:07] mgz: thanks [20:11] rogpeppe3: are you actually here? [20:35] gah, I really hate it when people leave juju/juju in a state where it can't build against master of its dependent branches. it means I have to fix the build before I can land anything [21:48] well, not getting juju min version in today it seems. [21:49] natefinch: =/ b/c of the libs ? [21:50] katco: yeah. I can land the stuff in charm.v6, but the stuff in master won't be able to go in without fixes first... axw has what might be the fixes on a personal branch, but they're slightly above trivial, so I wouldn't want to touch them without talking to him [21:51] natefinch: makes sense. well at least you have ci building against your branch [21:52] katco: yeah, but I don't think I can make my branch compile for CI, even [21:54] katco: it works locally because I can hack things so that my commits are against an earlier version of charm.v6, but it can't be set up to run in Ci where it runs godeps [21:55] natefinch: oh. doh. [21:55] natefinch: well, once your fix to charm.v6 lands, you can patch your feature branch to use that version and run ci, yeah? [21:56] katco: this is the whole problem... landing the code in charm.v6 means landing after those breaking changes [21:57] katco: I fcan make it work locally because I can make a local branch off an earlier version of charm..v6 that excludes the breaking changes... but I can't create that kind of a branch in github.com/juju/charm [21:57] (I mean, In theory an admin could, but it would be a heinous thing to do) [21:58] I'll update my email to indicate that it's not just master but my feature branch that has this issue. [21:58] natefinch: sorry, not sure i understand. you're landing some fixes in charm.v6, why couldn't you patch up your feature branch to be compatible with those changes? [21:59] katco: oh, I could. They're just slightly beyond trivial... so I wouldn't want to get my feature branch in a bad state if I mess up those fixes [21:59] katco: if I talk to andrew and confirm that I just need to cherry picky a couple commits, then yeah, I can do that. [21:59] natefinch: oh i think i see.. you'd have to apply axw's changes to your feature branch and that is non trivial [21:59] correct [21:59] gotcha [22:00] natefinch: can you rebase your feature branch to a blessed version of master? [22:00] natefinch: at least then you've utilized your "dead time" and tested up to that point [22:01] katco: yes, but the problem is in charm.v6, so it would still be there. The solution could be to make a feature branch in charm.v6 off the earlier blessed version [22:01] natefinch: yeah, not a bad use of the "dead time". also looks like mgz just sent an email [22:01] natefinch: might have some other stuff to fix up [22:01] gah, yeah [22:02] just read it [22:02] ok, I'll work on that tonight. Gotta run to make dinner now. [22:02] natefinch: have a good thanksgiving [22:02] katco: you too! === natefinch is now known as natefinch-afk [22:08] hey, do we have a way of flagging long-running tests as not-for-normal-runs? [22:17] fwereade: I believe we do [22:44] perrito666, do you happen to know what it might be? [22:45] fwereade: nope, katco told me a gazillion times and I forgot a gazillion + 1 [22:45] * fwereade glances hopefully at katco [22:45] ...although I've almost convinced myself that it's ok to patch just this once [22:46] fwereade: o/ is this what you're looking for? https://github.com/juju/juju/blob/master/featuretests/package_test.go#L15 [22:46] katco, I was wondering if there was something that'd work globally [22:47] katco, but I am being silly anyway [22:47] fwereade: oh, no i don't think we've begun flagging certains tests as long or short yet [22:47] katco, cheers [22:48] fwereade: happy friday to you :) [22:49] katco, and to you :) [22:49] katco, by the way, do you recall state/testing.NotifyWatcherC et al? [22:50] fwereade: not at all... what is it? [22:50] katco, it's a test harness for a watcher [22:50] katco, was wondering, if you'd used it, whether you thought it was a sane interface [22:51] fwereade: i haven't used it, sorry :( [22:51] katco, no worries :) [22:51] fwereade: however, i do almost have a working charm that starts an emacs web server. so that must count for something. [22:51] katco, sweet :D [22:52] fwereade: this is going to be such an unholy abomination. i love it. [22:53] katco, in principle, I heartily approve, but please be ready to pull the plug in case skynet [22:53] fwereade: it's too late, dear william. it's already gotten to me. [22:54] * fwereade welcomes our new robots overlords [22:56] uh, robot overlords? sweet I could do with a break on all the thinking stuff [22:57] perrito666: i was very happy to see you still using emacs in your tweet :) [22:57] * perrito666 is as close to burnout as one can be before being actually burnt [22:57] katco: it totally stuck this time, Just need to get a bit more proficient, but I am on a good path [22:57] perrito666: good to hear [22:58] perrito666: the packaging system + use-package is a great combination [22:58] I might use a couple of hours of my holidays to learn a few tricks more [22:58] and perhaps tune it a little more [22:59] perrito666: this is a fun site: http://emacsrocks.com/ [22:59] I see you are not only working in new clients but also in retention :p [23:00] my next conquest is to try to get a marker in the lines that are not git committed yet, which is a feature I miss a lot [23:00] perrito666: i must do as the master suggests. [23:03] perrito666: there might be something better, but a quick scan: https://github.com/nonsequitur/git-gutter-plus [23:04] nice, actually I am going for something simpler, I am completely happy to use the command line for git :p [23:05] I just want the +, ~ and - next to the line number [23:05] Ill try all options I find :D [23:05] perrito666: i highly recommend magit as a better way to git [23:05] uhh, sweet, new github design [23:06] perrito666: it's so nice. remind me to show you in dec. if you'd like [23:06] katco: after years of git, I believe a better way to git is not to git at all :p [23:06] katco: totally, I always dig learning new tools [23:06] magit makes so many things much simpler [23:06] i think any gui for git might [23:06] but it's nice that it's built into emacs [23:07] katco: Ill be honest here, the only git guis that I liked where for OSX [23:07] and iirc, commercial :p [23:08] magit might change your mind. it is really nice. [23:08] git and diff displaying in general can benefit a lot from modern graphic paradigms [23:08] but hey, in linux Ill take what I can get :p [23:09] perrito666: with magit, you have a file you need to merge, you cursor down over the file, hit e and you get a side-by-side with merged at the bottom [23:09] red/green highlighting, etc. [23:10] * perrito666 watches screenshots [23:11] I really dig stuff like this https://deeperdesign.files.wordpress.com/2011/05/screen-shot-2011-05-27-at-12-59-14-pm.png [23:12] perrito666: here's what ediff looks like: http://www.tldp.org/LDP/LG/issue27/gx/marsden/ediff-merge.gif [23:12] perrito666: and then you hit a or b depending on which side you want merged [23:12] is that motif? [23:12] lol no idea [23:13] the bottom is the merged version [23:13] blast from the past :p but hey, I have seen worse, ill give it a shot [23:33] Bug #1518480 opened: Invalid storage prevents machine removal [23:33] Bug #1518482 opened: Invalid storage returned by storage-list causing charm breakage