[00:21] sinzui, katco, alexisb - here's the PR for reverting the rename of environments.yaml: http://reviews.vapour.ws/r/3629/ [00:22] rb, not pr. whatever :P [03:52] cherylj, sinzui: failure on machine-dep-engine appears to be bad-record-mac [03:53] thumper: yeah, I saw that. [03:53] If other tests pass, I'd say merge away [03:54] looks like the maas tests (and a few others) are still running / queued [03:57] yeah... not final yet === urulama__ is now known as urulama [09:12] dimitern: ping [09:13] dimitern: do you know the root cause of the trusty deploy CI failure we're seeing [09:13] dimitern: "dpkg-deb: error: --extract needs a target directory" [09:13] dimitern: it looks like a build failure [09:14] mgz: ping [09:15] we have two failures the same on xenial [09:16] a lxd-deploy and aws-deploy-trusty [09:17] voidspace, hmm let me have a look [09:17] dimitern: http://reports.vapour.ws/releases/3533 [09:20] voidspace, mgz, it looks like it failed to build xenial deb package, but the job still succeeded: http://data.vapour.ws/juju-ci/products/version-3533/build-binary-xenial-amd64/build-130/consoleText [09:22] debian/rules:26: recipe for target 'binary' failed [09:22] dh_install: juju-core missing files: src/github.com/juju/juju/cmd/plugins/juju-backup/juju-backup [09:22] dimitern: this doesn't fail for master though it seems [09:24] cmd/plugins/juju-backup is also missing on master [09:24] maybe a CI rule has been updated for master [09:25] voidspace, it seems the debian/rules for master and other branches are different [09:26] which tells me they fixed it for master, but apparently each branch uses its own rules [09:29] dimitern: ok, cool [09:29] dimitern: so it's their fault :-) [09:29] mgz: ^^^ [09:41] :) [09:46] dimitern: where did you check the debian rules? [09:47] voidspace, in the log - check for error 2 [09:47] at the end of that "successful" job log I pasted earlier [09:48] dimitern: thanks [10:21] hey guys, sorry I missed the standup. I am around, just spaced the actual event. [10:21] Anyone remember who setup reviewboard? I'm noticing that Tycho has submitted stuff to Github but it doesn't seem to end up in RB as well. [10:22] https://github.com/juju/juju/pulls/4191 [10:22] jam: it was ericsnow who setup reviewboard [10:22] we're still talking FWIW [12:55] Jam: i believe review board requires for the user to do the link with github account to work [12:55] Btw did you see axw email regarding xdg? [13:22] dimitern, voidspace, dooferlad: please could you take a look at my (mostly complete) bridge script that allows an explicit interface to be specified: http://reviews.vapour.ws/r/3630/ [13:23] dimitern, voidspace, dooferlad: if we're happy with the approach I'll take maas-spaces script to 1.25 and test accordingly. [13:27] frobware, sure, looking [13:33] dimitern, what's the right answer for this? http://pastebin.ubuntu.com/14671696/ [13:42] frobware, reviewed [13:42] frobware, re the paste - the only difference is no auto eth1 in initial, right? [13:43] frobware, a guess the answer should be "if it worked before the script shouldn't break it" :) [13:44] dimitern, correct. [13:44] s/script/script, running it/ [13:44] dimitern, I think we should mimic what we got, hence I just pushed an additional test to cover this case. [13:45] frobware, +1 [13:46] dimitern, I haven't see that in the wild, but some time ago I put a check in to only add the 'auto' stanza if there was one before. That all happened in the _bridge() functions but in the main loop we skip over interfaces and this now gets missed, hence the additional test. [13:48] dimitern, bridge-prefix is a default so it will always have a value even if bridge-name && interface-to-bridge are specified. [13:48] frobware, right - the more cases we cover in tests, the better I think [13:48] frobware, ah, well I missed that [13:48] dimitern, yeah, the new functionality is a little at odds with the existing behaviour. [13:48] frobware, feel free to ignore the second issue then :) [13:56] wow go 1.6b2 got smarter error messages :) [13:56] ../../state/watcher.go:2395: doc.MachineId undefined (type networkInterfaceDoc has no field or method MachineId, but does have MachineID) [13:58] [13:58] voidspace, ping [13:59] mgz: ping [14:03] frobware: is maas meeting on? [14:03] frobware, dimitern, voidspace: https://plus.google.com/hangouts/_/canonical.com/maas-juju-net [14:03] dooferlad: just grabbing coffee [14:03] be there in 2mins [14:24] does anyone know if we have specs to add relations to tabular format? [14:44] destroy-environment --force is gone and destroy-controller has no force option but just hangs... :-( === coreycb` is now known as coreycb [14:45] mgz: sinzui: ping [14:45] hi voidspace [14:45] sinzui: hey, hi [14:46] sinzui: maas-spaces CI jobs are failing the xenial tests for us because they're looking for juju-backups [14:46] sinzui: and it looks like this is configured in debian rules per CI task [14:46] sinzui: as the same doesn't happen on master [14:46] voidspace: when you need to use the force in that situation, use kill-controller, (a safe force) [14:46] sinzui: ah, thanks [14:47] voidspace: I fixed packaging last Friday to not look for a plugin [14:47] sinzui: that worked [14:47] voidspace: but lt me see if xenial has a different rule [14:48] sinzui: we have a fail on monday: http://reports.vapour.ws/releases/3533 [14:48] sinzui: aws-deploy-trusty...xenial and lxd-deploy-xenial both failed for that reason [14:49] voidspace: the lxd-deploy-xenial issue was a temporary network failure [14:50] ah, my apologies then [14:50] I'm fairly sure about the other one :-) [14:52] voidspace: we can see next test of the branch did have a good network http://reports.vapour.ws/releases/3542 [14:52] sinzui: ah, awesome - I hadn't seen that [14:52] sinzui: appreciated [14:52] still cursed though [14:52] *sigh* [14:53] frobware: dimitern: dooferlad: new CI run for maas-spaces http://reports.vapour.ws/releases/3542 [14:54] frobware: dimitern: dooferlad: those tests that failed now pass (rules fixed), but we have a wily unit test failure [14:54] I'm running wily and I don't think that test fails for me [14:54] investigating [14:54] frobware: dimitern: dooferlad: http://reports.vapour.ws/releases/3542/job/run-unit-tests-wily-amd64/attempt/1210 [14:57] voidspace, that looks like flakiness [14:59] voidspace, hmm maybe not - I can see it here on wily [14:59] dimitern: cool, we'll see in the next test run then [14:59] it's running for me [14:59] slowly... [14:59] heh [15:00] dimitern: so with my code that blocks logins until space discovery is complete [15:00] dimitern: bootstrap completes - but shows that error [15:00] dimitern: so bootstrap doesn't wait, it just appears to error out [15:00] so more work needed [15:00] frobware: ^^ [15:00] dimitern: that test passes for me on wily [15:00] hmm, I may not be up to date though [15:01] dimitern: right, I've pulled in the merge from master now [15:01] I wonder if that is the cause [15:13] Bug #1526072 changed: Juju-deployer 0.6.0 and juju-core 1.25 - Build doesn't time out and keeps running until aborted

[15:13] voidspace, I've run cmd/jujud/agent tests a few times - it fails some times, but no more than usual [15:17] dimitern: right, passes for me every time so far [15:23] I am seeing random failures while running test locally due to mgo not cleaning up (on several branches) [15:23] is there something I need to do locally to make sure things are getting cleaned up? [15:24] Nope, the errors you see might be the cleanup errors after a panic from something else [15:24] Tmp tends to fill out with mgo stuff though so you might want to flush that [15:24] alexisb: there really shouldn't be! [15:24] I gotta go pick up the daughter from school, sorry [15:25] ok, just seems test on master have gotten really flaky [15:25] around cmd/jujud [15:26] Alexisb which are failing? [15:26] agent and unit [15:26] Pastebin? [15:26] github.com/juju/juju/cmd/jujud/agent/... [15:27] fails consistently [15:27] I did stumble upon agent failing lately [15:29] alexisb, try GOMAXPROC=1 go test -check.v github.com/juju/juju/cmd/jujud/agent/... to see if it helps [15:38] GOMAXPROCS is not the cause. I run with GOMAXPROCS=8 all the time, and don't have many consistent failures. [15:40] alexisb: what version of Go? (run go version) [15:40] voidspace, dooferlad, frobware, care to review a -3080 line diff :) ? http://reviews.vapour.ws/r/3631/ [16:23] frobware, voidspace, dooferlad, any of you still there? [16:25] dimitern, otp [16:25] ok [16:25] just pestering for a review btw [16:25] dimitern: can take a look in a moment [16:26] dooferlad, cheers [16:32] dimitern, one question I have is the "risk" of landing your change in light of merging back to master. just curious... [16:32] dimitern, I don't want to derail the opportunity to be back in master. [16:33] natefinch, I am running 1.5 [16:33] dimitern, ooh. Now that I've actually looked at your PR it's all deletes.. \o/ [16:34] frobware, yeah - for a change :) [16:37] dimitern: +1 delete all the old things! [16:37] dimitern, so my question still stands though: do you see any risk for CI tests, stopping a merge into master. It's likely that master may be blocked for a few days so that we can land. [16:39] frobware, I'm not sure about that PR causing CI tests to fail - not that I know of, and since it's dropping code which was not running for quite a while, I think we're safe [16:39] dimitern, k [16:40] dimitern, just trying to avoid another 24 cycle on getting maas-spaces into master [16:41] alexisb: 1.4 is more reliable.... 1.5 has some known problems with juju still, last I heard [16:43] alexisb, I have pre-built deb of go 1.4.3 if it helps. http://178.62.20.154/~aim/go1.4.3_1.0.0-14_amd64.deb - it installs into /usr/local/go1.4.3 [16:56] natefinch, perrito666 this is the failure I am seeing consistently with latest master and golang 1.5: [16:56] https://pastebin.canonical.com/148393/ [16:57] alexisb: yeah, pretty sure I was seeing the sockets thing with go 1.5 too [16:57] yup, that is go 1.5 [16:57] I have smae issues only in my go1.5 machine [16:57] yay! [16:58] sort of :/ [16:58] any idea if it is addressed with 1.5.2+? [16:58] alexisb: I don't think it's a Go bug, I think it's a juju bug, but I haven't looked super deep into it. [16:58] natefinch, ok [17:02] ashipika, were you seeing that sockets error as well? ^^ (https://pastebin.canonical.com/148393/) [17:03] cmars: yes, i was [17:03] cmars: in cmd/jujud/agent package [17:04] cmars: and golang 1.5.3 [17:05] how would you call both ends of a relation ? [17:06] I am trying to add relations to tabular status and finding this problem [17:08] cmars, alexisb: https://pastebin.canonical.com/148394/ [17:10] ashipika, that is the same failure I am seeing [17:10] trying with go 1.4 now just to see if it is different [17:15] ericsnow: is your branch pushed in a good state for me to rebase? I just got my branch all building, tests passing, etc. [17:15] natefinch: working on it [17:21] natefinch: should be good now [17:25] ericsnow: cool [17:46] so it looks like go 1.4 is allowing the agent tests to pass but I am still seeing failures on master for the status test [17:46] anyone else seeing failures on master for the status tests? [18:13] Bug #1538241 opened: 2.0-alpha2 stabilization

[18:51] dammit [18:52] natefinch: ? [18:53] machine was frozen when I came back and now its doing that "running in low graphics mode" crap [18:53] natefinch: something is very wrong with some part of your computer [18:53] did you ever replace the battery? [18:53] yes I did [18:57] bad when I can't even get to a command prompt [18:59] ahh Ctrl alt f1 works at least [19:00] ericsnow: natefinch: finally finished revising the user-stories. i posted comments in the doc, but i'd like to review the diff with you two after i do lunch [19:00] katco: k [19:01] katco - hopefully I can figure out my laptop problems before then [19:01] natefinch: if not, do you have any other machine/phone you could attend on? i need you there [19:02] natefinch: also, is this going to affect your ability to finish your card? [19:04] I can attend on my tablet if needed. yes it'll screw up getting the card done, but I'm hopeful I can fix it [19:04] natefinch: k, keep me posted please. [19:04] * katco lunches [19:06] hmmm less hopeful now [19:09] natefinch: need a hand? [19:11] perrito666 yes [19:11] natefinch: lets privmsg [19:11] k [19:17] cherylj, ping; I'm around for 30 mins. anything I can do or help with regard to maas-spaces CI builds... ? [19:38] it would be really nice if linux would notice that all possible video drivers have been blacklisted and give me some kind of option to de-blacklist one of them [19:40] perrito666: thanks for the help... I figured it out. I foolishly tried to enable the nvidia driver, but it evidently didn't actualy take until I rebooted after ubuntu froze... but when it did that, it blacklisted the nouveau driver... === natefinch_ is now known as natefinch [19:41] but why would the nvidia kernel fail? [19:41] perrito666: because I'm special like that [19:43] dont worry, this is the year of linux on the desktop :p [19:49] frobware: I haven't looked at any of the failures, do you guys need me to? [19:49] (sorry, was out for a walk before the rain starts) [20:27] ok, something weird just happened [20:27] I ran a whole test suite and didnt got spammed by mongo === urulama is now known as urulama__ [20:34] natefinch: meeting time [20:36] katco: oops, sorry, coming [21:14] ericsnow: did you say you had a charm that utilized resource-get? [21:14] natefinch: yeah, yours :) [21:15] natefinch: check it out in my branch in test-charms/.../starsay [21:16] ericsnow: ahh, I see it. linux's default directory listing order always messes me up [21:18] yeah, yeah, blame the OS [21:19] perrito666: it's not my fault they decided to go top to bottom then left to right, rather than left to right, top to bottom [21:47] Bug #1538303 opened: 1.25.0: bootstrap failure - WARNING discarding API open error: EOF

[21:50] Bug #1538303 changed: 1.25.0: bootstrap failure - WARNING discarding API open error: EOF

[21:56] Bug #1538303 opened: 1.25.0: bootstrap failure - WARNING discarding API open error: EOF

[21:59] katco, ericsnow: trying to debug why the unit resources aren't getting returned when I call show-service-resources. I'm not sure resource-get is working correctly. [21:59] Bug #1538303 changed: 1.25.0: bootstrap failure - WARNING discarding API open error: EOF

[21:59] katco, ericsnow: but I have to run for a while for dinner. [21:59] natefinch: k === natefinch is now known as natefinch-afk [22:02] Bug #1538303 opened: 1.25.0: bootstrap failure - WARNING discarding API open error: EOF

[22:43] katco: FYI, I have 6 patches up on RB now for the resource-get stuff [22:45] ericsnow: awesome! in a meeting, i'll try and review tonight [22:45] katco: note that tests bloat the line count substantially on a couple of those patches :) [22:47] ericsnow: lol [22:47] katco: I'm going to land them all in 1 PR once they all pass review [22:48] ericsnow: that's clever. good idea [22:48] katco: yeah, rbt post --parent was my friend here [23:30] katco: so we'll revisit our iteration backlog after standup tomorrow? [23:31] ericsnow: yes [23:31] katco: sounds good [23:31] ericsnow: sorry not sure i can get to the reviews tonight. need to start dinner and the family gets home soon [23:31] katco: np [23:32] katco: you wrote a decent chunk of it :) [23:41] perrito666: we will end standup, can't hear you [23:42] wallyworld: yes, apparently my 3g network sucks at upload [23:42] perrito666: mongo3 will be next week's focus [23:42] after we get the next alpha out [23:42] wallyworld: my question was, regarding master [23:42] my last branch, merged with master, when running the test suite [23:43] it doesnt spam syslog [23:43] that is new [23:43] perrito666: rsyslog is removed from juju core, you talking about mongo syslog spam? [23:44] wallyworld: that is a problem from mongo not syslog [23:44] it was at least :p [23:44] you know, when running the full suite, there was a spam to all the terminals at one point [23:44] sure, but what i meant was, there will be no juju core syslog spam now [23:45] I assumed that spam was being done by mongo, not juju [23:45] i thought it was core, but not sure now. anyway, mongo2 still being used for now [23:46] wallyworld: k [23:46] bbl, ill go try to resurrect internet