| davecheney | suddently, https://github.com/juju/juju/pull/4212 | 00:30 |
|---|---|---|
| waigani | yeah, just hit the same :/ | 00:33 |
| davecheney | i guess it's time to apply all those security updates i'd been avoiding | 00:41 |
| thumper | fark... | 00:53 |
| thumper | who would have thought that the design on state/status.go would come back and bite? | 00:54 |
| thumper | / statusDoc represents a entity status in Mongodb. The implicit | 00:54 |
| thumper | / _id field is explicitly set to the global key of the associated | 00:54 |
| thumper | / entity in the document's creation transaction, but omitted to allow | 00:54 |
| thumper | / direct use of the document in both create and update transactions. | 00:54 |
| thumper | gee, we omit a vital piece of infomration to make a few other places of code easier... | 00:55 |
| thumper | I can't read all the status docs at once because we can't link the doc to the souce | 00:55 |
| thumper | stupid | 00:55 |
| thumper | I'll have to read out as maps | 00:56 |
| * thumper loos for that pic he saw davecheney use | 00:56 | |
| davecheney | google "burning bag of shit gif" | 00:57 |
| thumper | https://twitter.com/davecheney/status/691479334867988480 | 00:58 |
| * thumper throws in the towel and eats lunch | 00:59 | |
| axw | wallyworld: when you can, can you please take a look at http://reviews.vapour.ws/r/3654/ | 02:51 |
| axw | working on updating the bootstrap code now | 02:51 |
| wallyworld | sure, i'll just finish soemthing | 02:51 |
| davecheney | is github still down ? | 02:59 |
| thumper | it was | 02:59 |
| thumper | I'm finding it up again | 03:00 |
| davecheney | thumper: https://github.com/juju/juju/pull/4219 | 03:01 |
| davecheney | attempt number 2 to backport the leadership worked fix | 03:01 |
| davecheney | i modifeid the tests to work with the testing utility packages that exist on 1.25 | 03:01 |
| davecheney | which limited the scope of the backport | 03:01 |
| thumper | k | 03:03 |
| thumper | davecheney: what was the change between this backport on the other? | 03:04 |
| thumper | wow... first clean run through all the tests for a while with no intermittent failures | 03:39 |
| mup | Bug #1538868 opened: kill-controller failed: sudo: switch: command not found <ci> <kill-controller> <local-provider> <regression> <juju-core:Incomplete> <juju-core api-command-rename:Triaged> <https://launchpad.net/bugs/1538868> | 03:52 |
| waigani_ | cherylj: merge done. Very non trivial, sent you an email. | 03:56 |
| cherylj | waigani_: thanks. If you can merge your changes into the machine-dep-engine branch, it'll be queued for another CI run. | 03:57 |
| waigani_ | cherylj: okay will do | 03:59 |
| cherylj | thanks, waigani_, I appreciate the help. We've been trying to get that branch merged for days! | 03:59 |
| waigani_ | cherylj: Yep it's the syncing hell. Not a nice place to be. | 04:01 |
| cherylj | I feel bad for everyone who has to rebase after the api-command-rename branch lands | 04:02 |
| cherylj | that's going to be absolutely awful | 04:02 |
| waigani_ | yeah true ... I should get a move on with these workers! | 04:02 |
| waigani_ | ;) | 04:02 |
| natefinch | damn I hate that we sling around IDs as strings everywhere. Just wasted multiple hours because I swapped a serviceID and a unitID. We even *have* type-specific Tags to avoid that :/ | 04:12 |
| * thumper is about to bootstrap for the first time in freaking ages | 04:24 | |
| thumper | will try lxd first | 04:24 |
| * thumper wonders what he needs | 04:24 | |
| natefinch | thumper: go 1.3+, the lxd apt package.... and there's a couple manual steps before bootstrap | 04:24 |
| thumper | got go | 04:24 |
| thumper | got lxd | 04:24 |
| thumper | what are the manual steps? | 04:25 |
| * thumper wonders why we need manual steps | 04:25 | |
| davecheney | thumper: i modified leadership/fixture_test.go to work with the older version of the juju/testing.Clock type | 04:26 |
| thumper | davecheney: is that it? | 04:26 |
| davecheney | I have to modify testing.Clock as well to fit the test | 04:26 |
| davecheney | yup | 04:26 |
| thumper | shipit (when unfroze)( | 04:26 |
| davecheney | someone refactored testing.Clock after 1.25 | 04:26 |
| thumper | hazaah | 04:27 |
| davecheney | so I just applied as much of the refactoring to the 1.25 testing.Clock type to get the tests to pass | 04:27 |
| natefinch | thumper: the steps are spelled out in the environments.yml if you juju init.... but assuming you already have an environments.yml.... if you haven't logged out since installing lxd, you have to run newgrp lxd ... that's just lxd setup | 04:27 |
| natefinch | thumper: and then - | 04:28 |
| natefinch | lxc remote add images images.linuxcontainers.org | 04:28 |
| thumper | I've read the help | 04:28 |
| natefinch | lxd-images import ubuntu --alias ubuntu-trusty | 04:28 |
| thumper | I'm wondering why the lxd provider doesn't import the images by itself | 04:29 |
| davecheney | hold your breath people, i'm hitting merge | 04:30 |
| natefinch | thumper: *shrug* it's kinda slow, but yeah, it would be nice to remove any manual steps, especially if we expect this to replace the local provider | 04:30 |
| thumper | which I thought we were doing for 2.0 | 04:31 |
| natefinch | I have not heard anything concrete about removing local for 2.0 | 04:31 |
| * thumper bootstraps his first lxd env since 2.0 | 04:32 | |
| cherylj | natefinch: yes, we are going to be removing the local provider for 2.0 | 04:32 |
| natefinch | cherylj: awesome. I hope moonstone isn't on the hook to do it. | 04:32 |
| cherylj | ha, I think the pleasure is going to tanzanite | 04:32 |
| natefinch | as much as I'd love to rip out its beating heart, we're overcommitted just with resources | 04:33 |
| * davecheney cue dramatic chipmonk | 04:33 | |
| natefinch | we were overcommitted on resources BEFORE we lost wayne :/ | 04:33 |
| thumper | natefinch: FYI, everyone is overcommitted | 04:36 |
| thumper | somethings I think the powers that be think we work best that way | 04:37 |
| thumper | when the opposite is in fact true | 04:37 |
| natefinch | thumper: I'll let you know when I find a software company that doesn't continually overcommit. | 04:37 |
| thumper | hmm... | 04:40 |
| thumper | my attempt at getting a new environment with lxd is struggling | 04:40 |
| thumper | particularly adding machines to another environment | 04:40 |
| thumper | bah humbug, recreating with debug | 04:41 |
| axw | thumper wallyworld: do you recall the outcome of discussion at oakland about removing the live tests? | 04:46 |
| wallyworld | i think many were in facor | 04:46 |
| wallyworld | favour | 04:46 |
| wallyworld | since we have CI | 04:46 |
| wallyworld | and they are only for aws and openstack | 04:47 |
| axw | but no actual decision | 04:47 |
| wallyworld | um, not that i recall, but i reckon jfdi | 04:47 |
| axw | I'll send an email to juju-dev saying I'm doing it unless there's strong objections | 04:47 |
| thumper | hmm... | 04:48 |
| thumper | this time I am getting new machines | 04:49 |
| thumper | I wonder what happened last time | 04:49 |
| wallyworld | axw: keep the openstack test doubles though, just remove the actual live tests against a real cloud | 04:49 |
| axw | wallyworld: what does that mean? there's "local live", which runs the live tests against doubles | 04:50 |
| wallyworld | there's 3 types of test: local, loval live, and live | 04:50 |
| wallyworld | i think we can remove the live bits | 04:51 |
| wallyworld | the setup than runs the tests against an actual cloud | 04:51 |
| axw | wallyworld: so you want to keep local live? I don't think there's anything to gain from removing just the live ones then | 04:51 |
| wallyworld | from memory, the suites use a fixture which embed the same tests to be run either locally against doubles or live | 04:51 |
| wallyworld | if we remove the local live ones, there's nothing to replace then afaik (for openstack say) | 04:52 |
| thumper | hehehe | 04:52 |
| thumper | debug-log is so nice now | 04:52 |
| davecheney | boom | 04:52 |
| davecheney | merges pull request; drops mic | 04:52 |
| wallyworld | axw: so other providers would have been written with more unit test coverage, but openstack relied on the local / live test fixture i think | 04:53 |
| wallyworld | but i could be wrong, would need to check | 04:55 |
| axw | wallyworld: bleh, I'll leave it. I'm trying to avoid having to touch the live tests with credentials bits, unavoidable if they have to stay for local live | 04:55 |
| axw | wallyworld: the live tests aren't particularly exhaustive, and openstack at least has a bunch more local-only ones... but I cbf right now | 04:56 |
| wallyworld | axw: ok, i am just being cautious about removing test coverage. the actual live tests against a real cloud aren't run by default, only if a flag is passed to go test. but it would be nice to clean that stuff up at some point | 04:57 |
| davecheney | really 1.25 is going to live on for another 2 years | 04:57 |
| wallyworld | i know the openstack tests use hooks to test various failure sceanrios | 04:57 |
| davecheney | it's a tire fire | 04:57 |
| wallyworld | and deleting the tests will result in a loss of that coverage | 04:57 |
| thumper | and here is an exported model http://paste.ubuntu.com/14686150/ | 05:04 |
| thumper | for two machines... | 05:04 |
| thumper | I know it doesn't look a lot | 05:04 |
| * thumper is done for the day | 05:06 | |
| mup | Bug #1538742 changed: lxc containers stay pending when using maas provider and juju 2.0-alpha1 <cpec> <juju-core:Invalid> <https://launchpad.net/bugs/1538742> | 06:04 |
| mup | Bug #1538742 opened: lxc containers stay pending when using maas provider and juju 2.0-alpha1 <cpec> <juju-core:Invalid> <https://launchpad.net/bugs/1538742> | 06:16 |
| === urulama__ is now known as urulama | ||
| mup | Bug #1538742 changed: lxc containers stay pending when using maas provider and juju 2.0-alpha1 <cpec> <juju-core:Invalid> <https://launchpad.net/bugs/1538742> | 06:19 |
| wallyworld | axw: hey, if you get a chance at some stage today, there's a couple of PRs related to the api-command-rename branch. one small, one large. i hope that's the last major work that's needed besides smaller cleanup. the good news is that on what's landed to date, CI is happy | 06:57 |
| axw | wallyworld: ok. trying to get tests for bootstrap finished atm, will see how I go | 06:58 |
| wallyworld | sure, np. ty | 06:58 |
| wallyworld | i have soccer in about an hour anyway | 06:58 |
| voidspace | frobware: dooferlad: dimitern: another master merge http://reviews.vapour.ws/r/3661/ | 09:27 |
| dimitern | voidspace, cheers - looking | 09:28 |
| dimitern | voidspace, reviewed | 09:41 |
| voidspace | dimitern: thanks | 09:47 |
| === ionutbalutoiu_ is now known as ionutbalutoiu | ||
| dimitern | voidspace, jam, fwereade, standup? | 10:04 |
| voidspace | dimitern: sorry, omw | 10:05 |
| mup | Bug #1417526 changed: environs/bootstrap: EnsureNotBootstrapped should not rely on StateServerInstances <azure-provider> <bootstrap> <juju-core:Won't Fix> <https://launchpad.net/bugs/1417526> | 10:14 |
| axw | wallyworld: I'm not likely to be able to finish updating the bootstrap tests, but if you want to see what's there already, it's at https://github.com/juju/juju/compare/cloud-credentials...axw:cloud-credentials-bootstrap2 | 10:53 |
| axw | wallyworld: I've updated ec2, openstack, gce, and lxd so far | 10:53 |
| axw | wallyworld: other providers will be done in a followup | 10:53 |
| wallyworld | axw: ty, we can use what works to put the demo together and follow up with proper tests | 11:10 |
| perrito666 | morning | 11:48 |
| voidspace | dooferlad: ping | 12:59 |
| voidspace | dimitern: ping | 13:00 |
| voidspace | dimitern: dooferlad: unping | 13:01 |
| dimitern | :) | 13:02 |
| mup | Bug #1539116 opened: TestManageEnvironRunsPeergrouper panics <intermittent-failure> <panic> <juju-core:Triaged> <https://launchpad.net/bugs/1539116> | 15:23 |
| mup | Bug #1539116 changed: TestManageEnvironRunsPeergrouper panics <intermittent-failure> <panic> <juju-core:Triaged> <https://launchpad.net/bugs/1539116> | 15:26 |
| mup | Bug #1539116 opened: TestManageEnvironRunsPeergrouper panics <intermittent-failure> <panic> <juju-core:Triaged> <https://launchpad.net/bugs/1539116> | 15:29 |
| mup | Bug #1458585 opened: TestCommand fails <ci> <gccgo> <go1.5> <intermittent-failure> <regression> <test-failure> <wily> <xenial> <juju-core:Incomplete> <juju-core 1.23:Triaged> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1458585> | 15:35 |
| mup | Bug #1458585 changed: TestCommand fails <ci> <gccgo> <go1.5> <intermittent-failure> <regression> <test-failure> <wily> <xenial> <juju-core:Incomplete> <juju-core 1.23:Triaged> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1458585> | 15:38 |
| mup | Bug #1458585 opened: TestCommand fails <ci> <gccgo> <go1.5> <intermittent-failure> <regression> <test-failure> <wily> <xenial> <juju-core:Incomplete> <juju-core 1.23:Triaged> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1458585> | 15:41 |
| mup | Bug #1458585 changed: TestCommand fails <ci> <gccgo> <go1.5> <intermittent-failure> <regression> <test-failure> <wily> <xenial> <juju-core:Incomplete> <juju-core 1.23:Triaged> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1458585> | 15:45 |
| mup | Bug #1458585 opened: TestCommand fails <ci> <gccgo> <go1.5> <intermittent-failure> <regression> <test-failure> <wily> <xenial> <juju-core:Incomplete> <juju-core 1.23:Triaged> <juju-core 1.25:Triaged> <https://launchpad.net/bugs/1458585> | 15:54 |
| cherylj | hey frobware, I see you got some feedback on your changes for bug 1532167 | 15:54 |
| mup | Bug #1532167: maas bridge script handles VLAN NICs incorrectly <addressability> <maas-provider> <network> <juju-core:Triaged> <juju-core 1.25:In Progress by frobware> <https://launchpad.net/bugs/1532167> | 15:54 |
| cherylj | think it will be hard to address? | 15:54 |
| cherylj | hey thomnico - were you able to get everything up and running for your manual provider? | 15:58 |
| thomnico | cherylj, yes I figured it ou | 15:59 |
| thomnico | out | 15:59 |
| cherylj | great! | 15:59 |
| thomnico | but this is buggy | 15:59 |
| thomnico | and I had another alert | 15:59 |
| cherylj | thomnico: I'm going to try to bootstrap with a trusty machine and see if I run into the same issue you mentioned | 16:00 |
| thomnico | they use 10.0.3.0 as an actual network .... which conflict with default lxc | 16:00 |
| cherylj | I was using wily | 16:00 |
| thomnico | the issue is really cleaning after destroy | 16:00 |
| cherylj | ah, yeah | 16:00 |
| thomnico | even with systemctl I had to clean up manually for being able to restart | 16:01 |
| thomnico | but I got unblocked in the morning | 16:01 |
| cherylj | ok, good | 16:01 |
| thomnico | logs are not really telling what check failed .. so I had to dig into the code | 16:01 |
| thomnico | I hope you have enough to reproduce .. I am not blocked .. many thanks for helping | 16:02 |
| cherylj | thomnico: you can also specify a different subnet for the lxc bridge | 16:03 |
| thomnico | in the environment ? | 16:03 |
| thomnico | got lost looking for it :) | 16:03 |
| cherylj | thomnico, no just on the machines | 16:03 |
| thomnico | indeed that is how I fix it | 16:04 |
| cherylj | you can specify a different bridge device, but you'd still need to make sure it was using the right subnet | 16:04 |
| thomnico | did not bother you if I find how to do thing | 16:04 |
| thomnico | ok so it is the expected behavior, great | 16:04 |
| thomnico | in a nutshell if destroy-environment (not force) run can clean the manual bootstrap machine that will solve a lot of headaches | 16:06 |
| cherylj | yes, I've run into that myself and I saw you mentioned the bug in your email | 16:07 |
| thomnico | and a CI test for manual : boostrap , destroy , bootstrap again | 16:07 |
| thomnico | with the NfV realm will see the need for this more often | 16:07 |
| thomnico | so it worth it imho | 16:08 |
| thomnico | and try a machine without systemd :) | 16:08 |
| thomnico | please | 16:08 |
| thomnico | should I add the info in mail in the bug report ? | 16:10 |
| thomnico | have to go back to lab.. respond to mail | 16:10 |
| thomnico | if needed | 16:10 |
| thomnico | thanks for all | 16:11 |
| mup | Bug #1539156 opened: juju does not ignore 'bootstrap-host' when bootstrap --to specified <bootstrap> <ci> <juju-core:Triaged> <https://launchpad.net/bugs/1539156> | 16:12 |
| mup | Bug #1539156 changed: juju does not ignore 'bootstrap-host' when bootstrap --to specified <bootstrap> <ci> <juju-core:Triaged> <https://launchpad.net/bugs/1539156> | 16:15 |
| mup | Bug #1539156 opened: juju does not ignore 'bootstrap-host' when bootstrap --to specified <bootstrap> <ci> <juju-core:Triaged> <https://launchpad.net/bugs/1539156> | 16:18 |
| mup | Bug #1539167 opened: Intermittent test failure "Test left sockets in a dirty state" (github.com/juju/juju/cmd/jujud/agent) <juju-core:New> <https://launchpad.net/bugs/1539167> | 16:42 |
| voidspace | dimitern: ping | 16:48 |
| dimitern | voidspace, pong | 16:48 |
| voidspace | dimitern: I'm really struggling to test the MachineAgent changes that wait for space discovery to complete | 16:49 |
| voidspace | dimitern: ideally I'd like to start a machine agent with the discoverspaces worker patched out | 16:49 |
| voidspace | dimitern: assert that the discoverspaces worker was created | 16:49 |
| voidspace | dimitern: connect to state as a client and confirm that the login fails | 16:49 |
| voidspace | dimitern: then calll the "setDiscoverSpacesComplete" function that is given to the worker | 16:50 |
| voidspace | dimitern: confirm that login now succeeds | 16:50 |
| voidspace | dimitern: as far as I can tell in cmd/jujud/agent/machine_test.go the creation of all singular workers is patched out with the singularRecord | 16:50 |
| voidspace | dimitern: so I'm failing at the first hurdle - I can't create a MachineAgent with job ManageEnviron that will actually attempt to create a discoverspaces worker | 16:51 |
| dimitern | voidspace, yeah, well that part of the code changed a lot I think when the dep-engine stuff landed | 16:51 |
| dimitern | voidspace, which was supposed to make testing things like this easier | 16:53 |
| voidspace | dimitern: as far as I can tell the login limiting (for upgrades or restore) isn't tested at the MachineAgent level | 16:54 |
| dimitern | voidspace, hmm | 16:54 |
| dimitern | voidspace, well, can you reasonably test it in isolation? | 16:55 |
| voidspace | dimitern: at the moment my conclusion is that I can't | 16:55 |
| voidspace | dimitern: maybe I can test through the api | 16:55 |
| voidspace | dimitern: I'll look and see - maybe that's how it's tested for upgrade/restore | 16:55 |
| dimitern | voidspace, possibly | 16:56 |
| dimitern | voidspace, it should be easier to test that client logins are disabled once you start the worker? | 16:56 |
| voidspace | dimitern: only if it's started from the MachineAgent, which is where the disabling is done | 16:56 |
| voidspace | dimitern: and I can't yet create a MachineAgent that will actually start the worker | 16:57 |
| voidspace | it's the MachineAgent that limits logins, not the worker | 16:57 |
| voidspace | dimitern: commonMachineSuite, which all the machine agent tests use patches out the SingularRunner | 16:58 |
| voidspace | dimitern: so none of the workers started that way are actually created | 16:58 |
| dimitern | voidspace, can you test it in a feature test? | 16:59 |
| voidspace | dimitern: yes, probably | 17:00 |
| dimitern | voidspace, e.g. the way we test that end-to-end "juju space list" works | 17:00 |
| voidspace | dimitern: I'll look at that | 17:00 |
| voidspace | dimitern: thanks | 17:00 |
| dimitern | in featuretests | 17:00 |
| voidspace | hmmm... the singularRunnerRecord that has a fakeSingularRunner looks like it might have a real runner underlying it | 17:02 |
| voidspace | which is odd, because my start function *isn't* being called | 17:02 |
| voidspace | dimitern: I'll do a bit more investigation before I give in and go to a feature test | 17:03 |
| dimitern | voidspace, cheers | 17:03 |
| mup | Bug #1539167 changed: Intermittent test failure "Test left sockets in a dirty state" (github.com/juju/juju/cmd/jujud/agent) <juju-core:New> <https://launchpad.net/bugs/1539167> | 17:18 |
| mup | Bug #1539190 opened: manual provider requires bootstrap-host, even when --to specified <bootstrap> <manual-provider> <placement> <juju-core:Triaged> <https://launchpad.net/bugs/1539190> | 17:18 |
| === marcoceppi is now known as marcoceppi|airpl | ||
| mup | Bug #1539216 opened: bootstrap rootdisk full - need to compress rotated logs <juju-core:New> <https://launchpad.net/bugs/1539216> | 19:00 |
| natefinch | heh, I've had a couple requests for compressing the rotated logs in Lumberjack. Guess I should spend some time on it ;) | 19:01 |
| === marcoceppi|airpl is now known as marcoc|airplane | ||
| * thumper afk for ~30 min | 21:18 | |
| perrito666 | wallyworld: brt | 21:27 |
| === natefinch is now known as natefinch-afk | ||
| wallyworld | axw: just you and me for standup, wanna do it now a bit early? | 23:12 |
| axw | wallyworld: omw | 23:12 |
| cherylj | wallyworld: the fix we talked about for bug 1538868 is wrong | 23:16 |
| mup | Bug #1538868: kill-controller failed: sudo: switch: command not found <ci> <kill-controller> <local-provider> <regression> <juju-core:Incomplete> <juju-core api-command-rename:Triaged> <https://launchpad.net/bugs/1538868> | 23:16 |
| cherylj | I'll fix it | 23:16 |
| wallyworld | cherylj: only if you have time, i am happy to do it as it will need to go into the api-command-rename branch | 23:17 |
| cherylj | It'll just take me a minute. I'll do it now | 23:17 |
| cherylj | wallyworld: http://reviews.vapour.ws/r/3673/ | 23:26 |
| wallyworld | cherylj: awesome, ty, will look after current meeting | 23:27 |
| cherylj | wallyworld: k, I'm going to have dinner, so merge it for me if you ship it! | 23:28 |
| wallyworld | cherylj: you awesome ty | 23:28 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!