[14:04] <niluje> this build fails on jenkins: https://jenkins.ubuntu.com/server/job/cloud-init-ci/4/
[14:04] <niluje> but it works locally
[14:04] <niluje> any idea why anyone?
[14:14] <niluje> oh, found I guess.
[14:15] <dpb1> what did you find
[14:17] <niluje> it was working locally because I was running unittests on a scaleway instance
[14:22] <niluje> ok, should be fixed
[14:22] <niluje> just have to wait until the build is retriggered
[14:22] <dpb1> niluje: missing a mock?  something like that?
[14:22] <niluje> exactly
[14:22] <dpb1> k
[14:28] <niluje> annnd that's actually not good yet, there's still mocking issues :p
[14:29] <dpb1> hehe
[14:36] <blackboxsw> morning folks
[14:45] <blackboxsw> powersj: I'm looking over the proposed changes failing https://jenkins.ubuntu.com/server/job/cloud-init-integration-proposed-x/73/console  it's related to the branch smoser landed last night
[14:45] <blackboxsw> he switched bash calls in unit test setup. trying to reproduce the issue here
[14:47] <powersj> blackboxsw: so that test uses what is in proposed, which shouldn't change unless he did another upload
[14:50] <blackboxsw> hmm, ohh right, I just keyed in on the error traceback mentioning bash and made a quick assumption. right. Looking more closely
[14:51] <niluje> https://jenkins.ubuntu.com/server/job/cloud-init-ci/5/nodes=metal-amd64/console
[14:51] <niluje> any idea why the permission denied here?
[14:51] <niluje> the mock seems ok to me
[14:52] <blackboxsw> niluje: looking over the changes that landed there. it feels like the unit test is leaking a call .
[14:54] <niluje> i'm wondering if it's not because the tests are run as non root
[14:55] <niluje> when running tox locally I have a RuntimeError: Failed running ['git', 'describe', '--match=[0-9]*'] [rc=128] (, fatal: No names found, cannot describe anything.
[14:55] <niluje> lol ok
[14:55] <niluje> git remote add lp https://git.launchpad.net/cloud-init && git fetch lp fixes the issue :p
[14:55] <niluje> my clone doesn't have all the tags
[14:56] <powersj> yeah have to have the tags so it can run maketarball which uses the tags to determine the version.
[14:56] <niluje> maybe it could be optional
[15:00]  * niluje found the error -_-
[15:22] <blackboxsw> niluje: looks like you are mocking 'http://169.254.42.42/user_data/cloud-init but get_data calls just "user_data/cloud-init" is that right?
[15:27] <niluje> that's the not the issue
[15:28] <niluje> I'm mocking the HTTP call correctly
[15:28] <niluje> but I'm using a requests adadpter to bind on a specific source port (required by the Scaleway API)
[15:28] <niluje> which requires to be root
[15:28] <niluje> so I also need to fake the bind
[15:28] <niluje> will be good in a second
[15:34] <niluje> annnddd it should work now.
[15:34] <niluje> sorry for the failing builds :(
[15:35] <blackboxsw> thanks for the alert & fix
[15:35] <niluje> blackboxsw: is that you who retrigger the build?
[15:35] <niluje> or is it automatic?
[15:36] <niluje> retriggers*
[15:36] <blackboxsw> niluje: I'll check the builder to see ( I can retrigger if you've already pushed
[15:37] <niluje> (I pushed)
[15:37] <powersj> if you push again, we check for new reviews every 15mins.
[15:37] <niluje> oh, ok
[15:37] <niluje> A week or two ago I tested jenkins blue ocean
[15:37] <niluje> I see your are using jenkins-job-builder
[15:38] <niluje> using a Jenkinsfile + blue ocean is really a nice alternative
[15:38] <niluje> https://s3.amazonaws.com/img0.recordit.co/QRyeaiscAS.mp4?AWSAccessKeyId=AKIAINSRFOQXTN4DT46A&Expires=1497631105&Signature=SipNfzd6Mk1vE5DP8KjcCE1972g%3D
[15:38] <niluje> if you want to see what it looks like :)
[15:39] <niluje> (it was just a job to discover the features of the blue ocean interface, it basically does nothing)
[15:40] <powersj> wow that is a clean interface
[15:41] <niluje> yep
[15:41] <powersj> I do have a new merge review pipeline coming, I was hoping to have it up today, but ran into an issue with jenkins-launchpad-plugin. I'll try to resolve that Monday.
[15:42] <powersj> Essentially for every merge we will run this: https://jenkins.ubuntu.com/server/job/cloud-init-ci-nightly/
[15:42] <niluje> and with this setup, you only need to create a job from jenkins and fill a git URL
[15:42] <niluje> containing the Jenkinsfile
[15:42] <niluje> so since the job configuration is saved on git, you no longer have to use jenkins job builder :)
[15:42] <powersj> yeah that would be nice
[15:43] <niluje> can you show me the corresponding jenkinsfile powersj ?
[15:43] <powersj> http://paste.ubuntu.com/24873098/
[15:43] <blackboxsw> thx powers for the build kick
[15:44] <niluje> oO
[15:44] <niluje> is it the full jenkinsfile powersj ? or just a part of it
[15:44] <niluje> I thought they had to start with "pipeline {"
[15:45] <powersj> That's what I have inside the pipeline section right now. It isn't checking out from git, so I'm not sure if that affects it.
[15:45] <niluje> ok
[15:45] <niluje> I thought the first block inside a "stage" block needed to be a "steps" block
[15:46] <niluje> http://blackhole.brmzkw.info/2017-06-16/Jenkinsfile <- one of my tests with jenkinsfile
[15:47] <powersj> interesting
[15:47] <powersj> it does say "required" on the doc site
[15:47] <niluje> powersj: the "steps" section?
[15:47] <powersj> yeah
[15:47] <niluje> yeahh. build is passing. Thanks for your help
[15:48] <niluje> yep, without steps I had an error
[15:48] <niluje> honestly I'm totally blind when working with jenkins
[15:48] <niluje> they have the worst documentation possible
[15:49] <niluje> in case you are interested (because I lost some time to search how to do it): it is possible, in a pipeline, to interactively ask for administrator input (like "do you want to approve this build? yes/no" or "can you fill this text input so I can later use the content int he build? <input value>")
[15:49] <niluje> I have Jenkinsfiles doing that
[15:49] <niluje> it is also possible to use the jenkins API to fill these inputs
[15:50] <powersj> that is cool, the thing I would like to figure out is parallel steps
[15:50] <niluje> wait for it :p
[15:50] <powersj> that way I can reduce our CI time
[15:50] <powersj> haha
[15:50] <niluje> powersj: http://blackhole.brmzkw.info/2017-06-16/JENKINS.md
[15:51] <niluje> (if you see errors feel free to report :))
[15:51] <niluje> i wrote this documentation for my coworkers who discover jenkin
[15:51] <niluje> s
[15:51] <powersj> nice thx!
[16:01]  * powersj look's at proposed failures
[16:16] <powersj> blackboxsw: I am 90% certain it is due to me updating the version of pylxd on the system
[16:23] <powersj> blackboxsw: it is. we are currently playing games with the version of tox + pylxd + lxd because of some incompatibilities between the versions. The backported code does not work with the latest lxd + pylxd combo, even when running with a tox specified version of lxd.
[16:24] <blackboxsw> bah
[16:24] <powersj> because we have two good clean runs of proposed I am going to say we are good, disable the job for now so we don't keep randomly running and failing and re-enable for the next proposed test.
[16:24] <powersj> bad timing in we had an upload to proposed + broken pylxd + major update to integration tests
[16:26] <blackboxsw> powersj: I was wondering  if this is also impacted by  read-dependencies now attempting to upgrade system-packages each integration test run now as well
[16:26] <powersj> It may, but on an LTS I would hope it would not break
[16:48] <ajorg> is it okay to delete a branch in my fork once it's been merged?
[16:49] <blackboxsw> ajorg: should be fine there
[21:34] <blackboxsw> ok 4 SRU bug verifications needed still
[21:50]  * blackboxsw is setting up an azure account
[22:47] <blackboxsw> found two other bugs that were part of the SRU, but weren't listed in our trello card. adding them now