[08:11] <meena> we have a logging configuration. you can change it
[11:04] <meena> smoser: fixed the mysteriously failing kenv test! https://github.com/canonical/cloud-init/pull/621 now to replace dmidecode with kenv in ds-identify
[11:27] <meena> maybe i should make my console fonts smaller, so i can fit more python backtrace into one screen
[13:20] <smoser> meena: what was it ?
[13:21] <smoser> https://github.com/canonical/cloud-init/commit/963690e5f8867d71a5106ce05a768dcf7eea646f i guess.
[13:22] <paride> Odd_Bloke, could you please create and push a new branch for groovy? `git checkout -b ubuntu/groovy ubuntu/20.3-15-g6d332e5c-0ubuntu1` should do it
[13:22] <paride> ubuntu/devel is already ahead of what ubuntu/groovy should have
[13:22] <paride> I can't prepare a PR for it as the target branch is nonexisting
[13:27] <paride> once that's done let's see if https://code.launchpad.net/~cloud-init-dev/+recipe/cloud-init-daily-groovy builds
[13:41] <meena> smoser, yes, the closure was checking the wrong (translation table)
[13:58] <smoser> meena: i put  a review there.
[13:58] <smoser> paride: the second argument there to 'checkout -b' sets the tracking branch
[13:59] <smoser> doesnt it? which isn't what you want. or maybe if you feed it a tag it does teh right thing.
[13:59] <smoser> i always branch with: git checkout -b fix/my-new-branch upstream/master
[14:01] <paride> smoser, right, but in this case we couldn't cut ubuntu/groovy from ubuntu/devel as ubuntu/devel is one commit ahead of what it's in groovy
[14:01] <paride> no big deal probably? Should we SRU to groovy we're going to merge everything anyway
[14:01] <smoser> right. i'm just saying that you wouldnt want to set the tracking to a tag.
[14:01] <smoser> that is what i'm not familiar with.
[14:02] <smoser> i guess iwth a tag it does not set the upstream tracking.
[14:02] <smoser> so... never mind  me
[14:47] <meena> smoser, thank you for the review. I'll make those changes before moving on to ds-identify fix is_container and ds-identify
[14:49] <Odd_Bloke> falcojr: I was just trying to use IN_PLACE for testing, and it looks to me like the mount happens too late (I've added some logging lines locally for this): https://paste.ubuntu.com/p/nXRRWRFJbg/
[14:51] <Odd_Bloke> It might take a bit of refactoring to get it in the right place: I think we'll need to create but not start the container, perform the mount, then start the container?
[14:56] <falcojr> yeah, good catch. Even without that it's a bit of an outlier in terms of timing
[15:00] <meena> re ds-identify, i'll need some guidance on how to best do that in python, because it's done like this https://github.com/canonical/cloud-init/blob/master/tools/ds-identify#L405
[15:48] <smoser> meena: do what in python ?
[15:52] <Odd_Bloke> paride: +1 on the Jenkins jobs changes, leaving to you to land as you'll need to do some manual cleanup on the no-longer-generated jobs.
[15:52] <paride> thanks Odd_Bloke
[15:53] <meena> smoser, basically, I'm wondering if I should hack that sysctl call into CONTAINER_HELPERS
[15:55] <smoser> oh. i see.
[15:56] <meena> CONTAINER_TESTS,actually
[15:57] <meena> test sysctl call -eq 1
[15:58] <smoser> meena: i m ight re-do CONTAINER_TESTS to be a list of functions.
[15:59] <meena> or i might
[15:59] <meena> or lambdas?
[15:59] <smoser> then just insert one named 'is_container_bsd' that checks uname and returns None.
[15:59] <smoser> you coudl try with lamndas' but theire' harder to mock and such
[16:00] <meena> bah
[16:01] <meena> I'ma let you do the ground work then
[16:06] <paride> Odd_Bloke, I'm tempted to switch to `jenkins-jobs update --delete-old`. I did that in the iso-testing jenkins and it really helped with housekeeping. Can you think of a reason why that might not be a good idea?>
[16:07] <powersj> we have this same debate about once a year :)
[16:12] <paride> powersj, I thought twice :P
[16:12] <smoser> meena: https://paste.ubuntu.com/p/k9VpYPGvwC/
[16:12] <smoser> thoughts ?
[16:12] <powersj> paride, I have had it turned on before, but folks wanted to be able to create temp jobs or deploy dev jobs
[16:18] <meena> smoser, looks good
[16:18] <meena> smoser, wanna add that to my branch?
[16:19] <paride> powersj, I remember you saying that it was not a good idea for some reason. Thanks for chiming in :)
[16:22] <paride> anyway manually added jobs should not be auto-deleted if the '<!-- Managed by Jenkins Job Builder -->' string is deleted from their description
[16:22] <paride> as jjb doesn't recognize them as its own
[16:23] <paride> alternative 2: add a bool parameter to admin-jobs-update so it runs --delete-all on request
[16:25] <paride> I'll make a PR with this and let's discuss it there
[16:28] <smoser> meena: you can add it and flesh out tesets. it will change how the tests work i'm sure. :-(
[16:28] <smoser> but i suspect most things mock is_container() so implementation woudlnt matter
[16:34] <powersj> paride, I like having the param so you can cleanup easily when you need/want to
[16:40] <meena> smoser, I'll attribute it to you
[16:46] <smoser> s/attribute/blame/ thanks
[16:46] <smoser> ;)
[16:51] <paride> powersj,  https://github.com/canonical/server-jenkins-jobs/pull/147
[17:41] <Odd_Bloke> falcojr: Easy one: https://github.com/canonical/cloud-init/pull/656
[17:48] <Odd_Bloke> falcojr: Actually, I think we lost the line that emitted _all_ settings to the log during the refactor, so I'll submit that in a second instead.
[17:56] <Odd_Bloke> falcojr: Take two: https://github.com/canonical/cloud-init/pull/657
[19:46] <meena> somebody please make paste.ubuntu.com not require a login, to download the raw text
[19:51] <meena> smoser: you can't build a FreeBSD system without sysctl
[19:52] <meena> smoser: /rescue (all statically compiled stuff that you can use as … well, rescue system) has sysctl
[19:52] <meena> like we can't build FreeBSD without geom
[20:10] <Odd_Bloke> meena: It requires a login to avoid it being used for spam/malware automation, I believe.
[20:12] <Odd_Bloke> falcojr: Another one for the integration testing nits: I don't think I updated .travis.yml for s/IMAGE_SOURCE/CLOUD_INIT_SOURCE/ before it landed, so all of the tests thus far have been running against the cloud-init baked into our OS image.
[20:31] <falcojr> That's half hilarious half disturbing