=== vrubiolo1 is now known as vrubiolo [14:14] Hi folks, typically I see Looking for data source in: ['Azure'], via packages ['', u'cloudinit.sources'] that matches dependencies ['FILESYSTEM'] in a working image. We have some cloud-init package created by a distro that is showing in log "Looking for data source in: ['Azure', 'None'], via packages ['', u'cloudinit.sources'] that matches [14:14] dependencies ['FILESYSTEM'] and it's not working as expected. Where should I begin to look at? /etc/cloud/cloud.cfg.d/ has the same file between two deployment and indicating Azure in the datasource list. How did 'None' get in there? [14:57] I kept getting disconnected from this webirc and keep losing messages. Is there anyway to configure webirc to not disconnect? [15:10] ahosmanMSFT: uhm, you coud pay for cloud-irc? [15:12] Switched to a client. Hopefully this works better. [15:19] * meena uses thelounge [15:25] AnhVoMSFT: My guess is that the issue is not to do with that line, but something else. Could you file a bug and attach the `cloud-init collect-logs` tarball from an affected instance, please? [15:27] Odd_Bloke: he's generating another package for it. I felt like something else was wrong too; but from the log line that was the only thing difference up until that point. We will test the new package later today and see if issue persists. [15:27] I am curious though, on how the datasource list is constructed, when does None get added? [15:27] AnhVoMSFT: OK, cool, let us know. :) [15:30] If it isn't in the datasource_list configuration in /etc/cloud/..., then I'm not sure off the top of my head. [16:36] Odd_Bloke: blackboxsw: https://github.com/canonical/cloud-init/pull/205 [16:47] rharper: Approved, had to merge master so waiting for CI to land. [16:47] (Thanks!) [16:47] Odd_Bloke: \o/ [19:00] any tips for troubleshooting why static ip address informatin isnt being set? my localDS specificies eth0 in 'network-config' but cloud-init is writing out ens18 to /etc/netplan/* [19:00] and using dhcp instead of static [19:03] of course i figured it out,now .. curtin wrote something out to /etc/cloud/cloud.cfg.d/ . [19:44] pfak: curtin will write out the network-config that maas provided; [19:46] rharper: this was the -live dvd [19:46] but yeah, i blew that away and now cloud-init is working perfectly with new ifnames disabled [19:46] subiquity writes network config directly to /etc/netplan/ IIRC, I don't believe subiquity provides network-config to curtin to write through [19:47] /etc/cloud/cloud.cfg.d/50-curtin-networking.cfg was writen out by the ubuntu live installer [19:47] which version of subiquity ? [19:47] whatever is shipped with 18.04 installer [19:47] 18.04.3 [19:47] hrm, ok [19:47] idk how to tell, the package isnt installed [19:47] and i blew away /var/log/installer [19:48] ok [19:48] sorry i couldnt be more helpful [19:48] no worries [22:08] rharper: blackboxsw: powersj: I have a branch to convert cloud-init to use pytest (instead of the long-deprecated nose). Given that it may cause unexpected issues, I propose aiming to land it soon after we cut 19.1. What do you think? [22:09] 20.1 ? [22:09] Yes. [22:09] I think makes sense to me [22:09] I'm good with it [22:09] Odd_Bloke: what impact do you think this will have on existing branches which add/modify unittests ? [22:10] rharper: I haven't changed any test code in my branch, so I believe it will be minimal. [22:10] (I will send an email out to the list, but figured that shouldn't be the first that anyone else had heard of it. ;) [22:11] Odd_Bloke, what would come of say... the xenial branch? [22:11] the packaging branches [22:11] We would need to update Build-Depends there, at least. [22:11] deps are in main? [22:13] Odd_Bloke: ok [22:13] This is just changing Build-Depends, so they don't need to be. [22:13] (They do for trusty, but we don't need to worry about that for cloud-init.) [22:13] aahh right [22:13] thanks! [22:14] Odd_Bloke: and they are available on xenial (out side of main) at levels that work ? [22:15] rharper: I have Travis passing, which includes the xenial package build for cloud tests. [22:15] \o/ [22:15] perfect [22:15] that's a big +1 for landing post 20.1; definitely mention that for downstreams [22:15] That's using bddeb, but I would be surprised if we saw anything different in the packaging branches. [22:15] I wonder about non-ubuntu though [22:15] likely worth seeing what robjo and otubo think of the branch [22:15] pytest was in main in trusty, so it's been Around for long enough that we're probably fine. [22:15] yeah [22:17] rharper: Odd_Bloke What branch? [22:17] robjo: Converting to pytest for testing. [22:17] The branch is not yet proposed, I'm about to do that then send an email out to the list notifying and asking for feedback. [22:17] That should be fun [22:18] Glad I have no active work :D [22:26] Can anyone remind me how to build a cloud-init package from the packaging branches? [22:26] I don't have an orig tarball, and make-tarball produces something that includes .tox. [22:29] Odd_Bloke: one sec; [22:34] https://launchpad.net/ubuntu/+source/cloud-init has links to orig tarballs; [22:34] and then I use build-package after checking out the release branch, https://github.com/CanonicalLtd/uss-tableflip/blob/master/doc/ubuntu_release_process.md , search for "build-package" [22:35] it has references to downloading an orig tarball and build-package script and example usage [22:35] OH, I had forgotten about build-package. [22:35] Odd_Bloke: I'm surprised make-tarball included .tox [22:35] it didn't for me [22:36] Odd_Bloke: I've also done: git checkout -b mybuid; git merge origin/ubuntu/$release; ./tools/bddeb; [22:36] Doesn't bddeb generate its own debian/ dir? Or does it skip that if one is already present? [22:43] Odd_Bloke: it looks like it copies from the topdir, so if you merge over it; then it should be the "release branch" version of debian/ [22:43] and you should bddeb --release $release [23:31] OK, I've drafted the email but I'm well EOD now, so I'll reread it with fresh eyes in the morning and send it then. [23:45] tiny tiny pr for doc update related to swapfile and cc_mounts configuration. [23:45] https://github.com/canonical/cloud-init/pull/212 [23:45] * blackboxsw is reviewing the netbsd branch, I think Goneri got back to the review comments [23:46] https://github.com/canonical/cloud-init/pull/62 for me [23:49] hrm I think there are more commits to come there, like consolidation of FreeBSD and NetBSD distro modules into a common parent class maybe. I'm not sure there, will ping the author