| === tds9 is now known as tds | |
| [11:33] <meena> Odd_Bloke: i see you've started using types! yay! … uhm, what do we get from that? do we have type checkers in the test system? | |
| === hjensas is now known as hjensas|afk | |
| [13:38] <Odd_Bloke> meena: For now, they serve as documentation. Per https://cloudinit.readthedocs.io/en/latest/topics/hacking.html#type-annotations, we can't use the typing module yet, so we won't really be able to annotate much of the codebase. | |
| [13:39] <rick_h_> type hints vs type enforcement? | |
| [13:47] <Odd_Bloke> It's really all type hinting, because Python doesn't use the type annotations at runtime at all. | |
| [13:47] <Odd_Bloke> But yeah, we won't have a CI check that enforces them for now. | |
| [14:09] <Odd_Bloke> meena: I would expect @ing a person and commenting on a PR they opened to produce roughly the same notifications in GH. | |
| === hjensas|afk is now known as hjensas | |
| [16:16] <shibumi> hi, is dhclient really necessary? | |
| [16:16] <shibumi> why is cloud-init not using systemd-networkd for dhcp? | |
| [16:16] <shibumi> I got a bug report for it on arch: https://bugs.archlinux.org/task/66035?dev=16600&type%5B0%5D=&sev%5B0%5D=&due%5B0%5D=&cat%5B0%5D=&status%5B0%5D=open&percent%5B0%5D=&reported%5B0%5D= | |
| [16:18] <rharper> shibumi: dhclient is used in early boot only to bring up nic to talk to IMDS; after crawling metadata service, if the distro has a renderer cloud-init writes out network config in the requested format; | |
| [16:18] <rharper> dhclient is required, on AWS, for example to enable interaction with the IMDS; | |
| [16:19] <blackboxsw> aaaaaand... it's about that time again :) | |
| [16:19] <rharper> the system is not configured to use dhclient and we do not keep the dhclient around | |
| [16:19] <blackboxsw> #startmeeting cloud-init status meeting | |
| [16:19] <meetingology> Meeting started Tue Mar 31 16:19:14 2020 UTC. The chair is blackboxsw. Information about MeetBot at http://wiki.ubuntu.com/meetingology. | |
| [16:19] <meetingology> Available commands: action commands idea info link nick | |
| [16:19] <blackboxsw> Hello and welcome to another cloud-init community status meeting folks. please feel free to continue any current conversations | |
| [16:20] <blackboxsw> I'll interleave status meeting notes with existing conversations. | |
| [16:20] <blackboxsw> #chair Odd_Bloke smoser rharper | |
| [16:20] <meetingology> Current chairs: Odd_Bloke blackboxsw rharper smoser | |
| [16:20] <blackboxsw> our IRC channel topic carries the next planned status meeting for those that wish to participate. All are welcome to interject or drive converstation topics here | |
| [16:21] <blackboxsw> let's set that now. to +2 weeks from now | |
| === blackboxsw changed the topic of #cloud-init to: pull-requests https://git.io/JeVed | Meeting minutes: https://goo.gl/mrHdaj | Next status meeting April 14 16:15 UTC | 19.4 (Dec 17) drops Py2.7 : origin/stable-19.4 | 20.1 (Feb 18) | https://bugs.launchpad.net/cloud-init/+filebug | |
| [16:21] <blackboxsw> April 14th, same time | |
| [16:21] <shibumi> rharper: thanks! | |
| [16:22] <blackboxsw> Previous meeting notes are here | |
| [16:22] <blackboxsw> #link https://cloud-init.github.io/status-2020-03-10.html#status-2020-03-10 | |
| [16:22] <blackboxsw> The topics we generally cover in this meeting are the following: Previous Actions, Recent Changes, In-progress Development, Community Charter, Upcoming Meetings, Office Hours (~30 mins). | |
| [16:22] <blackboxsw> #topic Previous Actions | |
| [16:23] <blackboxsw> and from last meeting, no previous actions were unaccounted for. | |
| [16:23] <blackboxsw> #topic Recent Changes | |
| [16:24] <blackboxsw> recent changes landed in tip of master via git log --since 2020-03-10 | |
| [16:24] <blackboxsw> #link https://paste.ubuntu.com/p/55hqVCfnpV/ | |
| [16:24] <rharper> shibumi: also, the issue with openstack metadata service not being found is likely related to classless statci route support in the EphemeralDHCP class in cloud-init, we fixed an issue there last fall, https://github.com/canonical/cloud-init/commit/07b17236be5665bb552c7460102bcd07bf8f2be8 | |
| [16:24] <rharper> shibumi: long term, we would like to replace dhclient with a python-based one; we've just not had a change to implement a minimal dhclient in python only | |
| [16:27] <blackboxsw> in that paste or recent changes, some big items have landed since last we 'met': both NetBSD and OpenBSD distro support is now added to cloud-init. vmware support of guest info gc status, SAP Converged cloud gets identified as OpenStack and Ubuntu Focal prioritizing netplan over ifupdown if both are present | |
| [16:28] <blackboxsw> Also Odd_Bloke has been landing and improving cloud-init automated process with near daily branches. github actions/workflows and docs about review and coding style expectations are landing to make it a lot easier for upstream to help speed contributions and reviews | |
| [16:29] <blackboxsw> Also, you'll note a lot of dropping python six and other py2-related artifacts from our codebase. Since upstream support is py3.4 or later we can simplify and prune a lot of the vestigial py2 functionality. | |
| [16:31] <blackboxsw> Ahh I also forgot, Ec2 now by default (Ubuntu Focal or tip of cloud-init) renders full networking, including secondary IPv4/IPv6 addresses, for all interfaces attached to a VM based on network config supplied by IMDS. Old releases of cloud-init used to only render basic networking on the primary (eth0) nic. | |
| [16:32] <blackboxsw> #topic In-progress Development | |
| [16:33] <blackboxsw> thanks Goneri and meena for all your the BSD ongoing development and support work there BTW | |
| [16:33] * Goneri waves | |
| [16:38] <blackboxsw> :) upstream is focused a bit on continuing to clean up py2 remnants from tip, adding support for reading netplan configuration from initramfs and continuing to add automation to the github development and release process/tooling to speed reviews. | |
| [16:39] <blackboxsw> we are also hammering the review queue a bit better than in the past with daily PR assignments to ensure the system remains more efficient | |
| [16:40] <blackboxsw> thanks again for all the code submits folks! | |
| [16:41] <blackboxsw> waiting In the wings we will eventually get around to handling the network hotplug solution for cloud-init (if configured) | |
| [16:42] <blackboxsw> #topic Community Charter | |
| [16:43] <blackboxsw> This section is generally reserved to discuss any general community goals for cloud-init. Per last cloud-init summit we discussed prioritizing the following: | |
| [16:43] <blackboxsw> * json schema validation for each cloudinit/config/cc_*py module | |
| [16:43] <blackboxsw> * correcting, extending stale datasource documentation under doc/rtd/topics/datasources | |
| [16:44] <blackboxsw> those tasks are easy to split up and so we set a goal to try to chunk through it this year | |
| [16:44] <blackboxsw> they are also categorized as bugs for easy pickup/assignment for anyone interested. | |
| [16:45] <blackboxsw> #link https://bugs.launchpad.net/cloud-init/+bugs?field.tag=bitesize | |
| [16:45] <blackboxsw> #topic Office hours (next ~30 mins) | |
| [16:46] <blackboxsw> thanks for listening to the above...... In this section a couple upstream devs should be available with eyes on the channel for the next 30 minutes for and bug, feature, PR review questions or concerns. | |
| [16:46] <blackboxsw> again thanks for joining. We'll have this meeting again in 2 weeks | |
| [16:47] <blackboxsw> I'm going to spend this time working on our cherry-pick script for publishing the Netplan -> ENI work into Ubuntu Focal today. | |
| [17:17] <blackboxsw> Thanks for tuning in. | |
| [17:17] <blackboxsw> #endmeeting | |
| [17:17] <meetingology> Meeting ended Tue Mar 31 17:17:39 2020 UTC. | |
| [17:17] <meetingology> Minutes: http://ubottu.com/meetingology/logs/cloud-init/2020/cloud-init.2020-03-31-16.19.moin.txt | |
| [17:35] * meena blames baby for *not* doing BSD work or support lately. | |
| [17:47] <blackboxsw> ohhhh sssssuuuuure meena. blame it on the baby ;) that's why I have kids too ;) | |
| [17:48] <blackboxsw> that, and I needed someone in the house that also likes board games | |
| [17:49] <meena> currently, she loves screaming, a lot. And honestly, as the youth say: mood. | |
| [17:50] <blackboxsw> :) | |
| [17:56] <rick_h_> that's the problem, you think "a kid, yea we're going to do all these fun things together!" | |
| [17:56] <rick_h_> and then you realize none of that starts until you put in a LOT of foundation time (like 3+ years) | |
| [17:57] <blackboxsw> true story | |
| [17:58] <sarnold> "have you considered a puppy instead?" | |
| [18:27] <meena> sarnold: we already have 2 cats (they're not fans) | |
| [18:29] <sarnold> haha :) | |
| [20:13] <Gaffel> Do I have to resort to raw commands to create a system user and group where a home dir is created and its ownership and security label is set properly? (UID and GID below 1000, but create home) | |
| [20:22] <Odd_Bloke> Gaffel: It does look like not creating a homedir for a system user is hardcoded, unfortunately: https://github.com/canonical/cloud-init/blob/master/cloudinit/distros/__init__.py#L471-L478 | |
| [20:23] <Gaffel> I was worried about that. That's not how useradd works though, which makes me disappointed. :[ | |
| [20:24] <Gaffel> I want to create a user+group with a UID & GID between 500 and 999 that also creates a home directory with the right ownership and SELinux labels. But I guess I'll have to do it differently. | |
| [20:26] <Odd_Bloke> Yeah, I'm not sure what the reasoning there is, but it looks like that behaviour has been in place since 2012. | |
| [20:26] <Odd_Bloke> It certainly seems like overriding that default behaviour would be desirable. | |
| [20:26] <Odd_Bloke> This looks like https://bugs.launchpad.net/cloud-init/+bug/1864728 to em. | |
| [20:26] <ubot5> Ubuntu bug 1864728 in cloud-init "Unable to create interactive "system" user" [Undecided,Incomplete] | |
| [20:26] <Odd_Bloke> *me | |
| [20:27] <Odd_Bloke> Gaffel: Do you have any interest in addressing the issue in cloud-init? ^_^ | |
| [20:27] <Gaffel> Changing the behaviour will be tricky, won't it? | |
| [20:28] <Gaffel> I guess no_create_home could be checked and the current behaviour would remain unless you set that to false. | |
| [20:29] <Gaffel> I assume that leaving it blank results in none though. | |
| [20:30] <Odd_Bloke> Yeah, I think you're along the right path with using no_create_home. | |
| [20:32] <Odd_Bloke> The defaulting would be a little tricky, but not super hard once you've reasoned out the matrix of possible combinations. | |
| [20:39] <Gaffel> I have the same usecase and OS family as that person. I'm trying to introduce more security hardening at work. I'll look into this. :] | |
| [20:43] <Gaffel> I'm new to cloud-init so I'll do my best. :) | |
| [20:46] <Odd_Bloke> Gaffel: \o/ You'll want to read through https://cloudinit.readthedocs.io/en/latest/topics/hacking.html to get started. | |
| [20:46] <Odd_Bloke> Gaffel: Though actually the CLA info there is out-of-date, you can just fill in your GH username and that's sufficient. | |
| [20:46] <Odd_Bloke> (The wording here predates that field being in the CLA form. :p) | |
| [20:53] <Gaffel> You mean the field where it wants my Launchpad ID? | |
| [20:53] <Gaffel> Or just leave that blank? | |
| [20:56] <Gaffel> I can't even read the text in the verification image. :/ | |
| [20:58] <rharper> Odd_Bloke: I guess we should update the docs and mention the field in the form | |
| [20:58] <Odd_Bloke> rharper: Yep, addressing that now. | |
| [20:58] <rharper> awesome! | |
| [20:59] <Gaffel> And do I really need to enter my physical address? :| | |
| [21:00] <Odd_Bloke> AFAIK, you do. | |
| [21:01] <Gaffel> :/ | |
| [21:06] <Odd_Bloke> rharper: https://github.com/canonical/cloud-init/pull/297 | |
| [21:06] <rharper> reviewing | |