[10:10] I have a question about the init scripts: Is there every any reason to only enable one of the cloud-init daemons? === esv_ is now known as esv [13:46] meena: you mean scripts rather than daemons? There's only 1 "daemon" with cloud-init, the hotplug script [14:13] meena: not that I can think of. each init script corresponds to the 4 boot stages. Different parts of cloud-init discovery or configuration run during each of the boot stages. If you don't run init-local or init stages you risk not discovering certain datasources which are only detected in init-local or init stage. If you skip modules-config or modules-final you don't apply parts of the user-data configuration in the cc_* modules. [14:36] blackboxsw: aye. [14:36] meena: what's the hotplug script? and does that work on not-linux? (does it even work on Alpine?) [14:38] meena: it's for handling hotplugged network interfaces, i.e. on AWS [14:38] the shipped script does not work on (base) Alpine as it is written in Bash but I have packaged a derivative that runs on sh/ash [14:45] how does a script run as daemon? [15:05] meena: in the sense that it runs and keeps running rather than runs for a few seconds and exits [15:05] it listens on a pipe (from memory) and processes data from it [15:05] ah, [15:06] basically (again from memory) it listens for udev events and reacts to them [15:09] minimal: this thing? https://github.com/canonical/cloud-init/blob/HEAD/tools/hook-hotplug [15:10] yeah [15:13] how do people manage to write 20 lines of shell code, that's not sh compatible? [15:14] https://git.alpinelinux.org/aports/tree/community/cloud-init/03-hook-hotplug-sh.patch [15:18] minimal: oh, right, no arrays, … brrr… [15:18] but, still, easier to fix that than expected [15:18] how come you didn't push it upstream? [15:19] I haven't tested in yet, hotplug is only support by Ec2, ConfigDrive2, and OpenStack datasources from memory [15:23] meena: in a related note I have a patch for doas support which I haven't upstreamed as doas is fragmented (Alpine is using opendoas) and it relies in functionality only currently present in opendoas [15:25] meena: boooo [15:25] i mean, that sounds great [15:26] basically doas only supports a single config file /etc/doas.conf whereas opendoas added also support for a directory where multiple files can be read from - so my doas support acts in a similar way to c-i sudo support of creating file in /etc/doas.d/ [15:27] speaking of sudo / doas: somebody should really make the sudo thing idempotent [15:28] https://gist.github.com/620feecb55db46cdeab5c12fa986ae61 [15:33] doesn't the contents of /etc/sudoers.d/90-cloud-init-users get replaced rather than appended to? [15:34] haven't checked the code yet... [15:34] not on my machines [15:34] every time i run cloud-init clean -sl i get a new entry [15:36] add, the code appends to file is it exists [15:37] function write_sudo_rules in distros/__init_.py or freebsd.py, netbsd.py [15:39] minimal: indeed, i just saw that, too [15:39] so it's a feature [15:41] I think we should read the file, and if the contents is the same as what we have, skip writing it [15:42] I think just write the file regardless (unless the intention was for the image creator to prepopulate the file, in which case couldn't user-related entries in /etc/clould/cloud.cfg achieve the same?) [15:42] write as in overwrite [15:46] yeah, i think overwrite would be less… surprising. [15:47] meena: BTW I noticed you'd raised a launchpad issue some time ago about freebsd support for cc_ca-certs [15:47] I'm doing some work with that code currently [15:47] minimal: reminder to myself, to implement it [15:47] is there a clear document as to how freebsd handles certs? [15:48] I'm already trying to figure out if the existing RHEL support is fully functional or not [15:48] minimal: usually, our man pages are pretty good ;) https://man.freebsd.org/certctl [15:48] and if not, that's a bug. [15:49] ok, i'll have a look and try and add FreeBSD support while I'm at all [15:50] it's on my radar, but my radar is currently beeping out of control [15:51] well I'm rejigging the code anyway so it's not much more effort to try and add it [15:51] plus wouldn't want my changes to make it any harder to add other os support in future [15:52] once I get a PR together then you can test FreeBSD support for me ;-) [21:38] ok one more upstream release published to ubuntu Lunar: [ubuntu/lunar-proposed] cloud-init 22.4.1-0ubuntu1 (Accepted) thanks falcojr [21:39] falcojr: plan is to now queue these uploads for ubuntu/kinetic|jammy|focal|bionic. so that when our 22.4 SRU clears the queue, we ping S-R-U release vanguards to assess and approve the 22.4.1 upload containing only the fix https://github.com/canonical/cloud-init/pull/1853 [21:39] -ubottu:#cloud-init- Pull 1853 in canonical/cloud-init "net: skip duplicate mac check for netvsc nic and its VF" [Merged] [21:40] blackboxsw: so we can just let them sit in approved queue until current SRU clears? [21:42] @falcojr we can do either. let them sit or I can upload once we vet the ubuntu/* branches and we will only ping from S-R-U review once we've cleared solutions QA. Probably best if we ping/await solutions QA results on 22.4 before muddying the upload queue to unapproved (B, F, J, K) series [21:43] I'll see if there's any updates from them, but tomorrow is technically the end of the waiting period [21:43] yeah let's double check. I really don't want to push a release out if solutions QA doesn't have a +1 for us here [21:44] especially given upcoming US-holiday