[00:45] hey folks, when I add a key for a local package repository via the `user-data` file, it adds the key into /etc/trusted.gpg, which breaks attempting to run `apt update`. Has anyone else run into this? [00:46] I'm thinking to work around it by putting the keys into /etc/trusted.d/source.gpg manually, but that is a lot more trouble than using the built-in `sources:` section in `user-data`. [01:32] habys: How are you adding the key? Could you share a link to your user-data file? [01:34] habys: What version of cloud-init are you on? There were some recent feature additions in the apt module around key handling. [01:52] interesting, I will copy paste the relevant part in a sec [01:53] I was just taking the down time to write a script to pull down new iso, kernel, and initrd for our ipxe environment [01:53] I don't know the version of cloud-init, but the installer ISO was from 2022-11-03 [01:54] ah: /usr/bin/cloud-init 22.3.4-0ubuntu1~22.04.1 [01:58] here is what I hope is the relevant part of the config, here I am adding our local puppet mirror: https://termbin.com/rgfj [01:59] oh heh, that's not local, but the idea is the same. [02:49] I found my notes from setting this up manually, I would take the puppet repo key and put it in `/usr/share/keyrings/puppet7.gpg` then refer to it in `/etc/apt/sources.list.d/puppet7.list`: `deb [arch=amd64 signed-by=/usr/share/keyrings/puppet7.gpg] https://apt.puppetlabs.com jammy puppet` [03:36] it's pretty late here, so I will probably miss any response. Thank you for looking. I have another question that's killing me.. Every time something isn't perfect in my `user-data` file, I get this crash: https://i.imgur.com/UKBUYIR.png [03:37] afaict, there is nothing about what caused the crash on the screen and the listed crash file also does not tell what caused the crash. [03:38] for example I can put an intentionally bad command in `late-commands:` but I can not grep for it in these logs. Any help there would really really help me, and help me be more self sufficient. [03:38] ty! Good night [07:12] habys: autoinstall and cloud-init aren't the same thing [07:15] I think you wanna ask #ubuntu-server for help here [11:04] I thought cloud-init parses `user-data`? [11:10] habys: the autoinstall key is treated as an opaque config, the only validation cloud-init does is checking that a version subkey is present: https://github.com/canonical/cloud-init/blob/main/cloudinit/config/cc_ubuntu_autoinstall.py#L135 [11:13] habys: the autoinstall config is then handed-over to subiquity [11:14] habys: autoinstall documentation: https://ubuntu.com/server/docs/install/autoinstall [11:19] thanks for helping, I'll read this because I am definitely having trouble understanding what the different components are doing [13:46] Hi, [13:46] For file ownership issue, could we need to create a new module to handle it? [13:46] https://bugs.launchpad.net/cloud-init/+bug/1990513/comments/7 [13:46] I'd like to hear your advice whether a new module is valuable. [13:46] -ubottu:#cloud-init- Launchpad bug 1990513 in cloud-init "write_files defer not using ownership for new folders" [High, Triaged] [13:49] Jack: since we already have the functionality in 'write_files', I think the behavior there could be updated [13:49] if a directory already exists, don't modify the permissions, but if it doesn't exist, create it with the owner/permissions specified [13:55] Thanks for your suggestion, I'll try to implement it and file a PR for your review [15:24] Hi [15:24] I am trying to Sign the Canonical contributor license agreement. [15:24] What should be filled in 'Please add the Canonical Project Manager or contact' [15:24] It is a required field. [15:24] you can put James Falcon [15:25] Thanks [15:25] And Launchpad id is username of launchpad, right? [15:26] yes, that can be left blank if you don't have one [15:26] Got it, thx [15:37] tail g [15:37] oops [15:37] meena: I see you getting a mention: https://www.freebsd.org/status/report-2022-10-2022-12/#_freebsd_as_a_tier_1_cloud_init_platform [16:04] minimal: that's not a mention, that's my report [16:22] minimal: any reason you can't can't just mock cloudinit.distros.networking.subp.subp" each each test individually? [16:24] minimal: the patch in the shared function goes out of scope when the function ends https://docs.python.org/3/library/unittest.mock.html#unittest.mock.patch [16:24] I think meena mentioned that in the PR, maybe that got missed [16:28] my gut reaction to that set of tests is that there's a lot of repitition [16:28] 150 lines in that test class and it's mostly copy/pasta -> some parametrization could make that a lot more succinct [16:29] but obviously you didn't write them that way, don't feel like you have to do that [16:29] Hi guys [16:29] Feel free to review this PR if you have time. [16:29] https://github.com/canonical/cloud-init/pull/1980 [16:29] Fix https://bugs.launchpad.net/cloud-init/+bug/1990513 [16:29] -ubottu:#cloud-init- Pull 1980 in canonical/cloud-init "Set ownership for new folders" [Open] [16:29] -ubottu:#cloud-init- Launchpad bug 1990513 in cloud-init "write_files defer not using ownership for new folders" [High, Triaged] [16:38] holman: I did at first try mocking subp in each test and it didn't work - that was before I decided to move the mock to the _mock_init function [16:42] minimal: works for me https://dpaste.org/GPoOp [16:46] I haven't had time this week to update the PR with feedback, I'll be doing that later today [16:50] +1 sounds good === EugenMayer591 is now known as EugenMayer59 === EugenMayer599 is now known as EugenMayer59