3 Jun 2024 |
hexa | I am root on that VM, yeah | 16:09:01 |
hexa | but futzing around with a propriertary system is not very high on my list of fun things to do | 16:09:20 |
hexa | it will probably be yeeted on every upgrade etc. | 16:09:30 |
adamcstephens | yeah | 16:21:50 |
adamcstephens | the agent is a statically linked go binary | 16:22:05 |
hexa | fair | 16:25:15 |
hexa | https://github.com/canonical/lxd/issues/11603 🙂 | 16:25:27 |
adamcstephens | Linux containers requires signed off commits and I forget. Every. Time. | 18:32:27 |
4 Jun 2024 |
adamcstephens | Hoping 24.05 images will go live tonight. https://jenkins.linuxcontainers.org/job/image-nixos/ | 03:53:58 |
hexa | not yet 🙂 | 07:35:07 |
adamcstephens | ahh those times must be UTC | 12:11:44 |
adamcstephens | tonight then. | 12:11:47 |
5 Jun 2024 |
hexa | and live 🙂 | 08:33:09 |
hexa | looks like the junos vrr image does not support virtio-net 😱 | 08:52:46 |
11 Jun 2024 |
| 9pfs joined the room. | 20:24:19 |
15 Jun 2024 |
steveej | does systemd's LoadCredential work in NixOS within LXC? i'm trying to run a systemd container in containerd (i believe it uses runc as a runtime), and the credential directory exists but is empty. no errors in the journal so i'm pretty blind | 19:34:25 |
adamcstephens | yes it works in LXC | 19:52:41 |
adamcstephens | different container runtimes behave differently though | 19:53:22 |
steveej | thanks adamcstephens 🐝 . inspecting this, the only oddity that i can see in this is usage of ACLs on the credentials on my host (where they work) | 20:33:49 |
steveej | do you use the default pkgs.systemd in LXC images or a custom one? | 20:35:33 |
adamcstephens | standard systemd. | 20:35:51 |
steveej | i'm building a systemdMinimal one now. wondering if it behaves better when it doesn't have ACL and other features i might not need inside the container | 20:36:34 |
steveej | maybe i just need to yield eventually and use incus instead of k3s | 20:44:44 |
adamcstephens | it depends on what you're trying to do. incus doesn't really orchestrate | 20:48:25 |
adamcstephens | it's also a full OS, and most OCI workloads are single app containers | 20:50:57 |
steveej | i thought i could save time by finding a way to orchestrate containers that can reuse the nixos modules collection. i don't like how containers have to redefine the service semantics | 20:52:05 |
adamcstephens | full OS containers can definitely run NixOS modules :) | 20:53:14 |
adamcstephens | using NixOS modules elsewhere is a great idea, but it's the holy grail in some ways | 20:53:55 |
steveej | i agree. so much work is put into these service definitions. it'd be a shame to have to ditch that for the sake of dynamic/runtime orchestration | 20:55:13 |
25 Jun 2024 |
| thubrecht joined the room. | 21:28:51 |