| 22 Apr 2024 |
 hexa | * have a local change, thinking about a test | 18:04:27 |
 adamcstephens | maybe just add a tpm device and use incus exec to verify a VM has created /dev/tpm0 ? | 19:19:43 |
| adamcstephens | looking at a VM in my environment there's no /dev/tpm0 | 19:19:54 |
| adamcstephens | * looking at a VM in my environment there's no /dev/tpm0 ( as i don't have a tpm on it ) | 19:20:02 |
| hexa | yeah, that was the issue I was stumbling over as well | 19:20:29 |
| adamcstephens | the error you got prevented the VM from starting, right? | 19:20:33 |
| hexa | it prevented the adding | 19:20:50 |
| adamcstephens | ahh | 19:20:55 |
| hexa | * it prevented the adding of the config section iirc | 19:20:56 |
| 26 Apr 2024 |
 rungmc | I don't know if this is already on your radar but you can get some fun PATH errors passing through a physical wireless device (Incus apparently uses iw under the hood for that) or toggling on AppArmor (aa-exec specifically) | 22:11:29 |
| adamcstephens | are the errors on the daemon or in an instance? | 22:19:02 |
| adamcstephens | i wouldn't expect apparmor to work at all | 22:19:17 |
| rungmc | Yeah, I only enabled AA to see if it would work for shits and giggles. For iw, the instance log shows:ERROR network - ../src/lxc/network.c:lxc_netdev_move_wlan:1649 - Couldn't find the application iw in PATH | 22:34:28 |
| adamcstephens | Ok I suspect that’s the daemons path which is in the incus module | 22:41:12 |
| 27 Apr 2024 |
| adamcstephens | In reply to @rungmc:matrix.org
Yeah, I only enabled AA to see if it would work for shits and giggles. For iw, the instance log shows:ERROR network - ../src/lxc/network.c:lxc_netdev_move_wlan:1649 - Couldn't find the application iw in PATH Can you provide the device command or config you’re adding? | 02:54:24 |
| rungmc | incus config device add testContainer wlan0 nic name=wlan0 nictype=physical parent=wlp11s0... I think. I'm working at like 2% mental capacity with a migraine that I'm blaming on looking into the community drama earlier 🙃 | 03:38:59 |
| adamcstephens | i wonder if our quantity of tests is starting to make them flaky | 12:33:44 |
| adamcstephens | e.g. i see timeouts | 12:34:01 |
| adamcstephens | i could merge some of these tests to reduce the overall quantity, but i explicitly split them so they can be more directly targeted by packages as passthru tests. | 12:35:53 |
| adamcstephens | https://github.com/NixOS/nixpkgs/pull/307039 | 12:53:15 |
| adamcstephens | i just can't get those tests to pass on ofborg, even though they work locally. i guess i can roll back the test refactor but i'm a bit doubtful that will fix them | 12:53:50 |
| hexa | did you add any systemd hardening things? | 12:54:14 |
| hexa | because for some reason kanidm consistently fails on ofborg due to its hardening, but works locally | 12:54:48 |
| adamcstephens | no | 12:55:37 |
| adamcstephens | it would be interesting to try and run incus as another user, but i have a feeling it's not able to | 12:56:10 |
| hexa | if you can't reproduce test failures on the same platform as where ofborg fails … don't worry | 12:56:42 |
| adamcstephens | any thoughts on that PR? :) especially the name of the option I'm adding | 12:57:52 |
| hexa | sigh | 12:58:16 |
| adamcstephens | oh no, it's not that bad is it? | 12:58:29 |
| hexa | I'll try to take a look, but it isn't smol | 12:58:29 |
