| 8 Apr 2024 |
lassulus | not sure if we can precalculate the UUIDs somehow, but afaik they are randomly generated | 15:13:53 |
lassulus | I guess we could override them with some hash if we really wanted to | 15:14:11 |
n3v3r_a9a1n | same for partlabel:
warning: Git tree '/home/egor/my-little-flake' is dirty
error:
… while evaluating attribute 'nodes'
at «string»:4:25:
3| (deploy // {
4| nodes = {
| ^
5| inherit (deploy.nodes) "immortal";
… while evaluating attribute 'immortal'
at «string»:5:49:
4| nodes = {
5| inherit (deploy.nodes) "immortal";
| ^
6| };
(stack trace truncated; use '--show-trace' to show the full trace)
error:
Failed assertions:
- No filesystem or LUKS device with the name /dev/disk/by-partlabel/disk-nvme0n1-luks is declared in your configuration.
- No filesystem or LUKS device with the name /dev/disk/by-partlabel/disk-sda-luks is declared in your configuration.
🚀 ❌ [deploy] [ERROR] Failed to evaluate deployment data: Evaluation resulted in a bad exit code: Some(1)
| 15:17:54 |
n3v3r_a9a1n | In reply to @xxxcrow:matrix.org
same for partlabel:
warning: Git tree '/home/egor/my-little-flake' is dirty
error:
… while evaluating attribute 'nodes'
at «string»:4:25:
3| (deploy // {
4| nodes = {
| ^
5| inherit (deploy.nodes) "immortal";
… while evaluating attribute 'immortal'
at «string»:5:49:
4| nodes = {
5| inherit (deploy.nodes) "immortal";
| ^
6| };
(stack trace truncated; use '--show-trace' to show the full trace)
error:
Failed assertions:
- No filesystem or LUKS device with the name /dev/disk/by-partlabel/disk-nvme0n1-luks is declared in your configuration.
- No filesystem or LUKS device with the name /dev/disk/by-partlabel/disk-sda-luks is declared in your configuration.
🚀 ❌ [deploy] [ERROR] Failed to evaluate deployment data: Evaluation resulted in a bad exit code: Some(1)
Solved with boot.initrd.luks.devices | 15:26:01 |
| @lychee:lefishe.club changed their display name from lychee to kay. | 17:11:04 |
| 9 Apr 2024 |
matthewcroughan | lassulus: just wondering, have you seen the systemd-repart stuff in nixos? | 16:24:18 |
matthewcroughan | https://nixos.org/manual/nixos/unstable/#sec-image-repart | 16:24:36 |
matthewcroughan | it is just super fancy | 16:24:45 |
lassulus | Yes | 16:24:49 |
matthewcroughan | If you had to compare it in terms of feature parity to disko, what would you say? | 16:25:16 |
matthewcroughan | does it lack certain things ? | 16:25:24 |
matthewcroughan | I know for example it might lack the boot flow tests that disko has | 16:25:31 |
lassulus | No multi device. No lvm. No zfs. No bcachefs | 16:32:12 |
matthewcroughan | Actually at scale21x someone was showing it to me, and we seemed to get bcachefs working after I forked nixpkgs and changed two lines | 16:32:32 |
matthewcroughan | are you sure it doesn't work? I was going to try it today to see | 16:32:38 |
lassulus | well basic bcachefs works. But no advanced features | 16:33:00 |
lassulus | Mike encryption miltidevice | 16:33:11 |
lassulus | * Like encryption miltidevice | 16:33:20 |
matthewcroughan | At the least though, isn't there some json file that repart consumes? | 16:33:31 |
matthewcroughan | could disko consume that same json file | 16:33:35 |
matthewcroughan | * could disko consume that same json file? | 16:33:36 |
lassulus | * Like encryption or multidevice | 16:33:40 |
lassulus | I think it's unit file Syntax? | 16:34:02 |
matthewcroughan | hmm.. damn, I thought there was json for some reason | 16:34:52 |
matthewcroughan | so yes, disko is much more powerful for certain things still | 16:35:32 |
lassulus | Yes | 16:36:08 |
lassulus | And always will be. Because you can only do so much in a flat structure | 16:36:35 |
lassulus | https://www.freedesktop.org/software/systemd/man/latest/repart.d.html# | 16:36:40 |
matthewcroughan | The only reason to bother with repart is the improved performance then | 16:37:50 |
matthewcroughan | I still think it is quite important to be able to build an arm64 via binfmt in an accelerated x86 VM, since otherwise it's not possible to build disk images for other archs | 16:38:33 |