| 11 Sep 2024 |
 lassulus | you would have to run disko manually for that to work, there should be support now for incremental updates, but that is very new and can cause data loss. so do that with care :) I can you can also just run the commands for the new disk manually by inspecting the generated script | 14:08:57 |
 mannp ⚡️ | I see, thanks for the quick response, just adding context in case it matters, I am using luks and btrfs for my main disk and have created the defn for luks and btrfs with one subvol /storage device in disko, but I am not sure if I am going to wipe out my main boot disk too 🙂 | 14:19:46 |
| lassulus | well disko should not wipe it out, but better have backups ;) | 14:21:35 |
| mannp ⚡️ | lol, okay | 14:23:13 |
| mannp ⚡️ | so I can use --dry-run to just get the script it seems, but am I perhaps able to pull out just the storage defn in the flake and ignore the boot disk defn? | 14:24:06 |
| mannp ⚡️ | Does disko support 2 luks encrypted devices in the same config as nix is complaining of multiple luks device definitions ... back to the manual I guess 🙂 | 14:30:08 |
| lassulus | it should, but it needs different names | 14:30:46 |
| mannp ⚡️ | Apologies, rookie copy and paste error | 14:34:42 |
| mannp ⚡️ | I am using "passwordFile = "/tmp/secret.key";" and unlocking the root luks partition is interactive, but the second cryptstorage is not. it fails to find the partiton due to lack of password. Does the second luks device have to be defined differantly? | 15:05:31 |
| mannp ⚡️ | * I am using "passwordFile = "/tmp/secret.key";" and unlocking the root luks partition is interactive, but the second cryptstorage is not. it fails to find the partiton due to lack of password. Does the second luks device have to be defined differantly? edit: Trying keyFile | 15:12:25 |
| mannp ⚡️ | When I have the flake inputs.disko.nixosModules.disko defined in my machine config, is that purely for mounting the disks? It doesn't run any further formatting post initial installation? | 15:16:17 |
| mannp ⚡️ | seems its a pre install tool not post, unless I am missing something from the manual. Will go the manual formatting route. cheers | 15:36:06 |
| 12 Sep 2024 |
|  moved to @amadaluzia:tchncs.de changed their display name from (lambda (u) (format nil "~A is not using Matrix right now." u)) "Artur Manuel" to (lambda (u) (format nil "~a ~a")) "Artur" "Manuel". | 01:48:49 |
| moved to @amadaluzia:tchncs.de changed their display name from (lambda (u) (format nil "~a ~a")) "Artur" "Manuel" to (lambda (f l) (format nil "~a ~a")) "Artur" "Manuel". | 01:55:56 |
|  Enzime joined the room. | 16:13:23 |
| 13 Sep 2024 |
 colemickens | okay, filed the issue for xcp, and sent a PR for impure-memSize: https://github.com/nix-community/disko/pull/770 | 19:08:09 |
| 14 Sep 2024 |
|  relichunter set a profile picture. | 17:40:58 |
| 15 Sep 2024 |
| lassulus | Enzime: is helping out with maintaining disko from now on, please welcome him kindly :) | 11:46:47 |
| colemickens | I found it a bit tricky to do this right, how can I better specify these partition start/ends that don't leave such awkward gaps: https://github.com/colemickens/nixcfg/blob/06d838e8abe1392433c3ddf08cff23da381f7cee/hosts/h96maxv58/base.nix#L80-L110 | 21:39:51 |
| lassulus | You can do start = 0. sgdisk will choose the next free block then | 21:51:07 |
| lassulus | Which is also the default | 21:51:16 |
| colemickens | the problem with start=0 is ordering | 22:02:24 |
| colemickens | esp with it hoisting ESP to be partition 1 | 22:05:37 |
| colemickens | well I don't know why ESP is being made first, maybe it's alpha sorted by default (since I'm using EF00 not EF02). anyway, I'll set priority. Now just looking for growpart. | 22:18:42 |
| colemickens | I keep doing it by hand and forgetting to set the part label and causing myself a lot of pain. | 22:18:59 |
| colemickens | oh I guess that's a system-level nixos option. hm | 22:34:51 |
| colemickens | I seriously love disko so much. I wish I could fund someone to make a non-VM install process, with lanzaboote, bootspec, etc. I don't think it's that outlandish. If one could compose images with cross-compiled systems without needing to involve qemu, that would just be "end-game". | 22:51:10 |
| colemickens | still, doing a native build, using disko, and then pivoting to a cross-built system is still such a nicer experience than sdcard or other things people do to get nixos on weird platforms. I'm excited to use it for ec2 image as well. | 22:52:21 |
| 16 Sep 2024 |
 arjan5 | colemickens Not sure if it's what you're looking for, but nixos-anywhere can install different-arch systems remotely, even for configs containing disko, lanzaboote, impermanence, etc. I have done this before. | 09:12:30 |
| lassulus | Lanzaboote needs interaction with the efi/bios to deploy your keys there. But I guess everything else could be done by the image builder | 12:43:32 |
