| 13 Feb 2024 |
 maralorn | Are there by now examples for a disko + impermanence setup with bcachefs? | 00:52:20 |
 lillecarl | adamcstephens 🐝: You won't find any success there really, disko isn't meant to manage your disks, it's meant to get your partitions right while also outputting nixos module config. | 00:53:50 |
 @adam:robins.wtf | but it does manage disks and partition them… | 01:09:42 |
| @adam:robins.wtf | Literally you give disko a set of disks to manage | 01:11:31 |
| lillecarl | adamcstephens 🐝: It doesn't change your existing layout, it writes a new one | 01:11:37 |
| @adam:robins.wtf | And in theory it could be told not to destroy the entire table. There are still other benefits to using disko such as formatting and multiple volumes | 01:13:43 |
| lillecarl | I guess it's time to go implement the state machine to handle transforming your layout, brb 1 year? | 01:14:53 |
| lillecarl | * I guess it's time to go implement the state machine to handle transforming every layout from one to another, brb 1 year? | 01:15:23 |
| @adam:robins.wtf | that seems a bit extreme and unnecessary for what i'm suggesting | 01:18:01 |
 lassulus | https://github.com/nix-community/disko/pull/435 | 01:18:52 |
| lassulus | In reply to @lillecarl:matrix.org
Has anyone else experienced the issue where your md device appears as /dev/md127 rather than anything useful?
hostname: nixos
kernel: 6.7.4
+ mdadm --create /dev/md/data1 --level=1 --raid-devices=2 --metadata=default --force --homehost=any /dev/disk/by-partlabel/disk-a-data /dev/disk/by-partlabel/disk-b-data
mdadm: array /dev/md/data1 started.
mdadm: timeout waiting for /dev/md/data1
+ partprobe /dev/md/data1
The device is created as /dev/md127 instead (no md folder in /dev). It says the name is "any:data1" but it doesn't exist there.
In that case a udev rule is missing. Is this happening on nixos? Maybe the live cd? | 01:19:56 |
| lillecarl | lassulus: Yeah a very very barebones NixOS. I installed it on a third disk to be able to test the config easily before deploying to OVH. | 01:21:32 |
| @adam:robins.wtf | In reply to @lassulus:lassul.us
https://github.com/nix-community/disko/pull/435 Yeah, this in-place updating probably could do what I am looking for. Maybe paired with a script which would generate an initial disk configuration based on what's already in the table | 01:22:11 |
| @adam:robins.wtf | meaning, generated and added to your configuration. but as a way to bootstrap an existing host where you don't want to wipe the entire table | 01:23:09 |
| @adam:robins.wtf | * meaning, generated and added to your configuration. as a way to bootstrap an existing host where you don't want to wipe the entire table | 01:23:17 |
| lillecarl | adamcstephens 🐝: There's --dry-run for comparing too | 01:26:21 |
| lassulus | In reply to @lillecarl:matrix.org
lassulus: Yeah a very very barebones NixOS. I installed it on a third disk to be able to test the config easily before deploying to OVH. do you have boot.swraid.enable? | 01:28:18 |
| lillecarl | In reply to @lassulus:lassul.us
do you have boot.swraid.enable? Not in the configuration no, I found some udev rules in mdadm. Trying to just add it to systemPackages right now. | 01:29:04 |
| lillecarl | In reply to @lassulus:lassul.us
do you have boot.swraid.enable? * Not in the "host" no, I found some udev rules in mdadm. Trying to just add it to systemPackages right now. | 01:29:15 |
| lassulus | In reply to @adam:robins.wtf
meaning, generated and added to your configuration. as a way to bootstrap an existing host where you don't want to wipe the entire table yeah, that part wouldn't probably be too hard, usually only the partitions care deeply about the exact sectors they are on, but the updateScript probably wouldn't change them if they are missaligned, just print a warning, so you could just roughly sketch out your current disk layout and then apply the update script and it would fill in the missing bits | 01:30:04 |
| lassulus | I was a bit lazy with the update script thing though | 01:30:11 |
| maralorn | adamcstephens 🐝: I am looking at https://github.com/nix-community/disko/pull/499. From the comment at your example I am expecting something like bind mounts to mount the subvolumes at their mountpoints. But I see no code in the module which does that.
(Maybe I don’t understand what "they will be remounted automatically" means.) | 01:30:37 |
| maralorn | Speaking of: Can someone point me to a bind mount example with disko? I somehow can’ find it. | 01:33:17 |
| lillecarl | In reply to @lassulus:lassul.us
do you have boot.swraid.enable? Thanks 😸 I've been wondering for awhile | 01:34:28 |
| lassulus | In reply to @maralorn:maralorn.de
Speaking of: Can someone point me to a bind mount example with disko? I somehow can’ find it. usually bind mounts are just managed with fileSystems.${target} = { device = src; options = [ "bind" ]; }; | 01:37:19 |
| lassulus | but if you need them during installation you have to do them with the hooks and a shell snippet | 01:37:40 |
| maralorn | In reply to @lassulus:lassul.us
usually bind mounts are just managed with fileSystems.${target} = { device = src; options = [ "bind" ]; }; Does that work for /nix? | 01:39:36 |
| lassulus | good question, never tried | 01:39:50 |
| @adam:robins.wtf | In reply to @maralorn:maralorn.de
adamcstephens 🐝: I am looking at https://github.com/nix-community/disko/pull/499. From the comment at your example I am expecting something like bind mounts to mount the subvolumes at their mountpoints. But I see no code in the module which does that.
(Maybe I don’t understand what "they will be remounted automatically" means.) bcachefs subvolumes aren't separate mount points | 01:39:55 |
| maralorn | In reply to @lassulus:lassul.us
but if you need them during installation you have to do them with the hooks and a shell snippet Well, I am quite used to manually mounting my stuff, so that’s fine. | 01:40:15 |
