!oNSIfazDqEcwhcOjSL:matrix.org

disko

353 Members
disko - declarative disk partitioning - https://github.com/nix-community/disko88 Servers

Load older messages


SenderMessageTime
5 Jul 2024
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)Ah there it is.. yeah we need a module option namespace now21:33:34
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) disko.image-builder.pkgs for example should exist and be overridable, and all the other options we have should be inside of disko.image-builder 21:33:56
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)I will pr that shortly21:59:05
8 Jul 2024
@wolfeexd:matrix.orgwolfeexd joined the room.03:35:38
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)After so much hacking I finally managed to solve that problem, but it's a huge hack14:53:54
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)we'll end up maintaining almost all of `vmTools14:54:00
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) * we'll end up maintaining almost all of `vmTools` 14:54:04
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) * we'll end up maintaining almost all of vmTools\ 14:54:06
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) * we'll end up maintaining almost all of vmTools 14:54:09
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)because vmTools doesn't let you directly override the qemu version used by all of the tools15:05:08
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)maybe I should make a PR to vmTools instead15:05:14
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) it just needs to stop using that damned qemuBinary function 15:05:44
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) * it just needs to stop using that damned qemuBinary function and pass qemu as an input rather than the whole of pkgs 15:05:56
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)https://github.com/NixOS/nixpkgs/pull/32561919:04:57
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)important for disko19:05:00
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)unless anyone knows a better way19:05:07
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) I've tried to make it as minimal as possible lassulus but it requires a vmTools change in nixpkgs 19:16:09
@wolfeexd:matrix.orgwolfeexd

attempt: on a BIOS boot only system, I am trying to assign a id or uudi to a disko config because /dev/sdx gets switched around and becomes less viable in a raid setup
problem: switching to or using /dev/disk/by-id/<some-device> causes a GRUB2 error after rebuild: "This GPT partition label contains no BIOS Boot Partition; embedding won't be possible."
extra context: this is on a system that already has a disko config setup and im trying to switch it, i cannot bootstrap with /dev/disk/by-id disk either.

I've been stuck on this for a few days, just a nudge in the right direction even if its pointing out something in the docs, or a workaround would be very helpful.

config:
""
{
device ? throw "Set this to your disk device, e.g. /dev/sda",
...
}: {
disko.devices = {
disk.main = {
inherit device;
type = "disk";
content = {
type = "gpt";
partitions = {
boot = {
name = "boot";
size = "1M";
type = "EF02"; # for grub MBR
};
esp = {
name = "ESP";
size = "500M";
type = "EF00";
content = {
type = "filesystem";
format = "vfat";
mountpoint = "/boot";
};
};
swap = {
size = "4G";
content = {
type = "swap";
resumeDevice = true;
};
};
root = {
name = "root";
size = "100%";
content = {
type = "lvm_pv";
vg = "root_vg";
};
};
};
};
};
lvm_vg = {
root_vg = {
type = "lvm_vg";
lvs = {
root = {
size = "100%FREE";
content = {
type = "btrfs";
extraArgs = ["-f"];

          subvolumes = {
            "/root" = {
              mountpoint = "/";
            };

            "/persist" = {
              mountOptions = ["subvol=persist" "noatime"];
              mountpoint = "/persist";
            };

            "/nix" = {
              mountOptions = ["subvol=nix" "noatime"];
              mountpoint = "/nix";
            };
          };
        };
      };
    };
  };
};

};
}
""

23:26:32
@wolfeexd:matrix.orgwolfeexdif /dev/disk/by-id/xxx is just a link to /dev/sdx then im not sure what would grub to fail23:27:42
@philiptaron:matrix.orgPhilip Taron (UTC-8)
In reply to @wolfeexd:matrix.org
if /dev/disk/by-id/xxx is just a link to /dev/sdx then im not sure what would grub to fail
I think this link is made by systemd-udevd and happens quite late
23:29:00
9 Jul 2024
@wolfeexd:matrix.orgwolfeexd * if /dev/disk/by-id/xxx is just a link to /dev/sdx then im not sure what would cause grub to fail00:22:47
@eyjhb:eyjhb.dkeyJhb joined the room.10:23:44
@guanran928:mozilla.org@guanran928:mozilla.org left the room.17:48:01
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)Anyone wanna review https://github.com/NixOS/nixpkgs/pull/325619 so I can improve disko with https://github.com/nix-community/disko/pull/69818:24:04
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)Or offer a better idea18:24:30
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) lassulus: fwiw k900 suggested something about binfmt namespaces making this less of an issue 18:32:13
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) binfmt is an impurity, so in flakes can be modelled as packages.x86_64-linux-binfmt for example 18:32:28
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192) * binfmt is an impurity, so in flakes can be modelled as packages.x86_64-linux.binfmt-foo for example 18:32:37
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)dare we make platform quads instead of platform triples? Lol18:32:52
@matthewcroughan:defenestrate.itmatthewcroughan @ 39c3 (DECT 94667 or 97340 or 67192)pkgsBinfmt could exist in a binfmt namespace world18:33:09

Show newer messages


Back to Room ListRoom Version: 10