!oNSIfazDqEcwhcOjSL:matrix.org

disko

364 Members
disko - declarative disk partitioning - https://github.com/nix-community/disko93 Servers

Load older messages

SenderMessageTime
14 Jan 2025
@zvzg:matrix.orgSávio* It seems diskoImagesScript is broken on nixpkgs master, as reported at https://github.com/nix-community/disko/issues/900#issuecomment-2556781723 I have run into the same issue, and I tried to fix it. Ever since vmRunCommand started depending on the Nix build environment and the switch from 9p to virtiofs, it has become awkward to make it work. The generated script doesn't run in the Nix build environment, so the only way I found to do it was saving the environment at build time and loading it in the script. It works, but it's a hack. Then I found out virtiofsd must be run as root or as fake root in a user namespace, which requires some code changes. diskoImages still works fine though. Is diskoImagesScript still supported?04:26:13
@dustee:matrix.orgdusteeRedacted or Malformed Event16:39:50
@dustee:matrix.orgdustee * when i connect to sshfs with nix filesystems.<>.fsType = "sshfs" the mountpoint freezes and doesnt work properly but doing it with a manual command with the same options works as expected. anyone experiencing anything similar? 16:40:00
15 Jan 2025
@usernamedelta89:matrix.orgusernamedelta89 joined the room.00:34:32
@py6m0e7i:matrix.orgpy6m0e7i joined the room.10:01:37
16 Jan 2025
@nxtk:matrix.orgnxtk joined the room.18:15:00
@nxtk:matrix.orgnxtk Hi, to encrypt drives during nixos-anywhere deployment, one must provide a temporary passphrase key file and add it in disko to the LUKS partition as well.
However, setting key file under boot.initrd.luks.devices."device".keyFile leads to incorrect initrd-crypttab entry, which prevents automatic TPM unlock down the line since key file isn't available anymore after installation, thus it seems like I need to remove key file entries after disko ran and partitions were mounted, but before the installation starts, do you have any ideas how to deal with that situation? 		18:16:25
@lassulus:lassul.uslassulusthere is passwordFIle18:40:33
@nxtk:matrix.orgnxtkepic, missed that18:44:53
@nxtk:matrix.orgnxtk* epic, missed that 🙏18:45:01
@zvzg:matrix.orgSáviohttps://github.com/nix-community/disko/pull/943 binfmt emulation21:34:23
@zvzg:matrix.orgSávioI tried using QEMU, but it was too slow without KVM21:35:10
17 Jan 2025
@zvzg:matrix.orgSávioIs there a way to run into the VM after finishing the build but before exiting? I would like to modify some files in the image00:47:42
@zvzg:matrix.orgSávio* Is there a way to run commands in the VM after finishing the build but before exiting? I would like to modify some files in the image00:47:54
@zvzg:matrix.orgSáviosystem.activationScripts seems to do the trick02:00:58
@rts-kaede:thewired.chかえでちゃんis it possible to have the swap inside a luks lvm and have working hibernation?14:14:54
@rts-kaede:thewired.chかえでちゃん* is it possible to have the swap inside a luks lvm and have working hibernation with disko?14:15:28
@lassulus:lassul.uslassulusyes14:16:04
@lassulus:lassul.uslassulusI think you need to configure some kernel parameters for that, but I could be mistaken14:16:27
18 Jan 2025
@lanice:matrix.orglanice joined the room.23:03:28
@lanice:matrix.orglanice

Hi! First time using disko, started out with an example using nixos-anywhere with just one ext4 partition (besides boot), and it worked great!

My question: Is it possible to have just /nix mounted on my ext4 partition, and the rest of root (/, /var, etc.) mounted on a differend SSD with zfs?

23:05:59
@lassulus:lassul.uslassulusyeah, just point / on the other one and /nix on the one you want /nix to be, you just need to set the mountpoint23:23:15
@lanice:matrix.orglaniceThanks!23:28:42
@lanice:matrix.orglanice One other question: I've seen a few configs that used a post creat hook, something like postCreateHook = "zfs snapshot zroot/local/root@blank";, while other configs don't. What are the implications here. Is it just the inital snapshot that's missing, the "empty" one basically? Do I need that? 23:29:38
19 Jan 2025
@enzime:nixos.devEnzime people make the empty snapshot for impermanence setups where you wipe your rootfs on every boot 02:38:54
@lanice:matrix.orglaniceAh, right, I've seen that before. That makes sense, thank you! Not doing that for now.03:00:51
@lanice:matrix.orglanice

For my HDD's, I have a zpool "data" configured like this:

data = {
  type = "zpool";
  mode = {
    topology = {
      type = "topology";
      vdev = [ { members = ["hdd24tb_0" "hdd24tb_1" "hdd24tb_2"]; mode = "raidz1"; } ];
    };
  };
  rootFsOptions = {
    atime = "off";
    xattr = "sa";
    compression = "zstd";
  };
  options.ashift = "12";
  datasets = {
    "data" = { type = "zfs_fs"; mountpoint = "/data"; };
    "media" = { type = "zfs_fs";  mountpoint = "/data/media"; };
    "backup" = { type = "zfs_fs"; mountpoint = "/data/backup"; };
  };
};

After finishing the install with nixos-anywhere, when I do zfs list, the output for that pool is:

data                1.41M  43.5T   139K  /data
data/backup  128K   43.5T   128K  /data/backup
data/data       139K   43.5T   139K  /data
data/media    128K   43.5T   128K  /data/media

What confuses me is the double-use of mountpoint /data. It seems both data and data/data is mounted on /data. Why did that happen? I probably did something wrong in my disko config, I'm doing a lot of trial and error and looking at other configs.

03:21:46
@lanice:matrix.orglanice *

For my HDD's, I have a zpool "data" configured like this:

disko.devices.zpool.data = {
  type = "zpool";
  mode = {
    topology = {
      type = "topology";
      vdev = [ { members = ["hdd24tb_0" "hdd24tb_1" "hdd24tb_2"]; mode = "raidz1"; } ];
    };
  };
  rootFsOptions = {
    atime = "off";
    xattr = "sa";
    compression = "zstd";
  };
  options.ashift = "12";
  datasets = {
    "data" = { type = "zfs_fs"; mountpoint = "/data"; };
    "media" = { type = "zfs_fs";  mountpoint = "/data/media"; };
    "backup" = { type = "zfs_fs"; mountpoint = "/data/backup"; };
  };
};

After finishing the install with nixos-anywhere, when I do zfs list, the output for that pool is:

data                1.41M  43.5T   139K  /data
data/backup  128K   43.5T   128K  /data/backup
data/data       139K   43.5T   139K  /data
data/media    128K   43.5T   128K  /data/media

What confuses me is the double-use of mountpoint /data. It seems both data and data/data is mounted on /data. Why did that happen? I probably did something wrong in my disko config, I'm doing a lot of trial and error and looking at other configs.

03:22:36
@lanice:matrix.orglanice *

For my HDD's, I have a zpool "data" configured like this:

disko.devices.zpool.data = {
  type = "zpool";
  mode = {
    topology = {
      type = "topology";
      vdev = [ { members = ["hdd24tb_0" "hdd24tb_1" "hdd24tb_2"]; mode = "raidz1"; } ];
    };
  };
  rootFsOptions = {
    atime = "off";
    xattr = "sa";
    compression = "zstd";
  };
  options.ashift = "12";
  datasets = {
    "data" = { type = "zfs_fs"; mountpoint = "/data"; };
    "media" = { type = "zfs_fs";  mountpoint = "/data/media"; };
    "backup" = { type = "zfs_fs"; mountpoint = "/data/backup"; };
  };
};

After finishing the install with nixos-anywhere, when I do zfs list, the output for that pool is:

data                1.41M  43.5T   139K  /data
data/backup  128K   43.5T   128K  /data/backup
data/data  139K   43.5T   139K  /data
data/media    128K   43.5T   128K  /data/media

What confuses me is the double-use of mountpoint /data. It seems both data and data/data is mounted on /data. Why did that happen? I probably did something wrong in my disko config, I'm doing a lot of trial and error and looking at other configs.

03:24:58
@lanice:matrix.orglanice *

For my HDD's, I have a zpool "data" configured like this:

disko.devices.zpool.data = {
  type = "zpool";
  mode = {
    topology = {
      type = "topology";
      vdev = [ { members = ["hdd24tb_0" "hdd24tb_1" "hdd24tb_2"]; mode = "raidz1"; } ];
    };
  };
  rootFsOptions = {
    atime = "off";
    xattr = "sa";
    compression = "zstd";
  };
  options.ashift = "12";
  datasets = {
    "data" = { type = "zfs_fs"; mountpoint = "/data"; };
    "media" = { type = "zfs_fs";  mountpoint = "/data/media"; };
    "backup" = { type = "zfs_fs"; mountpoint = "/data/backup"; };
  };
};

After finishing the install with nixos-anywhere, when I do zfs list, the output for that pool is:

data                1.41M  43.5T   139K  /data
data/backup  128K   43.5T   128K  /data/backup
data/data    139K   43.5T   139K  /data
data/media    128K   43.5T   128K  /data/media

What confuses me is the double-use of mountpoint /data. It seems both data and data/data is mounted on /data. Why did that happen? I probably did something wrong in my disko config, I'm doing a lot of trial and error and looking at other configs.

03:25:06

Show newer messages

Back to Room ListRoom Version: 10