Hi Nix and Disko community,

I'm new to the Nix ecosystem, and I have a lot of questions, so I really appreciate your patience while I learn 😊.

I'm trying to set up a home NAS server and was planning to use Disko to manage the disks. My setup is simple:

• Use an SSD for the operating system (ESP, swap, and root).
• Use two 10TB HDDs to create a Btrfs RAID1 pool for long-term storage.

My concern is whether Disko is designed for this type of scenario. From what I understand in the documentation, Disko can format and prepare the two HDDs with Btrfs during the initial installation, but if I later add a third 10TB disk, I cannot use Disko to add it to the existing pool and rebalance declaratively, and I would need to do this manually using btrfs device add and btrfs balance.

Is this understanding correct?

Is there any tool that allows fully declarative management and expansion of Btrfs arrays, or is Disko primarily intended for the installation and reprovisioning phase rather than for post-installation expansion?

Thanks in advance!

I cannot use Disko to add it to the existing pool and rebalance declaratively

I believe that's correct. I don't know of any tool that can do that, maybe you should make one!

In reply to @steeringwheelrules:tchncs.de
are there any examples of zfs setup that works with hibernation? as far as I'm aware it only needs to not be used for the swap partition, but the rest of the drives (including nixos) can be zfs, right?

This might work for hibernation:

  disko.devices = {
    disk = {
      nvme0n1 = {
        device = "/dev/nvme0n1";
        type = "disk";
        content = {
          type = "gpt";
          partitions = {
            boot = {
              priority = 1;
              start = "34s";
              end = "2047s";
              type = "EF02"; # for grub MBR
            };
            ESP = {
              priority = 2;
              start = "2048s";
              end = "2099199s";
              type = "EF00";
              content = {
                type = "filesystem";
                format = "vfat";
                mountpoint = "/boot";
              };
            };
            luks = {
              priority = 3;
              start = "2099200s";
              end = "100%";
              type = "8309";
              content = {
                type = "luks";
                name = "crypted-nvme0n1";
                settings = {
                  allowDiscards = true;
                };
                content = {
                  type = "lvm_pv";
                  vg = "nvme0n1";
                };
              };
            };
          };
        };
      };
    };
    lvm_vg = {
      nvme0n1 = {
        type = "lvm_vg";
        lvs = {
          # LVs are created in alphabetic order for some reason
          # Prefixing them with a letter like this ensures desired creation order
          a_swap = {
            size = "64GiB"; # equal size to maximum available RAM
            content = {
              type = "swap";
            };
          };
          z_data = {
            size = "100%FREE";
            content = {
              type = "zfs";
              pool = "root";
            };
          };
        };
      };
    };
    zpool = {
      root = {
        type = "zpool";
        rootFsOptions = {
          compression = "lz4";
          "com.sun:auto-snapshot" = "false";
          mountpoint = "none";
          reservation = "16G";
        };
        postCreateHook = ''
          zfs snapshot root@blank;
          zfs snapshot root/nixos@blank;
          zfs snapshot root/nix@blank;
          zfs snapshot root/home@blank;
          zfs snapshot root/persist@blank
        '';

        datasets = {
          nixos = {
            type = "zfs_fs";
            mountpoint = "/";
            options.mountpoint = "legacy";
          };
          nix = {
            type = "zfs_fs";
            mountpoint = "/nix";
            options.mountpoint = "legacy";
          };
          home = {
            type = "zfs_fs";
            mountpoint = "/home";
            options.mountpoint = "legacy";
            options."com.sun:auto-snapshot" = "true";
          };
          persist = {
            type = "zfs_fs";
            mountpoint = "/persist";
            options.mountpoint = "legacy";
            options."com.sun:auto-snapshot" = "true";
          };
        };
      };
    };
  };

But be aware of zfs#12842 - make Linux hibernation (suspend-to-disk) more robust and nixpkgs#208037 - nixos/zfs: mitigate data loss issues when resuming from hibernate

You might also need these to enter the danger territory (make yourself familiar with zpool recovery strategy):

  boot.zfs.allowHibernation = true;
  boot.zfs.forceImportRoot = false;
  boot.zfs.forceImportAll = false;