agentleFS
soulmachinesoulmachine/skills/ubuntu-zfs-mirror-install/SKILL.md

ubuntu-zfs-mirror-install

Skill3 starsChanged yesterdayMITView on GitHub

Install Ubuntu Server onto a 2-disk ZFS mirror root (rpool→/) plus a striped data pool (dpool→/data) via debootstrap + ZFSBootMenu, or fall back to a plain single-disk ext4 install (GRUB-EFI) when there aren't two equal-size disks. Use when asked to install Ubuntu on ZFS, build a ZFS mirror root / rpool+dpool layout, set up root-on-ZFS with ZFSBootMenu, or do a scripted Ubuntu Server install on a UEFI x86 server with one or two SSDs/NVMes.

The file

Files
SKILL.md

Ubuntu ZFS-mirror (or ext4) Server Install

Validated end-to-end (proven on an 8-GPU ASUS ESC8000-E12P). Every fix that makes the ZFS path boot hands-free and reboot-safe is baked into the scripts — see REFERENCE.md for what they are and how to recover if something's off.

Decision rule (choose the path from the disks)

  • Exactly two fixed disks of the SAME size → ZFS mirror (scripts/install.sh): per disk ESP | rpool-member | dpool-member; rpool mirror → /, dpool stripe → /data (no redundancy).
  • One disk, or two disks of different sizes → plain ext4 single-disk (scripts/install-ext4.sh): ESP | ext4 root, GRUB-EFI. (install.sh refuses unequal/single disks and points here.)

Prerequisites

  • Target booted into an Ubuntu Server LIVE environment (24.04+/26.04), UEFI mode (CSM off).
  • A root shell on it (local console, or SSH — enable ssh + set a password in the live env).
  • Internet reachable from the live env (debootstrap + ZFSBootMenu download).
  • Copy scripts/ to the target (e.g. scp -r scripts root@<live-ip>:).

Workflow

  1. List disks: sudo bash scripts/install.sh --list — shows fixed disks with by-id paths + sizes; the live/USB medium is marked [LIVE MEDIUM, do NOT pick].
  2. Gather settings from the user (ask; defaults in parens):
    SettingNotes
    disk(s)by-id path(s); count+size decides ZFS vs ext4
    rpool size (ZFS only)128 / 256 / 512 / 1024 GB → RPOOL_SIZE=+128G/+256G/+512G/+1T (default +1T)
    admin usernamedefault developer
    admin passworddefault passw0rd (root gets the same)
    SSH public keyoptional; if set, admin is key-only (password auth off)
    hostnamedefault ubuntu
    timezoneoptional (default Etc/UTC)
  3. Run the matching installer (type ERASE when prompted, or pass --yes once the user has confirmed):
    # ZFS mirror (two equal disks):
    DISK1=<by-id> DISK2=<by-id> RPOOL_SIZE=+1T ADMIN_USER=developer INSTALL_PASSWORD='...' \
      SSH_AUTHORIZED_KEY='ssh-ed25519 ...' HOSTNAME=ubuntu sudo -E bash scripts/install.sh
    # ext4 single disk:
    DISK1=<by-id> ADMIN_USER=developer INSTALL_PASSWORD='...' HOSTNAME=ubuntu sudo -E bash scripts/install-ext4.sh
    Takes ~10–20 min (debootstrap dominates). Over a flaky SSH link, run detached and tail the log: … nohup bash scripts/install.sh --yes >/root/install.log 2>&1 & then tail -f /root/install.log.
  4. Reboot — remove the install media (or set the target disk first in UEFI boot order). It should boot ZFSBootMenu → Ubuntu (or GRUB → Ubuntu) hands-free, with / and /data (ZFS) mounted.
  5. Validate: sudo bash scripts/validate.sh — expect pools//data ONLINE (ZFS) or / ext4, and systemctl is-system-running = running, 0 failed units.

Safety

  • Destructive. The installers refuse a disk that has mounted partitions (the live medium) and require typing ERASE (skip only with --yes after the user explicitly confirms the disks).
  • Single-NVMe failure on the ZFS layout: rpool stays bootable (degraded), dpool//data is lost (no redundancy by design).
  • ZFS path has no GRUB — the kernel cmdline lives in the org.zfsbootmenu:commandline ZFS property (it's set to quiet zfs_force=1; keep zfs_force=1 if you append e.g. intel_iommu=on iommu=pt).

See REFERENCE.md for the baked-in fixes, recovery steps, and hardware notes (incl. ASUS ESC8000 BMC quirks).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.