Setting up Ubuntu Server on WD PRx100

Disclaimer: do this at your own risk. No fancy web gui here, just raw unix power.

---

Overview

Original article was not covering all topics important for me so I had to do some reverse engineering and add some tweaks.

This tutorial covers how to install Ubuntu Server on WD PR4100 or PR2100.

It goes from preparation, downloading required packages, running installation, initial configuration and extras that most likely are intended to be used.

The whole process can be accomplished on any linux-like system equipped with KVM.

HOWEVER if Your are looking for something more extendable, serverish, and installable with classical way.. give a try older brother - HPE ProLiant MicroServer GEN10.

---

Ansible (automated way)

There is ansible folder with automatization of all steps from extras and more.

---

PR4100 Spec

* Release date: 2016
* CPU: Intel Pentium N3710 quad-core @ 1.6 GHz
* RAM: 4 GB DDR3
* USB: 3 x 3.0 ports
* Bays: 4 x 3.5" SATA III
* LAN: 2 x 1 Gbit/s Ethernet

---

Links:

• https://community.wd.com/t/guide-how-to-install-ubuntu-18-04-server-on-the-my-cloud-pr4100-nas/232786
• https://packages.debian.org/bullseye/all/ovmf/download
• https://netplan.io
• https://github.com/michaelroland
• https://community.wd.com/u/dswv42
• https://github.com/WDCommunity/wdnas-hwtools
• https://wiki.archlinux.org/title/ZFS/Virtual_disks
• https://linuxhint.com/zfs-concepts-and-tutorial
• https://ubuntu.com/tutorials/setup-zfs-storage-pool#1-overview
• https://arstechnica.com/gadgets/2021/06/a-quick-start-guide-to-openzfs-native-encryption/
• https://www.theurbanpenguin.com/creating-zfs-data-sets-and-compression/

---

Supported devices

• WD PR2100
• WD PR4100

WD DL2100 and WD DL4100 are not supported because of ARM architecture.

---

Requirements

• ~13G free space
• KVM/QEMU
• USB flash drive (8GB+)
• brew (macos only)
• pv
• python

---

Preparation (manual way)

Common

#Prepare a working directory
mkdir ubuntu && cd ubuntu

Ubuntu

sudo apt install qemu-kvm ovmf

#Copy the UEFI bootloader to a local file named bios.bin
cp /usr/share/ovmf/OVMF.fd bios.bin

MacOS (native M1 not supported)

brew install qemu
brew install gtk+

# Get the OVMF debian package 6
# https://packages.debian.org/bullseye/all/ovmf/download
curl http://ftp.us.debian.org/debian/pool/main/e/edk2/ovmf_2020.11-2_all.deb -o ovmf_2020.11-2_all.deb

# Unpack and get the UEFI bios file.
ar -x ovmf*.deb
tar -xf data.tar.xz
mv usr/share/OVMF/OVMF.fd bios.bin

# compatibility alias
alias kvm="qemu-system-x86_64"

---

Download the Ubuntu Server

Download chosen iso from here.

---

Main process

Find out the name of your USB flash drive with lsblk.

I'll use /dev/sdX here.

Boot the iso installer:

sudo kvm -bios ./bios.bin -L . -cdrom <path_to_iso> -drive format=raw,file=/dev/sdX -boot once=d -m 1G

It should boot with a black grub screen to install Ubuntu.

Complete the installation with the defaults and any extra package that you may be interested in (e.g. Nextcloud).

Note down the user and password, you need it to login into the machine later.

At the end, it will reboot and ask you to remove the cdrom.

Just close the whole window to shutdown the whole virtual machine.

Then boot without cdrom straight from the USB flash drive.

Login in the virtual machine and update packages if you like:

sudo kvm -bios ./bios.bin -L . -drive format=raw,file=/dev/sdX -m 1G

---

Post installation (while kvm is still running)

Networking dynamic

Ubuntu is now installed for a virtual network interface with the new udev persistent networking naming.

ip addr show

You'll see the current network interface is called ens3 or similar.

This won't work on actual My Cloud hardware.

Create the netplan configuration with dhcp support:

sudo <editor> /etc/netplan/01-netconfg.yaml

network:
  version: 2
  renderer: networkd
  ethernets:
    eno1:
      match:
        macaddress: <mac1>
      dhcp4: yes
      set-name: eno1
    eno2:
      match:
        macaddress: <mac2>
      dhcp4: yes
      set-name: eno2

MACADDRESSES could be found on device or on the box, however there is hackish way in EXTRAS at the very end

This causes the NAS to get a dynamic IPv4 address on both of its onboard eno interfaces.

However, here's how to combine the throughput of the 2 network interfaces on a single IP address.

network:
  version: 2
  renderer: networkd
  ethernets:
    eno1:
      match:
        macaddress: <mac1>
      dhcp4: no
      set-name: eno1
    eno2:
      match:
        macaddress: <mac2>
      dhcp4: no
      set-name: eno2
  bonds:
    bond0:
      interfaces: [eno1, eno2]
      dhcp4: yes
      parameters:
        mode: 802.3ad
        mii-monitor-interval: 1

Static IP config should be easy too.

More info (static IP, bonding, etc) on https://netplan.io.

The Ubuntu boot disk is now ready.

Shutdown with:

sudo halt -p

Plug the USB drive in the NAS.

Boot up and enjoy!

---

Extras (meant to be run on NAS directly)

Now you can SSH to your NAS and start installing extras.

Hardware Control

Thanks to the research of Michael Roland and @dswv42 we now have full control over the fan, lcd, buttons and sensors.

Ubuntu ships with the 8250_lpss module, so you don't need to build a custom kernel.

The PMC is accessible at serial port /dev/ttyS5.

You need some packages from the universe repo.

sudo add-apt-repository universe
cd /opt
git clone https://github.com/WDCommunity/wdnas-hwtools
cd /opt/wdnas-hwtools
sudo ./install.sh

---

Create a new ZFS array

Here's a great overview on the core features of ZFS, also this might help.

Now let's create a ZFS array on the PRx100.

Insert your disks (hotplug is allowed).

List them:

$ lsblk -d
NAME         MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
...
sda            8:0    0  1.8T  0 disk
sdb            8:16   0  1.8T  0 disk
sdc            8:32   0  1.8T  0 disk
sdd            8:48   0  1.8T  0 disk
...

If you are migrating from old CloudOS and RAID was setup, you need to do following (wipeout raid info and format disks):

WARNING THIS STEP IS IRREVERSIBLE, PROCEED AT YOUR OWN RISK

# For each disk
sudo wipefs --all --force /dev/sd[a-d]

# For each disk (plus following commands)
sudo fdisc /dev/sd[a-d]

# This step is optional
sudo reboot

Create a mirror pool over /dev/sda and /dev/sdb based on the Ubuntu Tutorial.

sudo zpool create media mirror /dev/sda /dev/sdb

Alternatively, create a raidz pool over 4 disks.

This is similar to a RAID5 pool, using 1 disk for parity.

sudo zpool create media raidz /dev/sda /dev/sdb /dev/sdc /dev/sdd

Or alternatively, create a raidz2 pool over 4 disks.

This is similar to a RAID6 pool, using 2 disk for parity.

sudo zpool create media raidz2 /dev/sda /dev/sdb /dev/sdc /dev/sdd

In order to use it, you need to create a file system (also called dataset) on the zpool.

This is similar to a 'share' in the CloudOS.

Here's an example

# The file system gets mounted automatically at /media/pictures.
sudo zfs create media/pictures

or if you used my FreeNAS image to create a ZFS array

sudo zpool import

Follow the instructions.

ZFS native encryption

Additionally you can setup encryption on dataset.

sudo zfs create -o encryption=aes-256-gcm -o keylocation=prompt -o keyformat=passphrase media/pictures

After reboot:

sudo zfs mount -a

More info here.

---

Disable internal flash memory

If the internal flash memory is completely broken, you may be unable to restore the original CloudOS.

Installing Ubuntu is a solution, but you'll see system freezes when polling the disks in the dmesg output.

A solution is to blacklist the mmc_block driver.

sudo <editor> /etc/modprobe.d/blacklist.conf

Add a line with:

blacklist mmc_block

Then:

sudo update-initramfs -u

---

Hackish way to obtain MACADDRESSES

• run Ubuntu Server from USB drive on NAS without netplan config
• wait ~5min since boot
• unplug USB drive from NAS; plug USB drive into PC/MAC
• run Ubuntu Server locally:
  • sudo kvm -bios ./bios.bin -L . -drive format=raw,file=/dev/sdX -m 1G
• run journalctl | grep "ci-info" | less:

<date> <hostname> cloud-init[1279]: ci-info: +++++++++++++++++++++++++++Net device info++++++++++++++++++++++++++++
<date> <hostname> cloud-init[1279]: ci-info: +--------+-------+-----------+-----------+-------+-------------------+
<date> <hostname> cloud-init[1279]: ci-info: | Device |   Up  |  Address  |    Mask   | Scope |     Hw-Address    |
<date> <hostname> cloud-init[1279]: ci-info: +--------+-------+-----------+-----------+-------+-------------------+
<date> <hostname> cloud-init[1279]: ci-info: |  eno1  | False |     .     |     .     |   .   | 00:01:02:03:04:05 |
<date> <hostname> cloud-init[1279]: ci-info: |  eth1  | False |     .     |     .     |   .   | 06:07:08:09:10:11 |
<date> <hostname> cloud-init[1279]: ci-info: |   lo   |  True | 127.0.0.1 | 255.0.0.0 |  host |         .         |
<date> <hostname> cloud-init[1279]: ci-info: |   lo   |  True |  ::1/128  |     .     |  host |         .         |
<date> <hostname> cloud-init[1279]: ci-info: +--------+-------+-----------+-----------+-------+-------------------+
<date> <hostname> cloud-init[1279]: ci-info: +++++++++++++++++++Route IPv6 info+++++++++++++++++++
<date> <hostname> cloud-init[1279]: ci-info: +-------+-------------+---------+-----------+-------+
<date> <hostname> cloud-init[1279]: ci-info: | Route | Destination | Gateway | Interface | Flags |
<date> <hostname> cloud-init[1279]: ci-info: +-------+-------------+---------+-----------+-------+
<date> <hostname> cloud-init[1279]: ci-info: +-------+-------------+---------+-----------+-------+

• there might be a few more blocks like that, but Hw-Address contains values that we are looking for

For some reason default config renames eno2 to eth1 and tries to do the same with eno1, however our netplan config fixes that issue.

Ansible

Pre-requirements

Browse vars/ folder and do necessary changes like:

• packer:
  • macaddress
  • authorized_keys
  • hostname
  • username
  • password
• zfs:
  • zfs_pools
  • zfs_datasets
  • zfs_dataset_pass

as well as:

• ansible.cfg:
  • remote_user

Image building

ansible-playbook packer.yml --ask-become-pass

then wait 30-40min and image should create itself via packer in ansible/tmp/output

Image burn

To burn image to /dev/disk5 (disk5 used as example), type:

cd tmp/output

sudo su

#as sudo
pv -tpreb nas-ubuntu-server.img | dd of=/dev/disk5 bs=4096 conv=notrunc,noerror

NAS Part

Unplug USB from PC and plug into NAS (no matter which USB port).

Now You need to localize NAS IP, run the same nmap:

NAS bond0 interface should have 00:00:00:00:00:01 macaddress.

nmap -p 22 10.0.0.0/24

(...)
Nmap scan report for 10.0.0.101
Host is up (0.012s latency).

PORT   STATE SERVICE
22/tcp open  ssh
MAC Address: 00:00:00:00:00:01 (Xerox)
(...)

If You can ssh to NAS - go to Ansible part.

If not - debug time.

Ansible part

Just run:

ansible-playbook all_in_one.yml --ask-become-pass -e "target=10.0.0.101" -i 10.0.0.101,

And watch the magic.

After its done, it is done :)

Helpful stuff

Aliases

# Alias for loading password and mounting
function zload() {
  sudo zfs load-key -L prompt $1 && sudo zfs mount $1
}

# Alias for unloading password and unmounting
function zunload() {
  sudo zfs unmount -f $1 && sudo zfs unload-key $1
}

# Wipes whole zfs datasets and pools
zfs destroy -r nas

Manually converting qcow2 to img

qemu-img convert nas-ubuntu-server.qcow2 -O raw nas-ubuntu-server.img

Existing ZFS pool

It may happen that You already have old ZFS pool:

TASK [zfs : Create ZFS pool] ***************************************************
fatal: [0.0.0.0]: FAILED! => changed=true
  cmd: zpool create -o autoexpand=on -o autoreplace=on -o delegation=on -o dedupditto=1.5 -o failmode=continue -o listsnaps=on nas raidz2 /dev/sda /dev/sdb /dev/sdc /dev/sdd
  msg: non-zero return code
  rc: 1
  stderr: |-
    invalid vdev specification
    use '-f' to override the following errors:
    /dev/sda1 is part of potentially active pool 'nas'
    /dev/sdb1 is part of potentially active pool 'nas'
    /dev/sdc1 is part of potentially active pool 'nas'
    /dev/sdd1 is part of potentially active pool 'nas'
  stderr_lines: <omitted>
  stdout: ''
  stdout_lines: <omitted>

In that situation just type:

$ sudo zpool import
   pool: nas
     id: 0000000000000000000
  state: ONLINE
status: The pool was last accessed by another system.
 action: The pool can be imported using its name or numeric identifier and
        the '-f' flag.
   see: https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-EY
 config:

        nas         ONLINE
          raidz2-0  ONLINE
            sda     ONLINE
            sdb     ONLINE
            sdc     ONLINE
            sdd     ONLINE
$ sudo zpool import -f 0000000000000000000
$ zload nas/data

Changing password for zfs dataset

Existing keys has to be loaded.

zfs change-key \
  -o keylocation=prompt \
  -o keyformat=passphrase \
  nas/data

Changing LVM size

* parted /dev/sde
* print (fix if needed)
* resizepart 3
* <input size>
* quit
* pvresize /dev/sde3
* lvextend -r -l100%free /dev/mapper/ubuntu--vg-ubuntu--lv

Debug

While You cannot connect to NAS, unplug USB from it, and plug back to PC. Go to tmp folder (BIOS is there), and run command (replace disk5 with whatever suits your case):

sudo kvm -bios ./bios.bin -L . -drive format=raw,file=/dev/disk5 -m 4G

Log into system, and dive into logs.

And that's it.