NanoPi M4 NAS

    Table of Contents

    Introduction

    The NanoPi M4 is an ARM board that supports up to 4x SATA drives when enquipped with this SATA Hat. The board also has 4x USB 3.0 ports, meaning we can theorectially have up to 8 drives connected to this little thing. My goal is to set this up with a few HDDs/SSDs in a ZFS RAIDZ pool and use it to backup photos.

    Hardware

    Flashing an OS

    FriendlyCore

    Initially, I tried using one of the OS images provided by FriendlyElec, called FriendlyCore. This is a Ubuntu-based OS, and comes prepackaged with some NanoPi specific utilities and headers.

    However, the image is for an 8GB SD card, and it comes with 2 partitions, one that is 3GB for the base OS, and the remaining 4.5GB for a copy of the OS that is meant to flashed into the eMMC on the NanoPi. The eMMC would be a lot faster to use than an SD card, but alas, I didn't purchase an eMMC module when I bought the card. It's worth trying later, though.

    The issue is that 3GB leaves me running out of disk space after any trivial operation. The mount is set up as an overlayfs, which I'm not too familiar with. I tried to delete the unnecessary partition, but it just left me unable to boot.

    Armbian

    Instead, I decided to try out Armbian, which has prebuilt images for many different SBCs, including the NanoPi M4. The setup on Armbian was really smooth.

    ZFS

    ZFS is an open-source file system and volume manager with some really cool features. With it, we can make a file system that spans across multiple drives, and extend it with new drives whenever we need.

    Installing ZFS

    Once Armbian is set up, I ran the following to install ZFS.

    Typically you can use the armbian-config utility to install kernel headers, but it didn't seem to work for me, so I installed them manually.

    sudo apt install linux-headers-current-rockchip64
sudo apt install zfs-dkms zfsutils-linux

    The latter command may take some time to compile the ZFS kernel modules, so take a break, stretch, etc.

    Make sure you're able to load the zfs kernel module with sudo modprobe zfs. If not, something went wrong and you might need to do some googling.

    ZFS RAIDZ Expansion

    ZFS offers a number of software based RAID modes for data redundancy. RAIDZ stripes parity information across all disks, letting up to one drive fail without losing data.

    As of 2022, you can expand a RAIDZ group with a new drive without having to recreate the zpool. The caveat is that existing data uses the old data/parity ratio, while new data will use the new data/parity ratio.

    For example, if I have 2 disks, then my data/partiy ratio is 1:1, meaning that only 1/2 of my total raw storage is actually usable. If I add another disk to the RAIDZ group, my data/partiy ratio improves to 2:1, meaning 2/3 of my total storage is usable. With this caveat, though, existing data will keep the old data/parity ratio, leaving my usable disk space somewhere between 1/2 and 2/3 of the total.

    However, we can get around this by rewriting the old data, giving it the new data/parity ratio.

    Create ZFS Pool

    Creating a basic RAIDZ1 pool that allows for one drive to fail without losing data.

    # find disk by serial
lsblk -o name,model,serial

# add discs by serial instead of `/dev/sda`...
sudo zpool create lake raidz /dev/disk/by-id/XXX /dev/disk/by-id/YYY

    Applications

    Backing Up Data

    TODO

    Photo Library

    There is a plugin for Nextcloud I want to try out called memories. This would work as a Google Photos replacement.

    Streaming Music

    TODO

    Streaming Video

    TODO