Category Archives: ZFS

ZFS on Ubuntu server

Ubuntu Home Server Setup Part II

Welcome to Part II of my Ubuntu Home Server build! In Part I, I did a very basic Ubuntu Server install. In this part, I’ll be creating a ZFS pool and volumes to store all my data on.

Other parts of this guide can be found at:

Home Server With Ubuntu

Setup

I’ll be setting up a server with 8 physical drives.

Disk 0: SSD for OS

Disk 1: SSD for ZFS Intent Log (improves write performance)
(read fantastic information about it here: http://nex7.blogspot.com/2013/04/zfs-intent-log.html)

Disk 2: SSD for L2ARC caching (improves read performance)

Disk 3 – 7: HDDs for ZFS Pool (where all my data with be stored)

Quick disclosure: I’m *far* from a ZFS expert. From what I’ve gleaned, this should suffice for home / small business use. If you’re planning something enterprise-grade, find an expert!

Install Ubuntu

Perform a regular Ubuntu server installation, or use an existing server.

SSH Into the server, rather than using the console. You’ll want to be able to copy and paste when you setup the zpool.

Install ZFS

sudo apt install zfsutils-linux

Create the ZPOOL

I’ll be using RAIDZ (which is like RAID-5) to get redundancy on my disks without losing too much usable space.

ZFS offers many other options, like RAID0, 1, 6, etc. Use whichever is appropriate for your workload.

It is very strongly recommended to not use disk names like sdb, sdc, etc. Those might change across reboots.

Many of the articles I’ve read suggest using UUIDs . However, my experience on Ubuntu Server is that these are not assigned to blank disks. Therefore, I will be using disk paths instead.

These are verbose and a bit of a pain to type, but they make sure you know exactly what disk you are referring to should you need to swap drives in the future. They will also not change on reboots.

To see your installed disks run:

ls -lh /dev/disk/by-path

My output looks like

adam@normandy:~$ ls -lh /dev/disk/by-path
 total 0
 lrwxrwxrwx 1 root root  9 Jul  8 09:06 pci-0000:00:1f.2-ata-5 -> ../../sr0
 lrwxrwxrwx 1 root root  9 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:0:0 -> ../../sda
 lrwxrwxrwx 1 root root 10 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:0:0-part1 -> ../../sda1
 lrwxrwxrwx 1 root root 10 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:0:0-part2 -> ../../sda2
 lrwxrwxrwx 1 root root 10 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:0:0-part3 -> ../../sda3
 lrwxrwxrwx 1 root root  9 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:1:0 -> ../../sdb
 lrwxrwxrwx 1 root root  9 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:2:0 -> ../../sdc
 lrwxrwxrwx 1 root root  9 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:3:0 -> ../../sdd
 lrwxrwxrwx 1 root root  9 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:4:0 -> ../../sde
 lrwxrwxrwx 1 root root  9 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:5:0 -> ../../sdf
 lrwxrwxrwx 1 root root  9 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:6:0 -> ../../sdg
 lrwxrwxrwx 1 root root  9 Jul  8 09:06 pci-0000:02:00.0-scsi-0:0:7:0 -> ../../sdh

I chose to install Linux on my 1st drive (sda). I’ll be using sdb for the ZIL, sdc for L2ARC, and sdd, sde, sdf, sdg, and sdh to for the data pool.

First, I’ll setup the data pool. This is where SSH is handy, since you can copy/paste your paths from above.

In my example below, I’m naming my pool “data.” You can use a different name if you’d like. If your setup is like mine, you’ll create one pool with many volumes in it.

I’m using drives with 4k physical sectors, so I’m adding the option: -o ashift=12
This should increase performance, but at the cost of total storage space. You an remove this option if you don’t think it’s a good fit for you.

sudo zpool create data -o ashift=12 raidz /dev/disk/by-path/pci-0000:02:00.0-scsi-0:0:3:0 /dev/disk/by-path/pci-0000:02:00.0-scsi-0:0:4:0 /dev/disk/by-path/pci-0000:02:00.0-scsi-0:0:5:0 /dev/disk/by-path/pci-0000:02:00.0-scsi-0:0:6:0 /dev/disk/by-path/pci-0000:02:00.0-scsi-0:0:7:0

To confirm this worked, run:

zpool list

You should have something like:

adam@normandy:~$ zpool list
NAME   SIZE  ALLOC   FREE  EXPANDSZ   FRAG    CAP  DEDUP  HEALTH  ALTROOT
data  18.1T   238K  18.1T         -     0%     0%  1.00x  ONLINE  -

Next I’ll tell ZFS to use sdb as the ZFS Intent Log

sudo zpool add data log /dev/disk/by-path/pci-0000:02:00.0-scsi-0:0:1:0

Then I’ll tell ZFS to use sdc as the L2ARCH cache

sudo zpool add data cache /dev/disk/by-path/pci-0000:02:00.0-scsi-0:0:2:0

If I run zpool status, I should see my data, ZIL, and cache drives

adam@normandy:/data/download/secure$ zpool status
  pool: data
 state: ONLINE
  scan: none requested
config:

        NAME                               STATE     READ WRITE CKSUM
        data                               ONLINE       0     0     0
          raidz1-0                         ONLINE       0     0     0
            pci-0000:02:00.0-scsi-0:0:3:0  ONLINE       0     0     0
            pci-0000:02:00.0-scsi-0:0:4:0  ONLINE       0     0     0
            pci-0000:02:00.0-scsi-0:0:5:0  ONLINE       0     0     0
            pci-0000:02:00.0-scsi-0:0:6:0  ONLINE       0     0     0
            pci-0000:02:00.0-scsi-0:0:7:0  ONLINE       0     0     0
        logs
          pci-0000:02:00.0-scsi-0:0:1:0    ONLINE       0     0     0
        cache
          pci-0000:02:00.0-scsi-0:0:2:0    ONLINE       0     0     0

errors: No known data errors

Create the Filesystem

Now that the zpool exists, we can create filesystems on top of it.
A pool can have multiple filesystems. I’ll create one for media, and one for virtual machines (because that’s what I need).

sudo zfs create data/media
sudo zfs create data/vm

To confirm it was created correctly run:

zfs list

And it should look something like this:

adam@normandy:~$ zfs list
 NAME         USED  AVAIL  REFER  MOUNTPOINT
 data         210K  14.0T  36.7K  /data
 data/media  35.1K  14.0T  35.1K  /data/media
 data/vm     35.1K  14.0T  35.1K  /data/vm

I would also suggest the following tweaks. Combined, they increased my zfs throughput 50-100%! They were recommended by https://unicolet.blogspot.com/2013/03/a-not-so-short-guide-to-zfs-on-linux.html and https://www.servethehome.com/the-case-for-using-zfs-compression/ as I searched for solutions to my less-than-stellar zfs performance.

zfs set xattr=sa data/media
zfs set atime=off data/media
zfs set compression=lz4

All of your zfs filesystems are automatically mounted.

adam@normandy:~$ mount
...
data on /data type zfs (rw,xattr,noacl)
data/media on /data/media type zfs (rw,xattr,noacl)
data/vm on /data/vm type zfs (rw,xattr,noacl

You can use them just as you would any mounted filesystem. That’s it!