The Funtoo Linux project cannot currently recommend ZFS on Linux for use in production. In fact, we recommend that you do not use ZFS in production. We have had a number of issues with this filesystem, even after making some very significant effort to make it work more reliably. Please use BTRFS instead. It's better than you've heard.
ZFS is an advanced filesystem that is available for use in Funtoo Linux, thanks to the ZFS on Linux project.
It is easy to set up and use ZFS. In this simple introduction, we're going to set up ZFS under Funtoo Linux using an existing
debian-sources-lts kernel like the one that comes pre-built for you with Funtoo Linux, and we will also be using our ZFS storage pool for storing data that isn't part of the Funtoo Linux installation itself. This means that we don't need to worry about enabling ZFS support in GRUB, or mounting ZFS to actually boot Funtoo Linux. Funtoo Linux will boot from a non-ZFS filesystem, and as part of the initialization process will initialize our ZFS storage pool and mount it at the location of our choice.
To install ZFS, perform the following steps:
root # emerge zfs
This will emerge the ZFS userspace tools (
zfs) as well as ZFS kernel modules (
spl). Once complete, enable ZFS in your default runlevel as follows:
root # rc-update add zfs-import default root # rc
ZFS is now initialized and ready for use.
Unlike traditional filesystems like ext4 and xfs, ZFS is an all-inclusive storage technology that manages its own filesystems without using
/etc/fstab. The ZFS concept of importing volumes and their associated filesystems makes them available for use by the operating system. This will be performed when the system boots via the
zfs-import startup script.
ZFS also generally manages the physical disks that it uses, and physical disks are added to a ZFS storage pool. Then, ZFS can create volumes from the storage pool on which files can be stored.
Unlike traditional Linux filesystems, ZFS filesystems will allocate storage on-demand from the underlying storage pool. Thus, we can set the "size" of a ZFS volume, but this space only actually allocated when files are stored on the filesystem. In contrast, traditional Linux filesystems like ext4 and xfs must be assigned underlying block storage in advance.
In ZFS terminology, a ZFS storage pool can hold the following things, all of which are considered to be datasets:
- filesystems - these are what get mounted and you store files in. Generally, this is the main thing people use ZFS for.
- clones - a filesystem that is created as a copy of an existing snapshot.
- snapshots - a read-only copy of a filesystem at a given point in time.
- volume - a dataset that acts as a block device, such as a swap device.
When you inspect the contents of a ZFS storage pool, you will see potentially all these different types of things listed as the contents of the pool, and their names will appear in a
pool/path[@snapshot] format. Pool is the name of the storage pool. Path is a slash-delimited path name for the component, and the slashes don't represent directories but a logical organizational hierarchy for the dataset in the pool.
Creating a Storage Pool
To create a basic ZFS storage pool, you will need an extra empty disk. Perform the following steps:
root # zpool create mypool /dev/sdxy
/dev/sdxy should be an unused disk. You may need to use the following command if this disk contains any pre-existing data on it:
root # zpool create -f mypool /dev/sdxy
Once your storage pool is created, you can verify its existence with the
zpool status command:
root # zpool status pool: mypool state: ONLINE scan: none requested config: NAME STATE READ WRITE CKSUM mypool ONLINE 0 0 0 sdb ONLINE 0 0 0 errors: No known data errors root #
And if you type
zfs list, you will likely see something like this:
root # # zfs list NAME USED AVAIL REFER MOUNTPOINT mypool 2.19G 459G 96K none
Notice the mountpoint entry of
None. While it is possible to mount your storage pool directly and use it as a filesystem, it is best to create a filesystem as a sub-path within your pool's namespace, as follows:
root # zfs create mypool/home root # zfs list NAME USED AVAIL REFER MOUNTPOINT mypool 2.19G 459G 96K none mypool/home 96K 459G 96K none
As you can see above, although we have created a ZFS filesystem, it is only using 96K of storage on our pool, although there are 459GB available. You can also see that the filesystem is currently not mounted. Rather than use the
mount command, let's change that the ZFS way:
root # mkdir /data/home root # zfs set mountpoint=/data/home mypool/home root # mount ... mypool/home on /data/home type zfs (rw,xattr,posixacl)
We have now set the
mountpoint property on our filesystem, and can see that it is now mounted where we want it. ZFS will remember that our
mypool/home filesystem gets mounted at
/data/home. Most people will want their filesystems to be automatically mounted at boot and will perform the following steps to make this happen:
root # rc-update add zfs-mount default
You should now be at the point where you can begin to use ZFS for a variety of tasks. While there is a lot more to ZFS than what is covered in this short introduction, you should now have a good understanding of the fundamental concepts on which ZFS is based.