Getting Started
The underlying storage used by ZFS filesystems involves pools. A pool may consist of one or more whole disks or disk partitions.
Basically, these whole disks and/or disk partitions can be combined in several different ways: dynamic striping, mirroring, or RAIDZ.
In all cases, the disks need to use the GUID Partition Table (GPT) and ZFS typically works best when it owns the entire disk due in
part to how conservative it is with the write cache.
Create a Simple 1 Disk Pool
In the most simple example, let’s start by using a single drive for our “puddle” storage pool. In the following example the commands
are issued as root.
First, find out which device node to use with a "diskutil list" command. In the example below, I'm going to work with /dev/disk2 which
currently has an APM (Apple Partition Map) label and an exsiting HFS filesystem. I'm going to replace the APM label with a GPT one and
blow away the HFS "FW" filesystem. You should unmount any mounted filesystems on the target drive at this point.
# diskutil list . . . /dev/disk2 #: type name size identifier 0: Apple_partition_scheme *9.4 GB disk2 1: Apple_partition_map 31.5 KB disk2s1 2: Apple_HFS FW 9.2 GB disk2s3
Now I'm going to place a GPT label on that disk: **Note this step is very important! You must format the disk before you create a ZFS pool on it
# diskutil partitiondisk /dev/disk2 GPTFormat ZFS %noformat% 100% Started partitioning on disk disk2 Creating partition map [ + 0%..10%..20%..30%..40%..50%..60%..70%..80%..90%..100% ] Finished partitioning on disk disk2 /dev/disk2 #: type name size identifier 0: GUID_partition_scheme *9.4 GB disk2 1: EFI 200.0 MB disk2s1 2: ZFS 9.0 GB disk2s2
And create our simple ZFS pool (named "puddle"):
# zpool create puddle /dev/disk2s2
And then check my work, noting that my new ZFS filesystem is available at /Volumes/puddle:
# zpool status puddle
pool: puddle
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
puddle ONLINE 0 0 0
disk2s2 ONLINE 0 0 0
errors: No known data errors
# df -hl /Volumes/puddle
Filesystem Size Used Available Capacity Mounted on
puddle 8.9Gi 19Ki 8.9Gi 1% /Volumes/puddle
Creating a Mirror or RAIDZ
To create a mirror or a RAIDZ, take a look at the following example where "tank" (a mirrored pair) and "dozer" (a RAIDZ set) are the names of our pools.
# zpool create tank mirror disk2s2 disk3s2
# zpool status tank
pool: tank
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
tank ONLINE 0 0 0
mirror ONLINE 0 0 0
disk2s2 ONLINE 0 0 0
disk3s2 ONLINE 0 0 0
errors: No known data errors
# zpool create dozer raidz disk4s2 disk5s2 disk6s2
# zpool status dozer
pool: dozer
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
dozer ONLINE 0 0 0
raidz1 ONLINE 0 0 0
disk4s2 ONLINE 0 0 0
disk5s2 ONLINE 0 0 0
disk6s2 ONLINE 0 0 0
errors: No known data errors
Add Storage to a Pool
To add more storage to an existing pool, like “tank” from the example above, take a look at the following example.
# zpool add tank mirror disk7s2 disk8s2
# zpool status tank
pool: tank
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
tank ONLINE 0 0 0
mirror ONLINE 0 0 0
disk2s2 ONLINE 0 0 0
disk3s2 ONLINE 0 0 0
mirror ONLINE 0 0 0
disk7s2 ONLINE 0 0 0
disk8s2 ONLINE 0 0 0
errors: No known data errors
Using an Existing Partition
If you have existing partitions you want to use, so long as the partition map on that drive is GPT and you've unmounted them and understand that pre-existing data on those partitions will be lost, you can use the "diskutil" command to change the label type in place. In the example below, I want to use the pre-existing HFS "blank" and Untitled 2" partitions for my "oddcouple" ZFS mirrored pool. Again, please note that with this example, any previous data on the HFS “blank” and “Untitled 2” partitions will (obviously) be overwritten! Note the Apple_HFS partition type for the “blank” and “Untitled 2” partitions:
# diskutil list /dev/disk0 #: type name size identifier 0: GUID_partition_scheme *149.1 GB disk0 1: EFI 200.0 MB disk0s1 2: Apple_HFS Leopard 29.7 GB disk0s2 3: Apple_HFS Leopard9A376 29.7 GB disk0s3 4: ZFS zfstest 29.7 GB disk0s4 5: Apple_HFS blank 29.7 GB disk0s5 6: Apple_HFS HFSJ_Boot 29.5 GB disk0s6 /dev/disk1 #: type name size identifier 0: GUID_partition_scheme *153.4 GB disk1 1: EFI 200.0 MB disk1s1 2: Apple_HFS Untitled 1 30.7 GB disk1s2 3: Apple_HFS Untitled 2 30.7 GB disk1s3 4: Apple_HFS Leopard9A376 30.7 GB disk1s4 5: Apple_HFS LaCieLeopard 30.7 GB disk1s5 6: Apple_HFS Leopard9A377a 29.9 GB disk1s6
Change the partition type to ZFS and check your work:
# diskutil eraseVolume ZFS %noformat% /dev/disk0s5 [ + 0%..10%..20%..30%..40%..50%..60%..70%..80%..90%..100% ] Finished erase on disk disk0s5 # diskutil eraseVolume ZFS %noformat% /dev/disk1s3 [ + 0%..10%..20%..30%..40%..50%..60%..70%..80%..90%..100% ] Finished erase on disk disk1s3 # diskutil list /dev/disk0 #: type name size identifier 0: GUID_partition_scheme *149.1 GB disk0 1: EFI 200.0 MB disk0s1 2: Apple_HFS Leopard 29.7 GB disk0s2 3: Apple_HFS Leopard9A376 29.7 GB disk0s3 4: ZFS zfstest 29.7 GB disk0s4 5: ZFS 29.7 GB disk0s5 6: Apple_HFS HFSJ_Boot 29.5 GB disk0s6 /dev/disk1 #: type name size identifier 0: GUID_partition_scheme *153.4 GB disk1 1: EFI 200.0 MB disk1s1 2: Apple_HFS Untitled 1 30.7 GB disk1s2 3: ZFS 30.7 GB disk1s3 4: Apple_HFS Leopard9A376 30.7 GB disk1s4 5: Apple_HFS LaCieLeopard 30.7 GB disk1s5 6: Apple_HFS Leopard9A377a 29.9 GB disk1s6
Now create the "oddcouple" pool and check your work:
# zpool create oddcouple mirror disk0s5 disk1s3
# zpool status oddcouple
pool: oddcouple
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
oddcouple ONLINE 0 0 0
mirror ONLINE 0 0 0
disk0s5 ONLINE 0 0 0
disk1s3 ONLINE 0 0 0
errors: No known data errors