Creating a ZFS pool
We can create a ZFS pool using different devices as:
a. using whole disks
b. using disk slices
c. using files
a. Using whole disks
# echo | format
Searching for disks...done
AVAILABLE DISK SELECTIONS:
0. c1t0d0
/pci@0,0/pci15ad,1976@10/sd@0,0
1. c1t1d0 <VMware,-VMware Virtual S-1.0-1.00GB>
/pci@0,0/pci15ad,1976@10/sd@1,0
2. c1t2d0
/pci@0,0/pci15ad,1976@10/sd@2,0
3. c1t3d0
/pci@0,0/pci15ad,1976@10/sd@3,0
4. c1t4d0
/pci@0,0/pci15ad,1976@10/sd@4,0
Specify disk (enter its number): Specify disk (enter its number):
I will not be using the OS disk (disk0).
# zpool create geekpool c1t1d0 # zpool list NAME SIZE ALLOC FREE CAP HEALTH ALTROOT geekpool 1008M 78.5K 1008M 0% ONLINE -
To destroy the pool :
# zpool destroy geekpool # zpool list no pools available
b. Using disk slices
Now we will create a disk slice on disk c1t1d0 as c1t1d0s0 of size 512 MB.
# zpool create geekpool c1t1d0s0 # zpool list NAME SIZE ALLOC FREE CAP HEALTH ALTROOT geekpool 504M 78.5K 504M 0% ONLINE -
c. Using files
We can also create a zpool with files. Make sure you give an absolute path while creating a zpool
# mkfile 100m file1 # zpool create geekpool /file1 # zpool list NAME SIZE ALLOC FREE CAP HEALTH ALTROOT geekpool 95.5M 115K 95.4M 0% ONLINE -
Creating pools with Different RAID levels
Now we can create a zfs pool with different RAID levels:
1. Dynamic strip – Its a very basic pool which can be created with a single disk or a concatenation of disk. We have already seen zpool creation using a single disk in the example of creating zpool with disks. Lets see how we can create concatenated zfs pool.
# zpool create geekpool c1t1d0 c1t2d0
# zpool list
NAME SIZE ALLOC FREE CAP HEALTH ALTROOT
geekpool 1.97G 80K 1.97G 0% ONLINE -
# zpool status
pool: geekpool
state: ONLINE
scan: none requested
config:
NAME STATE READ WRITE CKSUM
geekpool ONLINE 0 0 0
c1t1d0 ONLINE 0 0 0
c1t2d0 ONLINE 0 0 0
errors: No known data errors
This configuration does not provide any redundancy. Hence any disk failure will result in a data loss. Also note that once a disk is added in this fashion to a zfs pool may not be removed from the pool again. Only way to free the disk is to destroy entire pool. This happens due to the dynamic striping nature of the pool which uses both disk to store the data.
2. Mirrored pool
a. 2 way mirror
A mirrored pool provides you the redundancy which enables us to store multiple copies of data on different disks. Here you can also detach a disk from the pool as the data will be available on the another disks.
# zpool create geekpool mirror c1t1d0 c1t2d0
# zpool list
NAME SIZE ALLOC FREE CAP HEALTH ALTROOT
geekpool 1008M 78.5K 1008M 0% ONLINE -
# zpool status
pool: geekpool
state: ONLINE
scan: none requested
config:
NAME STATE READ WRITE CKSUM
geekpool ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
c1t1d0 ONLINE 0 0 0
c1t2d0 ONLINE 0 0 0
errors: No known data errors
b. 3 way mirror
# zpool destroy geekpool
# zpool create geekpool mirror c1t1d0 c1t2d0 c1t3d0
# zpool list
NAME SIZE ALLOC FREE CAP HEALTH ALTROOT
geekpool 1008M 78.5K 1008M 0% ONLINE -
# zpool status
pool: geekpool
state: ONLINE
scan: none requested
config:
NAME STATE READ WRITE CKSUM
geekpool ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
c1t1d0 ONLINE 0 0 0
c1t2d0 ONLINE 0 0 0
c1t3d0 ONLINE 0 0 0
errors: No known data errors
2. RAID-Z pools
Now we can also have a pool similar to a RAID-5 configuration called as RAID-Z. RAID-Z are of 3 types raidz1 (single parity) and raidz2 (double parity) and rzidz3 (triple parity). Lets us see how we can configure each type.
Minimum disk requirements for each type
Minimum disks required for each type of RAID-Z
1. raidz1 – 2 disks
2. raidz2 – 3 disks
3. raidz3 – 4 disks
a. raidz1
# zpool create geekpool raidz c1t1d0 c1t2d0 c1t3d0
# zpool list
NAME SIZE ALLOC FREE CAP HEALTH ALTROOT
geekpool 2.95G 166K 2.95G 0% ONLINE -
# zpool status
pool: geekpool
state: ONLINE
scan: none requested
config:
NAME STATE READ WRITE CKSUM
geekpool ONLINE 0 0 0
raidz1-0 ONLINE 0 0 0
c1t1d0 ONLINE 0 0 0
c1t2d0 ONLINE 0 0 0
c1t3d0 ONLINE 0 0 0
errors: No known data errors
b. raidz2
#zpool create geekpool raidz2 c1t1d0 c1t2d0 c1t3d0
#zpool list
NAME SIZE ALLOC FREE CAP HEALTH ALTROOT
geekpool 2.95G 186K 2.95G 0% ONLINE -
#zpool status
pool: geekpool
state: ONLINE
scan: none requested
config:
NAME STATE READ WRITE CKSUM
geekpool ONLINE 0 0 0
raidz2-0 ONLINE 0 0 0
c1t1d0 ONLINE 0 0 0
c1t2d0 ONLINE 0 0 0
c1t3d0 ONLINE 0 0 0
errors: No known data errors
c. raidz3
# zpool create geekpool raidz3 c1t1d0 c1t2d0 c1t3d0 c1t4d0
# zfs list
NAME USED AVAIL REFER MOUNTPOINT
geekpool 61K 976M 31K /geekpool
# zpool status
pool: geekpool
state: ONLINE
scan: none requested
config:
NAME STATE READ WRITE CKSUM
geekpool ONLINE 0 0 0
raidz3-0 ONLINE 0 0 0
c1t1d0 ONLINE 0 0 0
c1t2d0 ONLINE 0 0 0
c1t3d0 ONLINE 0 0 0
c1t4d0 ONLINE 0 0 0
errors: No known data errors
Adding spare device to zpool
By adding a spare device to a zfs pool the failed disks is automatically replaced by the space device and administrator can replace the failed diks ata later point in time. We can aslo share the spare device among multiple zfs pools.
# zpool add geekpool spare c1t3d0
# zpool status
pool: geekpool
state: ONLINE
scan: none requested
config:
NAME STATE READ WRITE CKSUM
geekpool ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
c1t1d0 ONLINE 0 0 0
c1t2d0 ONLINE 0 0 0
spares
c1t3d0 AVAIL
errors: No known data errors
Make sure you turn on the autoreplace feature (zfs attribute) on the geekpool
# zpool autoreplace=on mpool
Dry run on zpool creation
You can do a dry run and test the result of a pool creation before actually creating it.
# zpool create -n geekpool raidz2 c1t1d0 c1t2d0 c1t3d0
would create 'geekpool' with the following layout:
geekpool
raidz2
c1t1d0
c1t2d0
c1t3d0
Importing and exporting Pools
You may need to migrate the zfs pools between systems. ZFS makes this possible by exporting a pool from one system and importing it to another system.
a. Exporting a ZFS pool
To import a pool you must explicitly export a pool first from the source system. Exporting a pool, writes all the unwritten data to pool and remove all the information of the pool from the source system.
# zpool export geekpool # zpool list no pools available
In a case where you have some file systems mounted, you can force the export
# zpool export -f geekpool
b. Importing a ZFS pool
Now we can import the exported pool. To know which pools can be imported, run import command without any options.
# zpool import
pool: geekpool
id: 940735588853575716
state: ONLINE
action: The pool can be imported using its name or numeric identifier.
config:
geekpool ONLINE
raidz3-0 ONLINE
c1t1d0 ONLINE
c1t2d0 ONLINE
c1t3d0 ONLINE
c1t4d0 ONLINE
As you can see in the output each pool has a unique ID, which comes handy when you have multiple pools with same names. In that case a pool can be imported using the pool ID.
# zpool import 940735588853575716
Importing Pools with files
By default import command searches /dev/dsk for pool devices. So to see pools that are importable with files as their devices we can use :
# zpool import -d /
pool: geekfilepool
id: 8874031618221759977
state: ONLINE
action: The pool can be imported using its name or numeric identifier.
config:
geekfilepool ONLINE
//file1 ONLINE
//file2 ONLINE
Okay all said and done, Now we can import the pool we want :
# zpool import geekpool # zpool import -d / geekfilepool
Similar to export we can force a pool import
# zpool import -f geekpool
Creating a ZFS file system
The best part about zfs is that oracle(or should I say Sun) has kept the commands for it pretty easy to understand and remember. To create a file system fs1 in an existing zfs pool geekpool:
# zfs create geekpool/fs1 # zfs list NAME USED AVAIL REFER MOUNTPOINT geekpool 131K 976M 31K /geekpool geekpool/fs1 31K 976M 31K /geekpool/fs1
Now by default when you create a filesystem into a pool, it can take up all the space in the pool. So too limit the usage of file system we define reservation and quota. Let us consider an example to understand quota and reservation.
Suppose we assign quota = 500 MB and reservation = 200 MB to the file system fs1. We also create a new file system fs2 without any quota and reservation. So now for fs1 200 MB is reserved out of 1GB (pool size) , which no other file system can have it. It can also take upto 500 MB (quota) out of the pool , but if its is free. So fs2 has right to take up upto 800 MB (1000 MB – 200 MB) of pool space.
So if you don’t want the space of a file system to be taken up by other file system define reservation for it.
One more thing, reservation can’t be greater than quota if it is already defined. On ther other hand when you do a zfs list , you would be able to see the available space for the file system equal to the quota defined for it (if space not occupied by other file systems) and not the reservation as expected.
To set servation and quota on fs1 as stated above:
# zfs set quota=500m geekpool/fs1 # zfs set reservation=200m geekpool/fs1 # zfs list NAME USED AVAIL REFER MOUNTPOINT geekpool 200M 776M 32K /geekpool geekpool/fs1 31K 500M 31K /geekpool/fs1
To set mount point for the file system
By default a mount point (/poolname/fs_name) will be created for the file system if you don’t specify. In our case it was /geekpool/fs1. Also you do not have to have an entry of the mount point in /etc/vfstab as it is stored internally in the metadata of zfs pool and mounted automatically when system boots up. If you want to change the mount point :
# zfs set mountpoint=/test geekpool/fs1 # df -h |grep /test geekpool/fs1 500M 31K 500M 1% /test
Other important attributes
You may also change some other important attributes like compression, sharenfs etc.. Also we can specify attributes while creating the file system itself.
# zfs create -o mountpoint=/test geekpool/fs1
