ZFS as a volume manager

First, as I don't have any spare disks available, I'll create some files to use as pseudo disks. One of the many cool things about ZFS is that you can use it on anything from files like this for testing, to USB keyfobs, to honking great storage arrays. As a proof of concept, all I need is a couple of 100Mb files, which I'll then use to create a ZFS mirrored pool. I don't actually gain any redundancy here, but you can see how this could be used in a real-life scenario :

[mark@solaris:~] # mkfile 100m disk1
[mark@solaris:~] # mkfile 100m disk2
[mark@solaris:~] # sudo zpool create test mirror $PWD/disk1 $PWD/disk2

Now we can see the "test" pool is online and view it's status :

[mark@solaris:~] # zpool list
NAME                    SIZE    USED   AVAIL    CAP  HEALTH     ALTROOT
test                   95.5M   52.5K   95.4M     0%  ONLINE     -
[mark@solaris:~] # zpool status test
  pool: test
 state: ONLINE
 scrub: none requested
config:

        NAME                         STATE     READ WRITE CKSUM
        test                         ONLINE       0     0     0
          mirror                     ONLINE       0     0     0
            /export/home/mark/disk1  ONLINE       0     0     0
            /export/home/mark/disk2  ONLINE       0     0     0

errors: No known data errors
[mark@solaris:~] # zfs list      
NAME                   USED  AVAIL  REFER  MOUNTPOINT
test                  75.5K  63.4M  24.5K  /test

So I have around 63Mb to play with. Right now, I could just proceed as normal and create a few ZFS filesystems, but instead I'll create a block volume. This is done by using the "-V" flag with "zfs create", and specifying a size for our block device :

[mark@solaris:~] # sudo zfs create -V 60m test/testvol

This now creates an entry under /dev/zvol, much like Veritas creates things under /dev/vx/. We can then format it and mount it under /mnt/testvol :

[mark@solaris:~] # sudo newfs /dev/zvol/rdsk/test/testvol 
newfs: construct a new file system /dev/zvol/rdsk/test/testvol: (y/n)? y
/dev/zvol/rdsk/test/testvol:    122880 sectors in 20 cylinders of 48 tracks, 128 sectors
        60.0MB in 2 cyl groups (14 c/g, 42.00MB/g, 20160 i/g)
super-block backups (for fsck -F ufs -o b=#) at:
 32, 86176,

[mark@solaris:~] # sudo mkdir /mnt/testvol
[mark@solaris:~] # sudo mount -F ufs /dev/zvol/dsk/test/testvol /mnt/testvol 

Let's check it's really there :

[mark@solaris:~] # df -h /mnt/testvol 
Filesystem             size   used  avail capacity  Mounted on
/dev/zvol/dsk/test/testvol
                        55M   1.0M    49M     3%    /mnt/testvol

Now let's try snapshoting it. I'll create a file, snapshot the filesystem and then delete it to prove we could get it back.

[mark@solaris:~] # sudo touch /mnt/testvol/testfile
[mark@solaris:~] # ls -lh /mnt/testvol/testfile 
-rw-r--r-- 1 root root 0 2007-07-06 22:57 /mnt/testvol/testfile
[mark@solaris:~] # sudo zfs snapshot test/testvol@1

Now we've got our snapshot, we can delete that "testfile", as it's safely on the snapshot. The process of taking the snapshot created another block device under /dev/zvol, which can be mounted read-only :

[mark@solaris:~] # sudo rm /mnt/testvol/testfile 
[mark@solaris:~] # sudo mkdir /mnt/testsnapshot
[mark@solaris:~] # sudo mount -F ufs -o ro /dev/zvol/dsk/test/testvol@1 /mnt/testsnapshot
[mark@solaris:~] # ls /mnt/testsnapshot/
lost+found/  testfile

And just to confirm, zfs now shows our snapshotted volume :

[mark@solaris:~] # zfs list
NAME                   USED  AVAIL  REFER  MOUNTPOINT
test                  60.1M  3.39M  24.5K  /test
test/testvol          5.14M  58.2M  5.06M  -
test/testvol@1        74.5K      -  5.06M  -

How cool is that ?