ZFS Snapshots and Clones

Published 02 Jun 2009 Read time 5 min(s) 35 sec(s) (1487 views). Solaris

The difference between a snapshot and a clone is that a clone is a writeable, mountable copy of the file system. This capability allows us to store multiple copies of mostly-shared data in a very space-efficient way.

A snapshot is an read-only image of a file system,. When taking a snapshot, it is stored in a away so further transactions on a filesystem will only be carried out on the original file filesystem and not on the snapshot itself. This way it will be possible to rollback to get back to a previous state.

Clones and snapshots are not data copies but state copies which means they don't use any space when created. It is when the original filesystem is modified that the differences are stored.

If a rollback is performed, these differences are overwritten and space is freed-up once again.

Note: Clones can only be created from an existing snapshot.

zfs send and zfs receive allow clones of filesystems to be sent to for example a development environment.

To create a snapshot:

# zfs snapshot pool-name/filesystem-name@snapshot-name

To clone a snapshot:

# zfs clone snapshot-name filesystem-name

To roll back to a snapshot:

zfs rollback pool-name/filesystem-name@snapshot-name

Examples:

List current ZFS filesystems:

# zfs list
   NAME            USED  AVAIL  REFER  MOUNTPOINT
   pool            149K  1.95G    31K  /pool
   pool/data01      31K   512M    31K  /data01

Create a snapshot from a ZFS file system:

# zfs snapshot pool/data01@snapshot01
# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                       150K  1.95G    31K  /pool
   pool/data01                 31K   512M    31K  /data01
   pool/data01@snapshot01        0      -    31K  -

Rollback a ZFS file system to a previous state:

First let's do some modifications on filesystem:

# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                       152K  1.95G    31K  /pool
   pool/data01                 31K   512M    31K  /data01
   pool/data01@snapshot01        0      -    31K  -

# cd /data01
# mkfile 64m file01
# ls -l
   total 131095
   -rw------T   1 root     root     67108864 Jun  1 10:42 file01

Notice the difference in size of the snapshot:

# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                      63.7M  1.89G    31K  /pool
   pool/data01               63.6M   448M  63.5M  /data01
   pool/data01@snapshot01      19K      -    31K  -

Now perform the rollback:
```
# zfs rollback pool/data01@snapshot01
```

The file has disappeared, and the size of the snapshot has reduced.

# ls -l
   total 0
# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                       180K  1.95G    31K  /pool
   pool/data01                 32K   512M    31K  /data01
   pool/data01@snapshot01       1K      -    31K  -

Remove a snapshot:

# zfs destroy pool/data01@snapshot01
# zfs list
   NAME            USED  AVAIL  REFER  MOUNTPOINT
   pool            161K  1.95G    31K  /pool
   pool/data01    31K   512M    31K  /data01

Multiple snapshots can be taken at different times to have different points in time of recovery:

Create multiple snapshots:

# zfs snapshot pool/data01@snapshot01
# mkfile 64m /data01/file01
# zfs snapshot pool/data01@snapshot02
# mkfile 64m /data01/file02
# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                       118M  1.84G    31K  /pool
   pool/data01                117M   395M   117M  /data01
   pool/data01@snapshot01      19K      -    31K  -
   pool/data01@snapshot02      19K      -  64.0M  -
# ls -l /data01
   total 262186
   -rw------T   1 root     root     67108864 Jun  1 10:53 file01
   -rw------T   1 root     root     67108864 Jun  1 10:53 file02

In this scenario we can only rollback the most recent snapshot. If we attempt to rollback the oldest snapshot we receive the following:

# zfs rollback pool/data01@snapshot01
   cannot rollback to 'pool/data01@snapshot01': more recent snapshots exist
   use '-r' to force deletion of the following snapshots:
   pool/data01@snapshot02

If we need to rollback to first snapshot, first we have to rollback to the newer one, destroy it and, then rollback to the oldest snapshot:

# zfs rollback pool/data01@snapshot02
# ls -l /data01/
   total 131093
   -rw------T   1 root     root     67108864 Jun  1 10:53 file01
# zfs destroy pool/data01@snapshot02
# zfs rollback pool/data01@snapshot01
# ls -l /data01
   total 0

Displaying snapshots

# zfs list -t snapshot
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool/data01@snapshot01      18K      -    31K  -
   pool/data01@snapshot02        0      -    31K  -

Accessing snapshot contents through the .zfs/snapshot in the /pool-name directory. This can allow end users to recover their files without system administrator intervention.

# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                       220K  1.95G    31K  /pool
   pool/data01               49K   512M    31K  /data01
   pool/data01@snapshot01    18K      -    31K  -
   pool/data01@snapshot02      0      -    31K  -
# cd /data01/.zfs/snapshot/
# ls -l
   total 6
   drwxr-xr-x   2 root     root           2 Jun  1 10:38 snapshot01
   drwxr-xr-x   2 root     root           3 Jun  1 11:01 snapshot02

Each of the directories contains directory/file structures existing at the moment when snapshot was taken

Cloning a snapshot:

# zfs clone pool/data01@snapshot01 pool/data02
# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                      64.3M  1.89G    31K  /pool
   pool/data01             64.1M   448M  64.0M  /data01
   pool/data01@snapshot01    19K      -    31K  -
   pool/data02                1K  1.89G    31K  /pool/data02

Snapshot pool/data01@snapshot01 has been copied and will be writeable on pool/data02 clone

Removing a clone/snapshot:

# zfs destroy pool/data02
# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                      64.2M  1.89G    32K  /pool
   pool/data01               64.1M   448M  64.0M  /data01
   pool/data01@snapshot01      19K      -    31K  -

Note: If a snapshot has one or more clones we won't be able to destroy it unless clones are destroyed first:

# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                      64.3M  1.89G    31K  /pool
   pool/data01             64.1M   448M  64.0M  /data01
   pool/data01@snapshot01    19K      -    31K  -
   pool/data02                1K  1.89G    31K  /pool/data02
# zfs destroy pool/data01@snapshot01
   cannot destroy 'pool/data01@snapshot01': snapshot has dependent clones
   use '-R' to destroy the following datasets:
   pool/data02
# zfs destroy pool/data02
# zfs destroy pool/data01@snapshot01
# zfs list
   NAME            USED  AVAIL  REFER  MOUNTPOINT
   pool           64.2M  1.89G    31K  /pool
   pool/data01    64.0M   448M  64.0M  /data01

Promoting a clone

Once a clone in place, we can use to replace original dataset. We will make clone independent of snapshot it was created from and, then, remove snapshot(s) and origin filesystem so our clone will replace it

# zfs list
   NAME            USED  AVAIL  REFER  MOUNTPOINT
   pool           64.2M  1.89G    31K  /pool
   pool/data01    64.0M   448M  64.0M  /data01
# ll /data01
   total 131093
   -rw------T   1 root     root     67108864 Jun  1 11:01 file01
# zfs snapshot pool/data01@snapshot01
# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                      64.2M  1.89G    31K  /pool
   pool/data01               64.0M   448M  64.0M  /data01
   pool/data01@snapshot01        0      -  64.0M  -
# zfs clone -o mountpoint=/data02 pool/data01@snapshot01 pool/data02
# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                      64.2M  1.89G    31K  /pool
   pool/data01               64.0M   448M  64.0M  /data01
   pool/data01@snapshot01        0      -  64.0M  -
   pool/data02                  1K  1.89G  64.0M  /data02
# zfs promote pool/data02
# zfs list
   NAME                       USED  AVAIL  REFER  MOUNTPOINT
   pool                      64.2M  1.89G    31K  /pool
   pool/data01                 0   512M  64.0M  /data01
   pool/data02               64.0M  1.89G  64.0M  /data02
   pool/data02@snapshot01       1K      -  64.0M  -
# zfs destroy pool/data01
# zfs destroy pool/data02@snapshot01
# zfs list
   NAME            USED  AVAIL  REFER  MOUNTPOINT
   pool           64.2M  1.89G    31K  /pool
   pool/data02    64.0M  1.89G  64.0M  /data02
# ll /data01
   total 0
# ll /data02
   total 131093
   -rw------T   1 root     root     67108864 Jun  1 11:01 file01