5 As is detailed in the operational information below, snapshots
6 are definitely alpha-test code and are NOT yet ready for production
7 use. Much remains to be done to make them really useful, but I
8 wanted to let folks get a chance to try it out and start reporting
9 bugs and other shortcomings. Such reports should be sent to
10 Kirk McKusick <mckusick@mckusick.com>.
13 Snapshot Copyright Restrictions
15 Snapshots have been introduced to FreeBSD with a `Berkeley-style'
16 copyright. The file implementing snapshots resides in the sys/ufs/ffs
17 directory and is compiled into the generic kernel by default.
22 To create a snapshot of your /var filesystem, run the command:
24 mount -u -o snapshot /var/snapshot/snap1 /var
26 This command will take a snapshot of your /var filesystem and
27 leave it in the file /var/snapshot/snap1. Note that snapshot
28 files must be created in the filesystem that is being snapshotted.
29 I use the convention of putting a `snapshot' directory at the
30 root of each filesystem into which I can place snapshots.
31 You may create up to 20 snapshots per filesystem. Active snapshots
32 are recorded in the superblock, so they persist across unmount
33 and remount operations and across system reboots. When you
34 are done with a snapshot, it can be removed with the `rm'
35 command. Snapshots may be removed in any order, however you
36 may not get back all the space contained in the snapshot as
37 another snapshot may claim some of the blocks that it is releasing.
38 Note that the `schg' flag is set on snapshots to ensure that
39 not even the root user can write to them. The unlink command
40 makes an exception for snapshot files in that it allows them
41 to be removed even though they have the `schg' flag set, so it
42 is not necessary to clear the `schg' flag before removing a
45 Once you have taken a snapshot, there are three interesting
46 things that you can do with it:
48 1) Run fsck on the snapshot file. Assuming that the filesystem
49 was clean when it was mounted, you should always get a clean
50 (and unchanging) result from running fsck on the snapshot.
51 If you are running with soft updates and rebooted after a
52 crash without cleaning up the filesystem, then fsck of the
53 snapshot may find missing blocks and inodes or inodes with
54 link counts that are too high. I have not yet added the
55 system calls to allow fsck to add these missing resources
56 back to the filesystem - that will be added once the basic
57 snapshot code is working properly. So, view those reports
58 as informational for now.
60 2) Run dump on the snapshot. You will get a dump that is
61 consistent with the filesystem as of the timestamp of the
64 3) Mount the snapshot as a frozen image of the filesystem.
65 To mount the snapshot /var/snapshot/snap1:
67 mdconfig -a -t vnode -f /var/snapshot/snap1 -u 4
68 mount -r /dev/md4 /mnt
70 You can now cruise around your frozen /var filesystem
71 at /mnt. Everything will be in the same state that it
72 was at the time the snapshot was taken. The one exception
73 is that any earlier snapshots will appear as zero length
74 files. When you are done with the mounted snapshot:
79 Note that under some circumstances, the process accessing
80 the frozen filesystem may deadlock. I am aware of this
81 problem, but the solution is not simple. It requires
82 using buffer read locks rather than exclusive locks when
83 traversing the inode indirect blocks. Until this problem
84 is fixed, you should avoid putting mounted snapshots into
90 It takes about 30 seconds to create a snapshot of an 8Gb filesystem.
91 Of that time 25 seconds is spent in preparation; filesystem activity
92 is only suspended for the final 5 seconds of that period. Snapshot
93 removal of an 8Gb filesystem takes about two minutes. Filesystem
94 activity is never suspended during snapshot removal.
96 The suspend time may be expanded by several minutes if a process
97 is in the midst of removing many files as all the soft updates
98 backlog must be cleared. Generally snapshots do not slow the system
99 down appreciably except when removing many small files (i.e., any
100 file less than 96Kb whose last block is a fragment) that are claimed
101 by a snapshot. Here, the snapshot code must make a copy of every
102 released fragment which slows the rate of file removal to about
103 twenty files per second once the soft updates backlog limit is
109 For more general information on snapshots, please see:
110 http://www.mckusick.com/softdep/