1Device-mapper snapshot support
2==============================
3
4Device-mapper allows you, without massive data copying:
5
6*) To create snapshots of any block device i.e. mountable, saved states of
7the block device which are also writable without interfering with the
8original content;
9*) To create device "forks", i.e. multiple different versions of the
10same data stream.
11*) To merge a snapshot of a block device back into the snapshot's origin
12device.
13
14In the first two cases, dm copies only the chunks of data that get
15changed and uses a separate copy-on-write (COW) block device for
16storage.
17
18For snapshot merge the contents of the COW storage are merged back into
19the origin device.
20
21
22There are three dm targets available:
23snapshot, snapshot-origin, and snapshot-merge.
24
25*) snapshot-origin <origin>
26
27which will normally have one or more snapshots based on it.
28Reads will be mapped directly to the backing device. For each write, the
29original data will be saved in the <COW device> of each snapshot to keep
30its visible content unchanged, at least until the <COW device> fills up.
31
32
33*) snapshot <origin> <COW device> <persistent?> <chunksize>
34
35A snapshot of the <origin> block device is created. Changed chunks of
36<chunksize> sectors will be stored on the <COW device>.  Writes will
37only go to the <COW device>.  Reads will come from the <COW device> or
38from <origin> for unchanged data.  <COW device> will often be
39smaller than the origin and if it fills up the snapshot will become
40useless and be disabled, returning errors.  So it is important to monitor
41the amount of free space and expand the <COW device> before it fills up.
42
43<persistent?> is P (Persistent) or N (Not persistent - will not survive
44after reboot).  O (Overflow) can be added as a persistent store option
45to allow userspace to advertise its support for seeing "Overflow" in the
46snapshot status.  So supported store types are "P", "PO" and "N".
47
48The difference between persistent and transient is with transient
49snapshots less metadata must be saved on disk - they can be kept in
50memory by the kernel.
51
52When loading or unloading the snapshot target, the corresponding
53snapshot-origin or snapshot-merge target must be suspended. A failure to
54suspend the origin target could result in data corruption.
55
56
57* snapshot-merge <origin> <COW device> <persistent> <chunksize>
58
59takes the same table arguments as the snapshot target except it only
60works with persistent snapshots.  This target assumes the role of the
61"snapshot-origin" target and must not be loaded if the "snapshot-origin"
62is still present for <origin>.
63
64Creates a merging snapshot that takes control of the changed chunks
65stored in the <COW device> of an existing snapshot, through a handover
66procedure, and merges these chunks back into the <origin>.  Once merging
67has started (in the background) the <origin> may be opened and the merge
68will continue while I/O is flowing to it.  Changes to the <origin> are
69deferred until the merging snapshot's corresponding chunk(s) have been
70merged.  Once merging has started the snapshot device, associated with
71the "snapshot" target, will return -EIO when accessed.
72
73
74How snapshot is used by LVM2
75============================
76When you create the first LVM2 snapshot of a volume, four dm devices are used:
77
781) a device containing the original mapping table of the source volume;
792) a device used as the <COW device>;
803) a "snapshot" device, combining #1 and #2, which is the visible snapshot
81   volume;
824) the "original" volume (which uses the device number used by the original
83   source volume), whose table is replaced by a "snapshot-origin" mapping
84   from device #1.
85
86A fixed naming scheme is used, so with the following commands:
87
88lvcreate -L 1G -n base volumeGroup
89lvcreate -L 100M --snapshot -n snap volumeGroup/base
90
91we'll have this situation (with volumes in above order):
92
93# dmsetup table|grep volumeGroup
94
95volumeGroup-base-real: 0 2097152 linear 8:19 384
96volumeGroup-snap-cow: 0 204800 linear 8:19 2097536
97volumeGroup-snap: 0 2097152 snapshot 254:11 254:12 P 16
98volumeGroup-base: 0 2097152 snapshot-origin 254:11
99
100# ls -lL /dev/mapper/volumeGroup-*
101brw-------  1 root root 254, 11 29 ago 18:15 /dev/mapper/volumeGroup-base-real
102brw-------  1 root root 254, 12 29 ago 18:15 /dev/mapper/volumeGroup-snap-cow
103brw-------  1 root root 254, 13 29 ago 18:15 /dev/mapper/volumeGroup-snap
104brw-------  1 root root 254, 10 29 ago 18:14 /dev/mapper/volumeGroup-base
105
106
107How snapshot-merge is used by LVM2
108==================================
109A merging snapshot assumes the role of the "snapshot-origin" while
110merging.  As such the "snapshot-origin" is replaced with
111"snapshot-merge".  The "-real" device is not changed and the "-cow"
112device is renamed to <origin name>-cow to aid LVM2's cleanup of the
113merging snapshot after it completes.  The "snapshot" that hands over its
114COW device to the "snapshot-merge" is deactivated (unless using lvchange
115--refresh); but if it is left active it will simply return I/O errors.
116
117A snapshot will merge into its origin with the following command:
118
119lvconvert --merge volumeGroup/snap
120
121we'll now have this situation:
122
123# dmsetup table|grep volumeGroup
124
125volumeGroup-base-real: 0 2097152 linear 8:19 384
126volumeGroup-base-cow: 0 204800 linear 8:19 2097536
127volumeGroup-base: 0 2097152 snapshot-merge 254:11 254:12 P 16
128
129# ls -lL /dev/mapper/volumeGroup-*
130brw-------  1 root root 254, 11 29 ago 18:15 /dev/mapper/volumeGroup-base-real
131brw-------  1 root root 254, 12 29 ago 18:16 /dev/mapper/volumeGroup-base-cow
132brw-------  1 root root 254, 10 29 ago 18:16 /dev/mapper/volumeGroup-base
133
134
135How to determine when a merging is complete
136===========================================
137The snapshot-merge and snapshot status lines end with:
138  <sectors_allocated>/<total_sectors> <metadata_sectors>
139
140Both <sectors_allocated> and <total_sectors> include both data and metadata.
141During merging, the number of sectors allocated gets smaller and
142smaller.  Merging has finished when the number of sectors holding data
143is zero, in other words <sectors_allocated> == <metadata_sectors>.
144
145Here is a practical example (using a hybrid of lvm and dmsetup commands):
146
147# lvs
148  LV      VG          Attr   LSize Origin  Snap%  Move Log Copy%  Convert
149  base    volumeGroup owi-a- 4.00g
150  snap    volumeGroup swi-a- 1.00g base  18.97
151
152# dmsetup status volumeGroup-snap
1530 8388608 snapshot 397896/2097152 1560
154                                  ^^^^ metadata sectors
155
156# lvconvert --merge -b volumeGroup/snap
157  Merging of volume snap started.
158
159# lvs volumeGroup/snap
160  LV      VG          Attr   LSize Origin  Snap%  Move Log Copy%  Convert
161  base    volumeGroup Owi-a- 4.00g          17.23
162
163# dmsetup status volumeGroup-base
1640 8388608 snapshot-merge 281688/2097152 1104
165
166# dmsetup status volumeGroup-base
1670 8388608 snapshot-merge 180480/2097152 712
168
169# dmsetup status volumeGroup-base
1700 8388608 snapshot-merge 16/2097152 16
171
172Merging has finished.
173
174# lvs
175  LV      VG          Attr   LSize Origin  Snap%  Move Log Copy%  Convert
176  base    volumeGroup owi-a- 4.00g
177