Ceph支持一个非常好的特性，以COW（写时复制）的方式从RBD快照创建克隆，在Ceph中被称为快照分层。分层特性允许用户创建多个CEPH RBD克隆实例。这些特性应用于OpenStack等云平台中
，使用快照形式保护ceph RBD 镜像，快照是只读的，但COW克隆是完全可以写 ，可以多次来孵化实例，对云平台来说是非常有用的。
Ceph RBD镜像有format-1 和 format-2两种类型，RBD支持这两种类型，但是分层特性COW克隆特性只支持format-2镜像，默认RBD创建的镜像是format-1。
ceph snap几个主要命令
ceph snap create (snap add) Create a snapshot.
snap list (snap ls) Dump list of image snapshots.
snap protect Prevent a snapshot from being deleted.
snap purge Deletes all snapshots.
snap remove (snap rm) Deletes a snapshot.
snap rename Rename a snapshot.
snap rollback (snap revert) Rollback image to snapshot.
snap unprotect Allow a snapshot to be deleted.
一、创建RBD镜像实例：
1、在pool1上创建一个RBD镜像，不使用-p <pool name>默认在rbd存储池上创建
[[email protected] ~]# rbd create rbd1 -p pool1 --size 5G --image-format 1
rbd: image format 1 is deprecated
查看创建结果，使用-p指定存储池
[[email protected] ~]# rbd list -p pool1
rbd1
[[email protected] ~]# rbd info --image pool1/rbd1
rbd image ‘rbd1‘:
size 5120 MB in 1280 objects
order 22 (4096 kB objects)
block_name_prefix: rb.0.15d23.238e1f29
format: 1
2、映射块设备
[[email protected] ~]# rbd map --image pool1/rbd1
/dev/rbd3
查看映射的块设备
[[email protected] ~]# rbd showmapped
id pool image snap device
0 rbd block-device1 - /dev/rbd0
1 rbd block-device3 - /dev/rbd1
2 rbd block-device4 - /dev/rbd2
3 pool1 rbd1 - /dev/rbd3
3、挂载文件系统使用
[[email protected] ~]# mkfs.xfs /dev/rbd3
meta-data=/dev/rbd3 isize=256 agcount=9, agsize=162816 blks
= sectsz=512 attr=2, projid32bit=1
= crc=0 finobt=0
data = bsize=4096 blocks=1310720, imaxpct=25
= sunit=1024 swidth=1024 blks
naming =version 2 bsize=4096 ascii-ci=0 ftype=0
log =internal log bsize=4096 blocks=2560, version=2
= sectsz=512 sunit=8 blks, lazy-count=1
realtime =none extsz=4096 blocks=0, rtextents=0
[[email protected] ~]# mkdir /mnt/rbd3
[[email protected] ~]# mount /dev/rbd3 /mnt/rbd3
[[email protected] ~]# mount | grep rbd3
/dev/rbd3 on /mnt/rbd3 type xfs (rw,relatime,attr2,inode64,sunit=8192,swidth=8192,noquota)
[[email protected] ~]# touch /mnt/rbd3/test
[[email protected] ~]# ls /mnt/rbd3/
test
4、创建快照
[[email protected] ~]# rbd snap create pool1/[email protected]
[[email protected] ~]# rbd snap ls pool1/rbd1
SNAPID NAME SIZE
4 snapshot1 5120 MB
5、在rbd3目录下创建文件test1
[[email protected] ~]# touch /mnt/rbd3/test1
[[email protected] ~]# ls /mnt/rbd3/
test test1
6、恢复rbd1的快照
[[email protected] ~]# rbd snap rollback pool1/[email protected]
Rolling back to snapshot: 100% complete...done.
语法：rbd snap rollback <pool-name>/<image-name>@<snap-name>
7、重新挂载文件系统，发现已经恢复到之前的状态了
[[email protected] ~]# umount /mnt/rbd3
[[email protected] ~]# mount /dev/rbd3 /mnt/rbd3/
[[email protected] ~]# ls /mnt/rbd3/
test
[[email protected] ~]#
8、不再使用镜像时可以删除掉镜像
[[email protected]2 ~]# rbd snap rm pool1/[email protected]
[[email protected] ~]# rbd snap ls pool1/rbd1
语法：rbd snap rm <pool-name>/<image-name>@<snap-name>
9、删除多个快照
[[email protected] ~]# rbd snap purge pool1/rbd1
语法：rbd snap purge <pool-name>/<image-name>
二、使用RBD克隆
创建rbd2,类型为format2的RBD镜像
[[email protected] ~]# rbd create rbd2 --size 10240 --image-format 2
[[email protected] ~]# rbd info --image rbd2
rbd image ‘rbd2‘:
size 10240 MB in 2560 objects
order 22 (4096 kB objects)
block_name_prefix: rbd_data.15821238e1f29
format: 2
features: layering, exclusive-lock, object-map, fast-diff, deep-flatten
flags:
创建RBD镜像快照
[[email protected] ~]# rbd snap create rbd/[email protected]
[[email protected] ~]# rbd snap ls rbd/rbd2
SNAPID NAME SIZE
7 snapshot1 10240 MB
创建COW镜像前先要保护这个快照，这是非常重要的一步
[[email protected] ~]# rbd snap protect rbd/[email protected]
通过快照创建一个克隆的RBD镜像
[[email protected] ~]# rbd clone rbd/[email protected] rbd/clone_rbd2
[[email protected] ~]# rbd info --image rbd/clone_rbd2
rbd image ‘clone_rbd2‘:
size 10240 MB in 2560 objects
order 22 (4096 kB objects)
block_name_prefix: rbd_data.15c1b3d1b58ba
format: 2
features: layering, exclusive-lock, object-map, fast-diff, deep-flatten
flags:
parent: rbd/[email protected]
overlap: 10240 MB
创建克隆好了一个依赖于父镜像快照的RBD镜像后，为了让这个克隆的RBD镜像独立于父镜像，我们需要将父镜像的信息合并flattern到子镜像，一旦合并完成，RBD镜像和它的父镜像就不会存在任何关系了。
使用flatten合并
[[email protected] ~]# rbd flatten rbd/clone_rbd2
Image flatten: 100% complete...done.
合并完成后，查看这个镜像信息，发现父镜像/快照名字不存大了
[[email protected] ~]# rbd info --image clone_rbd2
rbd image ‘clone_rbd2‘:
size 10240 MB in 2560 objects
order 22 (4096 kB objects)
block_name_prefix: rbd_data.15c1b3d1b58ba
format: 2
features: layering, exclusive-lock, object-map, fast-diff, deep-flatten
flags:
[[email protected] ~]#
如果你不再使用父镜像，可以先移除它的保护状态，再删除它
[[email protected] ~]# rbd snap unprotect rbd/[email protected]
[[email protected] ~]# rbd snap rm rbd/[email protected]