RBD块设备在Ceph分布式存储中的具体应用

确保集群状态正常（具体配置过程略）：
可参照 https://blog.51cto.com/jdonghong/244175 上半部分配置。
CEPH环境配置
开始部署RBD or RADOS Block Device

客户端安装ceph（本案例客户端为192.168.27.210，192.168.26.112）
ceph-deploy install bddb.com
推送配置文件到客户端。
[[email protected] idc-cluster]# ceph-deploy admin bddb.com

客户端创建rdb块镜像设备

[[email protected] ceph]# rbd create idc --size 4096 --image-feature layering
[[email protected] ceph]# rbd ls
idc
[[email protected] ceph]#
客户端映射创建的块镜像设备
[[email protected] ceph]# rbd map idc
/dev/rbd0
[[email protected] ceph]#

格式化块设备创建文件系统（客户端节点上）。
[[email protected] ceph]# mkfs.xfs /dev/rbd0

挂载目录：[[email protected] ceph]# mount /dev/rbd/rbd/idc /ceph/rbd

查看状态并存放或创建测试文件：

[[email protected] ceph]# ls /ceph
rbd rbd2
[[email protected] ceph]# ls /ceph -l
total 0
drwxr-xr-x 2 ceph ceph 6 Aug 23 11:17 rbd
drwxr-xr-x 2 ceph ceph 6 Aug 26 10:01 rbd2
[[email protected] ceph]# mount /dev/rbd
rbd/ rbd0
[[email protected] ceph]# mount /dev/r
random raw/ rbd/ rbd0 rtc rtc0
[[email protected] ceph]# mount /dev/rbd
rbd/ rbd0
[[email protected] ceph]# mount /dev/rbd/rbd/idc /ceph/rbd
[[email protected] ceph]# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda2 60G 32G 29G 53% /
/dev/sda5 60G 3.5G 57G 6% /data
/dev/sda1 497M 148M 350M 30% /boot
tmpfs 184M 0 184M 0% /run/user/0
/dev/rbd0 3.9G 16M 3.6G 1% /ceph/rbd
[[email protected] ceph]# ls

ceph.client.admin.keyring ceph.conf rbdmap tmp3nza8m tmpGs8qYv tmpNTb5P9 tmpOJovru
[[email protected] ceph]# cd /ceph/rbd
[[email protected] rbd]# ls
123.txt 1.txt lost+found my1.txt my2.txt my3.txt
[[email protected] rbd]# rbd map isc
/dev/rbd1
[[email protected] rbd]# lsblk -f

NAME FSTYPE LABEL UUID MOUNTPOINT
sda
├─sda1 xfs 5f75d5de-3e02-43e3-a36d-57bc39e9a5ae /boot
├─sda2 xfs bab8e8ae-cb5e-4299-b879-54960f1a24b9 /
├─sda3 swap ce44af87-b8a7-40d2-8504-1c8fae81f613 [SWAP]
├─sda4
└─sda5 xfs cb8c9f72-8154-4ae9-aa57-54703dedfd06 /data
sr0
rbd0 ext4 cf1f5bc8-5dd1-44d4-87a6-0c55d39405fe /ceph/rbd
rbd1 xfs 42989fcc-0746-4848-a957-0c01704865b8
[[email protected] rbd]# mount /dev/rbd
rbd/ rbd0 rbd1
[[email protected] rbd]# mount /dev/rbd/rbd/isc
123.txt 1.txt lost+found/ my1.txt my2.txt my3.txt
[[email protected] rbd]# mount /dev/rbd/rbd/isc /ceph/rbd
rbd/ rbd2/
[[email protected] rbd]# mount /dev/rbd/rbd/isc /ceph/rbd2/
[[email protected] rbd]# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda2 60G 32G 29G 53% /
/dev/sda5 60G 3.5G 57G 6% /data
/dev/sda1 497M 148M 350M 30% /boot
tmpfs 184M 0 184M 0% /run/user/0
/dev/rbd0 3.9G 16M 3.6G 1% /ceph/rbd
/dev/rbd1 10G 33M 10G 1% /ceph/rbd2
[[email protected] rbd]# cd /ceph/rbd2/
[[email protected] rbd2]# ls
[[email protected] rbd2]# touch {1..3}.txt
[[email protected] rbd2]# ls
1.txt 2.txt 3.txt
[[email protected] rbd2]# echo my test >1.txt
[[email protected] rbd2]# ls
1.txt 2.txt 3.txt
[[email protected] rbd2]# cat 1.txt
my test
[[email protected] rbd2]#

再在另一个客户端映射块设备并挂载，观察RBD效果：
首先在另一个客户端观察设备镜像状态：
[[email protected] rbd]# rbd ls
idc
isc
[[email protected] rbd]# rbd info idc isc
rbd: too many arguments
[[email protected] rbd]# rbd info idc
rbd image ‘idc‘:
size 4096 MB in 1024 objects
order 22 (4096 kB objects)
block_name_prefix: rbd_data.85456b8b4567
format: 2
features: layering
flags:
[[email protected] rbd]# rbd info isc
rbd image ‘isc‘:
size 10240 MB in 2560 objects
order 22 (4096 kB objects)
block_name_prefix: rbd_data.148a56b8b4567
format: 2
features: layering
flags:
[[email protected] rbd]#
映射rbd块镜像设备到本机（map）
[[email protected] rbd]# lsblk -f
NAME FSTYPE LABEL UUID MOUNTPOINT
fd0
sda
├─sda1 xfs fb314ba6-93e0-4d7d-bb80-9c6e5a92fd61 /boot
└─sda2 LVM2_member 3Zne0f-m5MZ-OP67-TQ2a-Lnzr-SGME-UNJnMK
├─centos-root xfs d009c83b-2ca3-4642-a956-9f967fa249e6 /
├─centos-swap swap cf124d61-2df9-44a4-bba2-cee94056f547 [SWAP]
└─centos-data xfs 538b2348-c8cd-4755-9ba9-2f3b10fb8f33 /data
sr0
rbd0 ext4 cf1f5bc8-5dd1-44d4-87a6-0c55d39405fe /ceph/rbd
[[email protected] rbd]# rbd map isc
/dev/rbd1
[[email protected] rbd]#

查看map后的映射效果：
[[email protected] rbd]# lsblk -f

挂载映射设备并观察效果：（注意这里无需再格式，因为在另一个客户端已经格式化，生成文件系统类型，否则会造成数据破坏，甚至报错及集群出错等可能）
[[email protected] rbd]# mount /dev/rbd/rbd/isc /ceph/rbd2
[[email protected] rbd]# df -h

进入挂载目录观察另一客户端创建的文件及内容是否存在：
[[email protected] rbd]# cd /ceph/rbd2/
[[email protected] rbd2]# ls
1.txt 2.txt 3.txt
[[email protected] rbd2]# cat 1.txt
my test
[[email protected] rbd2]#

切到另一个客户端观察文件变化：

文件内容无变化，尝试重新映射观察变化：
[[email protected] rbd2]# cd
[[email protected] ~]# umount /ceph/rbd2/
[[email protected] ~]# rbd unmap isc
[[email protected] ~]# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda2 60G 32G 29G 53% /
/dev/sda5 60G 3.5G 57G 6% /data
/dev/sda1 497M 148M 350M 30% /boot
tmpfs 184M 0 184M 0% /run/user/0
/dev/rbd0 3.9G 16M 3.6G 1% /ceph/rbd
[[email protected] ~]#

重新映射后文件内容更新了：
[[email protected] ~]# lsblk -f
NAME FSTYPE LABEL UUID MOUNTPOINT
sda
├─sda1 xfs 5f75d5de-3e02-43e3-a36d-57bc39e9a5ae /boot
├─sda2 xfs bab8e8ae-cb5e-4299-b879-54960f1a24b9 /
├─sda3 swap ce44af87-b8a7-40d2-8504-1c8fae81f613 [SWAP]
├─sda4
└─sda5 xfs cb8c9f72-8154-4ae9-aa57-54703dedfd06 /data
sr0
rbd0 ext4 cf1f5bc8-5dd1-44d4-87a6-0c55d39405fe /ceph/rbd
rbd1 xfs 42989fcc-0746-4848-a957-0c01704865b8
[[email protected] ~]# mount /dev/rbd/rbd/isc /ceph/rbd2/
[[email protected] ~]# cd /ceph/rbd2/
[[email protected] rbd2]# ls
1.txt 2.txt 3.txt
[[email protected] rbd2]# cat 1.txt
my test
my test
my test
my test
my test

观察编辑同一文件效果：

并无提示
更改内容：

出现不一致。
扩（缩）容：
[[email protected] ~]# rbd resize --size 20480 rbd/idc

[[email protected] ~]# rbd info idc
rbd image ‘idc‘:
size 20480 MB in 5120 objects
order 22 (4096 kB objects)
block_name_prefix: rbd_data.85456b8b4567
format: 2
features: layering
flags:
[[email protected] ~]# resize2fs /dev/rbd0

查看文件依旧存在：
[[email protected] ~]# cd /ceph/rbd
[[email protected] rbd]# ls
123.txt 1.txt lost+found my1.txt my2.txt my3.txt
[[email protected] rbd]# cat my1.txt
127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4
::1 localhost localhost.localdomain localhost6 localhost6.localdomain6
192.168.27.210 master
192.168.27.211 client1
192.168.27.212 client2
192.168.27.213 client3
192.168.26.112 BDDB.com
192.168.26.112 bddb.com bddb

小结： rbd能实现在多客户端远程访问rbd块设备映射做为本地存储设备挂载使用，但不支持同一时刻多客户端同时挂载使用，会出现数据异步的情况，导致数据错乱，因此是非共享型异步传输方式，rbd支持在线扩(缩)容，最新ceph支持 layering,striping exclusive lock, object map,fast diff ,deep-flatten 等新的new features
layering。