CEPH RBD集群使用

准备阶段

RBD client端安装ceph工具集合

创建一个块设备映像

ceph-client-node@ceph-client-node:~$ rbd create harvis_rbd -m 192.168.1.252 -k /etc/ceph/ceph.client.admin.keyring –size 4096

将块设备映像映射到一个块设备

rbd feature disable harvis_rbd fast-diff

rbd feature disable harvis_rbd deep-flatten

rbd feature disable harvis_rbd object-map

rbd feature disable harvis_rbd exclusive-lock

ceph-client-node@ceph-client-node:~$ sudo rbd map harvis_rbd –name client.admin -m 192.168.1.252 -k /etc/ceph/ceph.client.admin.keyring

映射完成，用命令lsblk可以看到

ceph-client-node@ceph-client-node:~$ ls -l /dev/rbd/rbd/harvis_rbd

lrwxrwxrwx 1 root root 10 Apr 17 13:12 /dev/rbd/rbd/harvis_rbd -> ../../rbd0

ceph-client-node@ceph-client-node:~$ ls -l /dev/rbd0

brw-rw—- 1 root disk 251, 0 Apr 17 13:12 /dev/rbd0

将块设备格式化为xfs格式

ceph-client-node@ceph-client-node:~$ sudo mkfs.xfs /dev/rbd/rbd/harvis_rbd

ceph-client-node@ceph-client-node:~$ ls -l /dev/disk/by-uuid/

将新格式化的xfs找个地方挂载上

ceph-client-node@ceph-client-node:~$ sudo mkdir /mnt/ceph-block-device

ceph-client-node@ceph-client-node:~$ sudo mount /dev/rbd/rbd/harvis_rbd /mnt/ceph-block-device

ceph-client-node@ceph-client-node:/mnt/ceph-block-device$ sudo cp /boot/initrd.img-4.4.0-31-generic initrd.img

查看ceph -s观察集群消耗情况，拷贝前消耗178M，拷贝后消耗263M， 差值85M， 文件initrd.img大小21M. 这里有没有使用多副本的策略呢？