对接GlusterFS

# gfs01节点更新hostname
hostname gfs01
echo "gfs01" > /etc/hostname
# gfs02节点更新hostname
hostname gfs02
echo "gfs02" > /etc/hostname
# gfs01/gfs02配置hosts解析
[root@gfs01 ~]# cat /etc/hosts
10.10.10.13 gfs02
10.10.10.14 gfs01

# 查看可用磁盘
fdisk -l
# 分区并格式化
mkfs.xfs  /dev/vdb1
mkdir  -p /data
echo "/dev/vdb1  /data  xfs  defaults 1 2" >>/etc/fstab
# 挂载
mount -a

# 确定/data挂载
df -h | grep data

yum install centos-release-gluster -y
yum install glusterfs-server -y
systemctl start glusterd.service
systemctl enable glusterd.service
systemctl status glusterd.service

[root@gfs01 ~]# gluster peer probe gfs02
peer probe: success.

[root@gfs02 ~]# gluster peer status
Number of Peers: 1

Hostname: gfs01
Uuid: f88992e2-4a3f-4cbd-b79b-a5b7e28881dd
State: Peer in Cluster (Connected)

# 节点gfs01,gfs02都需执行  
mkdir  -p /data/glusterfs

# 在gfs01执行创建卷操作
gluster volume create data replica 2 gfs02:/data/glusterfs gfs01:/data/glusterfs

Replica 2 volumes are prone to split-brain. Use Arbiter or Replica 3 to avoid this. See: http://docs.gluster.org/en/latest/Administrator%20Guide/Split%20brain%20and%20ways%20to%20deal%20with%20it/.
Do you still want to continue?
 (y/n) y
volume create: data: success: please start the volume to access data

# 在gfs02 查看卷信息
[root@gfs02 ~]# gluster volume info

Volume Name: data
Type: Replicate
Volume ID: 1c27e490-af0c-4794-af45-60e960c6eb47
Status: Created
Snapshot Count: 0
Number of Bricks: 1 x 2 = 2
Transport-type: tcp
Bricks:
Brick1: gfs02:/data/glusterfs
Brick2: gfs01:/data/glusterfs
Options Reconfigured:
transport.address-family: inet
nfs.disable: on
performance.client-io-threads: off
# 启动卷
gluster volume start data

# 挂载
[root@gfs01 ~]# mount -t glusterfs gfs02:/data /mnt
[root@gfs02 ~]# mount -t glusterfs gfs01:/data /mnt

# gfs02 写文件
touch /mnt/{1..10}.txt

# gfs01 验证是否生成{1..10}.txt
ls /mnt/ | wc -l

10.10.10.13 gfs02
10.10.10.14 gfs01

yum install -y glusterfs-fuse

# 创建挂载点
mkdir /grdata
# 开机挂载,更新/etc/fstab
gfs01:/data /grdata glusterfs   backupvolfile-server=gfs02,use-readdirp=no,log-level=WARNING,log-file=/var/log/gluster.log 0 0
# 挂载
mount -a
# 验证,应该存有安装时的验证文件
ls /grdata/
rm -rf /grdata/*.txt

1. 先切换计算节点，后切换管理节点
2. 计算节点将默认的nfs存储摘掉，编辑fstab文件，切换到GlusterFS,重新挂载
3. 管理节点需要先停服务
systemctl stop node
systemctl stop rbd-repo
systemctl stop rbd-hub
systemctl stop rbd-app-ui
systemctl stop rbd-gateway

systemctl stop rbd-eventlog
systemctl stop rbd-worker
systemctl stop rbd-chaos
systemctl stop rbd-api
cclear
4. 确定服务都停了之后,第一个管理节点操作
mv /grdata /backup
5. 编辑/etc/fstab
mkdir /grdata
mount -a 
6. 同步数据
cp -a /backup/. /grdata/
7. 其他管理节点，修改/etc/fstab重新挂载/grdata(如果有)
8. 迁移完成确定集群状态: grctl cluster
9. 第一个管理节点关闭nfs服务
systemctl stop nfs-server
systemctl disable nfs-server

# 下载源码
git clone https://github.com/goodrain/rainbond-glusterfs.git
cd rainbond-glusterfs
# 编辑 inventory/hosts 文件
all下为所有节点信息
gfscluster下为gfscluster server端(存储节点)
rainbondcluster下为gfscluster client端(即管理节点，计算节点)
# 执行安装
./setup.sh