Glusterfs初试

Gluster的模式及介绍在此不表，这里只记录安装及配置过程。

1.整体环境

server1 : gfs1.cluster.com

server2 : gfs2.cluster.com

Client:

2.安装Gluster

• 下载软件

https://access.redhat.com/downloads/content/186/ver=3/rhel---7/3.4/x86_64/product-software

下载 Red Hat Gluster Storage Server 3.4 on RHEL 7 Installation DVD

安装RHEL 7.6的最小软件安装，将iso文件mount成cdrom, 然后修改yum源

mkdir -p /repo/base
mount /dev/cdrom /repo/base
vi /etc/yum.repos.d/base.repo

[rhel7.]
name=rhel7.
baseurl=file:///repo/base/
enabled=
gpgcheck=

• 安装

yum install -y redhat-storage-server
systemctl start glusterd
systemctl enable glusterd

systemctl status glusterd验证一下

[root@gfs1 mnt]# systemctl status glusterd
● glusterd.service - GlusterFS, a clustered file-system server
   Loaded: loaded (/usr/lib/systemd/system/glusterd.service; enabled; vendor preset: disabled)
   Active: active (running) since Fri -- :: CST; 6min ago
  Process:  ExecStart=/usr/sbin/glusterd -p /var/run/glusterd.pid --log-level $LOG_LEVEL $GLUSTERD_OPTIONS (code=exited, status=/SUCCESS)
 Main PID:  (glusterd)
    Tasks:
   CGroup: /system.slice/glusterd.service
           ├─ /usr/sbin/glusterd -p /var/run/glusterd.pid --log-level INFO
           ├─ /usr/sbin/glusterfs -s localhost --volfile-id gluster/glustershd -p /var/run/gluster/glustershd/glustershd.pid -l /var/lo...
           └─ /usr/sbin/glusterfsd -s gfs1.cluster.com --volfile-id gv0.gfs1.cluster.com.data-gluster-gv0 -p /var/run/gluster/vols/gv0/...

Feb  :: gfs1.cluster.com systemd[]: Starting GlusterFS, a clustered file-system server...
Feb  :: gfs1.cluster.com systemd[]: Started GlusterFS, a clustered file-system server.

• 配置防火墙

简单起见直接关闭了，以后补充开放具体网段

systemctl stop firewalld
systemctl disable firewalld

• 修改主机名以及/etc/hosts

每台机器执行，并修改/etc/hosts

hostnamectl set-hostname gfs1.cluster.com

• 添加存储

在每台glusterfs的server上加入一块存储盘，并进行初始化

fdisk /dev/sdb

mkfs.ext4 /dev/sdb1

在每个节点上运行以下命令挂载

mkdir -p /data/gluster
mount /dev/sdb1 /data/gluster
echo "/dev/sdb1 /data/gluster ext4 defaults 0 0" | tee --append /etc/fstab

3.配置Glusterfs

在节点1上运行

gluster peer probe gfs2.cluster.com

验证

[root@gfs1 mnt]# gluster peer status
Number of Peers: 

Hostname: gfs2.cluster.com
Uuid: 818cc628-85a7-4f5e-bd4e-34932c05de97
State: Peer in Cluster (Connected)

[root@gfs1 mnt]# gluster pool list
UUID                    Hostname            State
818cc628-85a7-4f5e-bd4e-34932c05de97    gfs2.cluster.com    Connected
dbcc01fc-3d2c-466f--57c46a9974be    localhost           Connected 

volume和brick的概念

3.1 复制卷

创建GFS卷gv0并配置复制模式

mkdir -p /data/gluster/gv0（在gfs1和gfs2上都建立brick)
gluster volume create gv0 replica  gfs1.cluster.local:/data/gluster/gv0 gfs2.cluster.local:/data/gluster/gv0

启动gv0卷

gluster volume start gv0
gluster volume info gv0

[root@gfs1 mnt]# gluster volume info gv0

Volume Name: gv0
Type: Replicate
Volume ID: 26d05ac6---ada4-5a423793fa20
Status: Started
Snapshot Count:
Number of Bricks:  x  =
Transport-type: tcp
Bricks:
Brick1: gfs1.cluster.com:/data/gluster/gv0
Brick2: gfs2.cluster.com:/data/gluster/gv0
Options Reconfigured:
performance.client-io-threads: off
nfs.disable: on
transport.address-family: inet

3.2 分布式卷(Distributed volume)

mkdir -p /data/gluster/brick

gluster volume create gv1 gfs1.cluster.com:/data/gluster/brick gfs2.cluster.com:/data/gluster/brick

gluster volume start gv1

[root@gfs1 mnt]# mkdir -p /data/gluster/brick
[root@gfs1 mnt]# gluster volume create gv1 gfs1.cluster.com:/data/gluster/brick gfs2.cluster.com:/data/gluster/brick
volume create: gv1: success: please start the volume to access data
[root@gfs1 mnt]# gluster volume start gv1
volume start: gv1: success
[root@gfs1 mnt]# gluster volume info gv1

Volume Name: gv1
Type: Distribute
Volume ID: 4782dd87-a411-44b3--70dfb072b5d0
Status: Started
Snapshot Count:
Number of Bricks:
Transport-type: tcp
Bricks:
Brick1: gfs1.cluster.com:/data/gluster/brick
Brick2: gfs2.cluster.com:/data/gluster/brick
Options Reconfigured:
transport.address-family: inet
nfs.disable: on

3.3 条带化卷(Stripe Volume)

mkdir -p /data/gluster/stripebrick

gluster volume create gv3 stripe  transport tcp gfs1.cluster.com:/data/gluster/stripebrick  gfs2.cluster.com:/data/gluster/stripebrick

gluster volume start gv3

[root@gfs1 mnt]# mkdir -p /data/gluster/stripebrick
[root@gfs1 mnt]# gluster volume create gv3 stripe  transport tcp gfs1.cluster.com:/data/gluster/stripebrick  gfs2.cluster.com:/data/gluster/stripebrick
volume create: gv3: success: please start the volume to access data
[root@gfs1 mnt]# gluster volume start gv3
volume start: gv3: success
[root@gfs1 mnt]# gluster volume info gv3

Volume Name: gv3
Type: Stripe
Volume ID: c25a10b8-a943-4c40-93be-088b972cbbaa
Status: Started
Snapshot Count:
Number of Bricks:  x  =
Transport-type: tcp
Bricks:
Brick1: gfs1.cluster.com:/data/gluster/stripebrick
Brick2: gfs2.cluster.com:/data/gluster/stripebrick
Options Reconfigured:
transport.address-family: inet
nfs.disable: on

3.4 分布式复制卷

更详细拓扑结构和管理说明参考官方文档，值得你阅读

https://access.redhat.com/documentation/en-us/red_hat_gluster_storage/3/html/administration_guide/

4.客户端配置

支持的客户端协议

yum install -y glusterfs-client
mkdir -p /mnt/glusterfs
mount -t glusterfs gfs1.cluster.com:/gv0 /mnt/glusterfs

验证挂载

[root@master ~]# df -hP /mnt/glusterfs
Filesystem             Size  Used Avail Use% Mounted on
gfs1.cluster.com:/gv0  .8G  136M  .2G   % /mnt/glusterfs

在node1和node2上也mount上glusterfs gv0,便于查看里面内容

[root@gfs1 ~]# mount -t glusterfs gfs2.cluster.com:/gv0 /mnt
[root@gfs2 ~]# mount -t glusterfs gfs1.cluster.com:/gv0 /mnt

然后基于客户端进行文件创建删除，同时将node1进行停机的高可用测试。