其他分享
首页 > 其他分享> > glusterfs部署及卷类型使用【转】

glusterfs部署及卷类型使用【转】

作者:互联网

glusterfs部署及卷类型使用

 

一、环境说明

hostnameIPdiskrole
gluster-node1 192.168.99.251 /dev/sdb、/dev/sdc、/dev/sdd server、client
gluster-node12 192.168.99.252 /dev/sdb、/dev/sdc、/dev/sdd server、client
gluster-node13 192.168.99.253 /dev/sdb、/dev/sdc、/dev/sdd server、client

二、GlusterFS安装

2.1 修改主机名

  $ hostnamectl set-hostname gluster-node1
  $ hostnamectl set-hostname gluster-node2
  $ hostnamectl set-hostname gluster-node3

2.2 添加host解析

  $ cat >> /etc/hosts <<EOF
  192.168.99.251 gluster-node1
  192.168.99.252 gluster-node2
  192.168.99.253 gluster-node3
  EOF
  $ scp /etc/hosts root@gluster-node2:/etc/hosts
  $ scp /etc/hosts root@gluster-node3:/etc/hosts

2.3 关闭SELinux、防火墙

  $ setenforce 0 && sed -i 's#SELINUX=enforcing#SELINUX=disabled#g' /etc/sysconfig/selinux
  $ systemctl stop firewalld && systemctl disable firewalld

2.4 glusterfs调优参数(可选)

  $ echo "vm.vfs_cache_pressure = 150" >> /etc/sysctl.conf
  $ echo "vm.swappiness = 0" >> /etc/sysctl.conf
  $ sysctl -p
  $ tuned-adm profile throughput-performance
  $ cat >> /var/lib/glusterd/groups/file-cache <<EOF
  performance.flush-behind=on
  performance.write-behind=on
  client.event-threads=8
  server.event-threads=4
  performance.client-io-threads=on
  io-thread-count=32
  network.tcp-window-size=1048576
  performance.write-behind-window-size=128MB
  server.outstanding-rpc-limit=128
  performance.parallel-readdir=on
  lookup-optimize=on
  performance.cache-size=4096MB
  performance.rda-cache-limit=100MB
  nfs.mem-factor=150
  performance.nfs.io-threads=on
  performance.nfs.stat-prefetch=on
  features.cache-invalidation=on
  features.cache-invalidation-timeout=600
  performance.stat-prefetch=on
  performance.cache-invalidation=on
  performance.md-cache-timeout=600
  network.inode-lru-limit=50000
  cluster.server-quorum-type=server
  cluster.quorum-type=auto
  cluster.favorite-child-policy=mtime
  network.ping-timeout=5
  EOF
  # 编辑该文件需使用以下命令关联到卷上,否则没什么意义
  $ gluster volume set gv1 group file-cache

2.5 安装epel源

  $ yum install epel-release -y

2.6 安装glusterfs源

  $ yum install centos-release-gluster -y

2.7 安装glusterfs

  $ yum install glusterfs-server -y

三、GlusterFS配置

  $ glusterfs -V # 查看glusterfs版本信息
  glusterfs 7.7
  $ systemctl start glusterd && systemctl enable glusterd
  # 启动gluster服务

在开始创建ClusterFS卷之前,需要创建一个称之为Trusted Storage的池,是一个可信的网络存储服务器,可以理解为集群。为卷提供bricks。

gluster peer probe<主机名或者IP地址>,建议是IP地址,或者内网用DNS做主机名解析,或者在/etc/hosts做好绑定。

3.1 添加GlusterFS对等体

  $ gluster peer probe gluster-node2
  peer probe: success.
  $ gluster peer probe gluster-node3
  peer probe: success.
  # 注意不要添加本地的服务器。这里在gluster-node1上操作,所以不需要添加gluster-node1

3.2 删除GlusterFS对等体

  $ gluster peer detach gluster-node2
  peer detach: success.

3.3 查看对等体状态

  $ gluster peer status
  Number of Peers: 2
  Hostname: gluster-node2
  Uuid: 5faada9a-1168-4a4d-bbb2-decf5b63990e
  State: Peer in Cluster (Connected)
  Hostname: gluster-node3
  Uuid: 0ba4f9f4-4cc5-4362-ac41-4941285cedbb
  State: Peer in Cluster (Connected)
   
  #在glutser-node2上查看
  $ gluster peer status
  Number of Peers: 2
  Hostname: gluster-node1
  Uuid: 6899d47f-1ce0-4c37-a6cb-fcd0defbb85f
  State: Peer in Cluster (Connected)
  Hostname: gluster-node3
  Uuid: 0ba4f9f4-4cc5-4362-ac41-4941285cedbb
  State: Peer in Cluster (Connected)

3.4 配置前准备工作

在各个节点上增加磁盘/dev/sdd,在gluster-node2和gluster-node3上都执行以下步骤:

  $ mkfs.xfs -f /dev/sdd
  meta-data=/dev/sdd isize=512 agcount=4, agsize=65536 blks
  = sectsz=512 attr=2, projid32bit=1
  = crc=1 finobt=0, sparse=0
  data = bsize=4096 blocks=262144, imaxpct=25
  = sunit=0 swidth=0 blks
  naming =version 2 bsize=4096 ascii-ci=0 ftype=1
  log =internal log bsize=4096 blocks=2560, version=2
  = sectsz=512 sunit=0 blks, lazy-count=1
  realtime =none extsz=4096 blocks=0, rtextents=0
  $ mkdir -p /storage/brick1
  $ mount /dev/sdd /storage/brick1
  $ df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda2 20G 1.5G 19G 8% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdd 1014M 33M 982M 4% /storage/brick1
  $ echo "/dev/sdd /storage/brick1 xfs default 0 0" >> /etc/fstab

3.5创建volume及其他操作

创建各种卷语法格式:

  volume create <NEW-VOLNAME> [stripe <COUNT>] [replica <COUNT> [arbiter <COUNT>]] [disperse [<COUNT>]] [disperse-data <COUNT>] [redundancy <COUNT>] [transport <tcp|rdma|tcp,rdma>] <NEW-BRICK>?<vg_name>... [force]

下面逐一演示7种卷的使用:

分布式卷架构示意图:

20200831214644

默认模式,既DHT, 也叫 分布卷: 将文件已hash算法随机分布到 一台服务器节点中存储。如:

gluster volume create test-volume server1:/exp1 server2:/exp2

  (1)查看创建命令的帮助
  [root@gluster-node1 ~]# gluster volume create -h
   
  Usage:
  volume create <NEW-VOLNAME> [stripe <COUNT>] [replica <COUNT> [arbiter <COUNT>]] [disperse [<COUNT>]] [disperse-data <COUNT>] [redundancy <COUNT>] [transport <tcp|rdma|tcp,rdma>] <NEW-BRICK>?<vg_name>... [force]
   
  (2)创建卷gv1
  [root@gluster-node1 ~]# gluster volume create gv1 gluster-node1:/storage/brick1 gluster-node2:/storage/brick1 force  #创建卷gv1,并将gluster-node1和gluster-node2上的/dev/sdb合并空间
  volume create: gv1: success: please start the volume to access data
  [root@gluster-node1 ~]# gluster volume start gv1  #启动gv1卷
  volume start: gv1: success
  [root@gluster-node3 ~]# gluster volume info #在glutser-node3上查看volume信息
   
  Volume Name: gv1
  Type: Distribute
  Volume ID: 5773c33b-2371-419a-a753-de89063f4e23
  Status: Started
  Snapshot Count: 0
  Number of Bricks: 2
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick1
  Brick2: gluster-node2:/storage/brick1
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: on
   
  (3)挂载数据卷
  [root@gluster-node1 ~]# mkdir /gv1  #创建挂载目录
  [root@gluster-node1 ~]# mount -t glusterfs 127.0.0.1:/gv1 /gv1  #挂载gv1卷
  [root@gluster-node1 ~]# df -h  #查看挂载信息
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda2 20G 1.5G 19G 8% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  127.0.0.1:/gv1 2.0G 85M 1.9G 5% /gv1  #可以看到挂载在gv1目录下的gv1卷磁盘大小为2个G,是gluster-node1和gluster-node2合并一起的磁盘大小
   
  [root@gluster-node2 ~]# mkdir /gv1  #在gluster-node2上同样创建挂载目录
  [root@gluster-node2 ~]# mount -t glusterfs 127.0.0.1:/gv1 /gv1  #在gluster-node2上挂载gv1卷
  [root@gluster-node2 ~]# df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda2 20G 1.3G 19G 7% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  127.0.0.1:/gv1 2.0G 85M 1.9G 5% /gv1
   
  [root@gluster-node3 ~]# mkdir /gv1
  [root@gluster-node3 ~]# mount -t glusterfs 127.0.0.1:/gv1 /gv1
  [root@gluster-node3 ~]# df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda3 28G 1.6G 27G 6% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  127.0.0.1:/gv1 2.0G 85M 1.9G 5% /gv1
   
  (4)测试在gluster-node1上写入数据,看挂载在gluster-node2和gluster-node3是否有同样的数据
  [root@gluster-node1 gv1]# touch 1.txt 2.txt 3.txt
  [root@gluster-node1 gv1]# ll
  total 0
  -rw-r--r--. 1 root root 0 Aug 5 06:33 1.txt
  -rw-r--r--. 1 root root 0 Aug 5 06:33 2.txt
  -rw-r--r--. 1 root root 0 Aug 5 06:33 3.txt
   
  [root@gluster-node2 ~]# ll /gv1/
  total 0
  -rw-r--r-- 1 root root 0 Aug 5 2018 1.txt
  -rw-r--r-- 1 root root 0 Aug 5 2018 2.txt
  -rw-r--r-- 1 root root 0 Aug 5 2018 3.txt
   
  [root@gluster-node3 ~]# ll /gv1/
  total 0
  -rw-r--r-- 1 root root 0 Aug 4 18:33 1.txt
  -rw-r--r-- 1 root root 0 Aug 4 18:33 2.txt
  -rw-r--r-- 1 root root 0 Aug 4 18:33 3.txt

复制卷的架构示意图:

20200831214725

复制模式,既AFR, 创建volume 时带 replica x 数量: 将文件复制到 replica x 个节点中。如:

*gluster volume create test-volume replica 2 transport tcp server1:/exp1 server2:/exp2*

  (1)将gluster-node1和gluster-node2上的/dev/sdc格式化并挂载到/storage/brick2
  [root@gluster-node1 ~]# mkfs.xfs -f /dev/sdc
  meta-data=/dev/sdc isize=512 agcount=4, agsize=65536 blks
  = sectsz=512 attr=2, projid32bit=1
  = crc=1 finobt=0, sparse=0
  data = bsize=4096 blocks=262144, imaxpct=25
  = sunit=0 swidth=0 blks
  naming =version 2 bsize=4096 ascii-ci=0 ftype=1
  log =internal log bsize=4096 blocks=2560, version=2
  = sectsz=512 sunit=0 blks, lazy-count=1
  realtime =none extsz=4096 blocks=0, rtextents=0
  [root@gluster-node1 ~]#
  [root@gluster-node1 ~]# mkdir -p /storage/brick2
  [root@gluster-node1 ~]# mount /dev/sdc /storage/brick2
  [root@gluster-node1 ~]# df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda2 20G 1.5G 19G 8% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  127.0.0.1:/gv1 2.0G 85M 1.9G 5% /gv1
  /dev/sdc 1014M 33M 982M 4% /storage/brick2
  [root@gluster-node1 ~]# echo "/dev/sdc /storage/brick2 xfs defaults 0 0" >> /etc/fstab
   
  (2)创建复制卷
  [root@gluster-node1 ~]# gluster volume create gv2 replica 2 gluster-node1:/storage/brick2 gluster-node2:/storage/brick2 force
  volume create: gv2: success: please start the volume to access data
  [root@gluster-node1 ~]# gluster volume start gv2 #启动volume gv2
  volume start: gv2: success
  [root@gluster-node1 ~]# gluster volume info
   
  Volume Name: gv1
  Type: Distribute #gv1为分布卷
  Volume ID: 5773c33b-2371-419a-a753-de89063f4e23
  Status: Started
  Snapshot Count: 0
  Number of Bricks: 2
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick1
  Brick2: gluster-node2:/storage/brick1
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: off
   
  Volume Name: gv2
  Type: Replicate #可以看到gv2为复制卷
  Volume ID: aee56c69-2c33-455d-8da0-aa4591a25065
  Status: Started
  Snapshot Count: 0
  Number of Bricks: 1 x 2 = 2
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick2
  Brick2: gluster-node2:/storage/brick2
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: on
  performance.client-io-threads: off
   
  (3)创建挂载目录/gv2并将gv2卷挂载
  [root@gluster-node1 ~]# mkdir /gv2
  [root@gluster-node1 ~]# mount -t glusterfs 127.0.0.1:/gv2 /gv2
  [root@gluster-node1 ~]# df -h #查看挂载信息,可以看到gv2的磁盘空间大小为1G,为2块磁盘空间的一般,这就是复制卷,可用空间只有一半
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda2 20G 1.5G 19G 8% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.9M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  127.0.0.1:/gv1 2.0G 85M 1.9G 5% /gv1
  /dev/sdc 1014M 33M 982M 4% /storage/brick2
  127.0.0.1:/gv2 1014M 43M 972M 5% /gv2
   
  (4)在挂载目录/gv2下创建文件测试
  [root@gluster-node1 ~]# cd /gv2/
  [root@gluster-node1 gv2]# touch file1 file2 file3
  [root@gluster-node1 gv2]# ls
  file1 file2 file3
  [root@gluster-node1 gv2]# ls /storage/brick2/
  file1 file2 file3
  [root@gluster-node2 ~]# ls /storage/brick2/
  file1 file2 file3
   
  可以看到创建的文件在gluster-node1和gluster-node2上磁盘挂载目录/storage/brick2目录中的文件是一样的,那么在gluster-node3上挂载后再创建测试
   
  [root@gluster-node3 ~]# mkdir /gv2
  [root@gluster-node3 ~]# mount -t glusterfs 127.0.0.1:/gv2 /gv2
  [root@gluster-node3 ~]# df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda3 28G 1.6G 27G 6% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  127.0.0.1:/gv2 1014M 43M 972M 5% /gv2
  [root@gluster-node3 ~]# cd /gv2/
  [root@gluster-node3 gv2]# touch file4
   
  [root@gluster-node2 ~]# ls /storage/brick2/
  file1 file2 file3 file4
   
  [root@gluster-node1 gv2]# ls
  file1 file2 file3 file4
  [root@gluster-node1 gv2]# ls /storage/brick2/
  file1 file2 file3 file4
  也可以看到gluster-node1和gluster-node2磁盘挂载目录/storage/brick2的文件是一样的

条带卷架构示意图:

20200831214757

条带模式,既Striped, 创建volume 时带 stripe x 数量: 将文件切割成数据块,分别存储到 stripe x 个节点中 ( 类似raid 0 )。如:

*gluster volume create test-volume stripe 2 transport tcp server1:/exp1 server2:/exp2*

  (1)格式化gluster-node1和gluster-node2上的磁盘/dev/sdd并挂载到/storage/brick3
  [root@gluster-node1 ~]# mkfs.xfs -f /dev/sdd
  meta-data=/dev/sdd isize=512 agcount=4, agsize=65536 blks
  = sectsz=512 attr=2, projid32bit=1
  = crc=1 finobt=0, sparse=0
  data = bsize=4096 blocks=262144, imaxpct=25
  = sunit=0 swidth=0 blks
  naming =version 2 bsize=4096 ascii-ci=0 ftype=1
  log =internal log bsize=4096 blocks=2560, version=2
  = sectsz=512 sunit=0 blks, lazy-count=1
  realtime =none extsz=4096 blocks=0, rtextents=0
  [root@gluster-node1 ~]# mkdir -p /storage/brick3
  [root@gluster-node1 ~]# mount /dev/sdd /storage/brick3
  [root@gluster-node1 ~]# echo "/dev/sdd /storage/brick3 xfs defaults 0 0" >> /etc/fstab
   
  (2)创建条带卷
  [root@gluster-node1 ~]# gluster volume create gv3 stripe 2 gluster-node1:/storage/brick3 gluster-node2:/storage/brick3 force
  volume create: gv3: success: please start the volume to access data
  [root@gluster-node1 ~]# gluster volume info
   
  Volume Name: gv1
  Type: Distribute
  Volume ID: 5773c33b-2371-419a-a753-de89063f4e23
  Status: Started
  Snapshot Count: 0
  Number of Bricks: 2
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick1
  Brick2: gluster-node2:/storage/brick1
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: off
   
  Volume Name: gv2
  Type: Replicate
  Volume ID: aee56c69-2c33-455d-8da0-aa4591a25065
  Status: Started
  Snapshot Count: 0
  Number of Bricks: 1 x 2 = 2
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick2
  Brick2: gluster-node2:/storage/brick2
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: on
  performance.client-io-threads: off
   
  Volume Name: gv3
  Type: Stripe
  Volume ID: 4d3c3a88-0c3f-4b6f-a2ef-25ab599cb015
  Status: Created
  Snapshot Count: 0
  Number of Bricks: 1 x 2 = 2
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick3
  Brick2: gluster-node2:/storage/brick3
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: on
  [root@gluster-node1 ~]# gluster volume start gv3
  volume start: gv3: success
   
  (3)创建挂载目录,并挂载
  [root@gluster-node1 ~]# mkdir /gv3
  [root@gluster-node1 ~]# mount -t glusterfs 127.0.0.1:/gv3 /gv3
  [root@gluster-node1 ~]# df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda2 20G 1.5G 19G 8% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  127.0.0.1:/gv1 2.0G 85M 1.9G 5% /gv1
  /dev/sdc 1014M 33M 982M 4% /storage/brick2
  127.0.0.1:/gv2 1014M 43M 972M 5% /gv2
  /dev/sdd 1014M 33M 982M 4% /storage/brick3
  127.0.0.1:/gv3 2.0G 85M 1.9G 5% /gv3
   
  (4)创建文件测试
  [root@gluster-node1 ~]# cd /gv3/
  [root@gluster-node1 gv3]# dd if=/dev/zero bs=1024 count=10000 of=/gv3/10M.file
  10000+0 records in
  10000+0 records out
  10240000 bytes (10 MB) copied, 2.46812 s, 4.1 MB/s
  [root@gluster-node1 gv3]# ll
  total 10000
  -rw-r--r--. 1 root root 10240000 Aug 6 04:14 10M.file
  [root@gluster-node1 gv3]# ll -h
  total 9.8M
  -rw-r--r--. 1 root root 9.8M Aug 6 04:14 10M.file
  [root@gluster-node1 gv3]# ll /storage/brick3/ #在gluster-node1实际存储目录中文件大小只有4.9M
  total 5008
  -rw-r--r--. 2 root root 5128192 Aug 6 04:14 10M.file
  [root@gluster-node1 gv3]# ll -h /storage/brick3/ #在gluster-node1实际存储目录中文件大小只有4.9M
  total 4.9M
  -rw-r--r--. 2 root root 4.9M Aug 6 04:14 10M.file

分布式条带卷架构示意图:

20200831214825

分布式条带模式(组合型),最少需要4台服务器或者4块磁盘才能创建。 创建volume 时 stripe 2 server = 4 个节点: 是DHT 与 Striped 的组合型。如:

gluster volume create test-volume stripe 2 transport tcp server1:/exp1 server2:/exp2 server3:/exp3 server4:/exp4

  (1)停止前面的所有卷
  [root@gluster-node1 ~]# gluster volume stop gv1
  Stopping volume will make its data inaccessible. Do you want to continue? (y/n) y
  volume stop: gv1: success
  [root@gluster-node1 ~]# gluster volume stop gv2
  Stopping volume will make its data inaccessible. Do you want to continue? (y/n) y
  volume stop: gv2: success
  [root@gluster-node1 ~]# gluster volume stop gv3
  Stopping volume will make its data inaccessible. Do you want to continue? (y/n) y
  volume stop: gv3: success
   
  (2)删除前面的所有卷
  [root@gluster-node1 ~]# gluster volume delete gv1
  Deleting volume will erase all information about the volume. Do you want to continue? (y/n) y
  volume delete: gv1: success
   
  [root@gluster-node1 ~]# gluster volume delete gv2
  Deleting volume will erase all information about the volume. Do you want to continue? (y/n) y
  volume delete: gv2: success
   
  [root@gluster-node1 ~]# gluster volume delete gv3
  Deleting volume will erase all information about the volume. Do you want to continue? (y/n) y
  volume delete: gv3: success
   
  (3)创建分布式条带卷
  [root@gluster-node1 ~]# gluster volume create gv1 stripe 2 gluster-node1:/storage/brick1 gluster-node1:/storage/brick2 gluster-node2:/storage/brick1 gluster-node2:/storage/brick2 force
  volume create: gv1: success: please start the volume to access data
  [root@gluster-node1 ~]# gluster volume info
   
  Volume Name: gv1
  Type: Distributed-Stripe #类型为分布式条带卷
  Volume ID: 4e6a8ddd-8403-4eb7-bd28-9365164755e3
  Status: Created
  Snapshot Count: 0
  Number of Bricks: 2 x 2 = 4
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick1
  Brick2: gluster-node1:/storage/brick2
  Brick3: gluster-node2:/storage/brick1
  Brick4: gluster-node2:/storage/brick2
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: on
   
  (4)启动gv1卷,并挂载到/gv1
  [root@gluster-node1 ~]# gluster volume start gv1
  volume start: gv1: success
  [root@gluster-node1 ~]# mount -t glusterfs 127.0.0.1:/gv1 /gv1
  [root@gluster-node1 ~]# df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda2 20G 1.5G 19G 8% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  /dev/sdc 1014M 33M 982M 4% /storage/brick2
  /dev/sdd 1014M 38M 977M 4% /storage/brick3
  127.0.0.1:/gv1 4.0G 170M 3.8G 5% /gv1
   
  (5)测试创建文件,并查看实际存储信息
  [root@gluster-node1 ~]# cd /gv1/
  [root@gluster-node1 gv1]# dd if=/dev/zero bs=1024 count=10000 of=/gv1/10M.file
  10000+0 records in
  10000+0 records out
  10240000 bytes (10 MB) copied, 1.94059 s, 5.3 MB/s
   
  [root@gluster-node1 gv1]# ll -h
  total 9.8M
  -rw-r--r--. 1 root root 9.8M Aug 6 06:58 10M.file
  [root@gluster-node1 gv1]# ll -h /storage/brick1/
  total 4.9M
  -rw-r--r--. 2 root root 4.9M Aug 6 06:58 10M.file
  [root@gluster-node1 gv1]# ll -h /storage/brick2/
  total 4.9M
  -rw-r--r--. 2 root root 4.9M Aug 6 06:58 10M.file

分布式复制卷架构示意图:

20200831214852

分布式复制模式(组合型), 最少需要4台服务器或者2台服务器4块磁盘才能创建。 创建volume 时 replica 2 server = 4 个节点:是DHT 与 AFR 的组合型。
gluster volume create test-volume replica 2 transport tcp server1:/exp1 server2:/exp2 server3:/exp3 server4:/exp4

  (1)卸载前面的分布式条带卷并删除
  [root@gluster-node1 ~]# umount /gv1/
  [root@gluster-node1 ~]# gluster volume stop gv1
  Stopping volume will make its data inaccessible. Do you want to continue? (y/n) y
  volume stop: gv1: success
  [root@gluster-node1 ~]# gluster volume delete gv1
  Deleting volume will erase all information about the volume. Do you want to continue? (y/n) y
  volume delete: gv1: success
   
  (2)创建分布式复制卷
  [root@gluster-node1 ~]# gluster volume create gv1 replica 2 gluster-node1:/storage/brick1/ gluster-node1:/storage/brick2 gluster-node2:/storage/brick1/ gluster-node2:/storage/brick2 force
  volume create: gv1: success: please start the volume to access data
  [root@gluster-node1 ~]# gluster volume info
   
  Volume Name: gv1
  Type: Distributed-Replicate
  Volume ID: 86f110cb-5075-493e-b49a-d7458b3def79
  Status: Created
  Snapshot Count: 0
  Number of Bricks: 2 x 2 = 4
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick1
  Brick2: gluster-node1:/storage/brick2
  Brick3: gluster-node2:/storage/brick1
  Brick4: gluster-node2:/storage/brick2
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: on
  performance.client-io-threads: off
  [root@gluster-node1 ~]# gluster volume start gv1
  volume start: gv1: success
   
  (3)挂载分布式复制卷/gv1
  [root@gluster-node3 ~]# mount -t glusterfs 127.0.0.1:/gv1 /gv1
  [root@gluster-node3 ~]# df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda3 28G 1.7G 27G 6% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  127.0.0.1:/gv1 2.0G 120M 1.9G 6% /gv1
   
  (4)创建文件测试
  [root@gluster-node3 gv1]# touch file1 file2 file3
  [root@gluster-node3 gv1]# ls
  file1 file2 file3
   
  [root@gluster-node1 ~]# ll /storage/brick1/
  total 0
  -rw-r--r--. 2 root root 0 Aug 9 2018 file3
  [root@gluster-node1 ~]# ll /storage/brick2/
  total 0
  -rw-r--r--. 2 root root 0 Aug 9 2018 file3
  [root@gluster-node2 ~]# ll /storage/brick1/
  total 0
  -rw-r--r-- 2 root root 0 Aug 9 2018 file1
  -rw-r--r-- 2 root root 0 Aug 9 2018 file2
  [root@gluster-node2 ~]# ll /storage/brick2/
  total 0
  -rw-r--r-- 2 root root 0 Aug 9 2018 file1
  -rw-r--r-- 2 root root 0 Aug 9 2018 file2
   
  在gluster-node3上挂载并创建文件,查看在gluster-node1和gluster-node2的实际存储位置/storage/brick1、/storage/brick2可以看到在gluster-node1上存储了file1文件并且在其存储位置brick1和brick2都有该文件,实现了复制功能。同理在gluster-node2上也可以看到file2、file3的存储情况。

条带+复制卷*架构示意图:*

20200831214932

条带复制卷模式(组合型), 最少需要4台服务器才能创建。 创建volume 时 stripe 2 replica 2 server = 4 个节点: 是 Striped 与 AFR 的组合型。
gluster volume create test-volume stripe 2 replica 2 transport tcp server1:/exp1 server2:/exp2 server3:/exp3 server4:/exp4

①卸载前面的分布式复制卷并删除

  [root@gluster-node3 ~]# umount /gv1
  [root@gluster-node1 ~]# gluster volume stop gv1
  Stopping volume will make its data inaccessible. Do you want to continue? (y/n) y
  volume stop: gv1: success
  [root@gluster-node1 ~]# gluster volume delete gv1
  Deleting volume will erase all information about the volume. Do you want to continue? (y/n) y
  volume delete: gv1: success

②创建条带复制卷并启动

  root@gluster-node1 ~]# gluster volume create gv1 stripe 2 replica 2 gluster-node1:/storage/brick1/ gluster-node1:/storage/brick2 gluster-node2:/storage/brick1/ gluster-node2:/storage/brick2 force
  volume create: gv1: success: please start the volume to access data
   
  [root@gluster-node1 ~]# gluster volume info
   
  Volume Name: gv1
  Type: Striped-Replicate
  Volume ID: b65aa061-044d-4a54-ac16-725077c2a0b8
  Status: Created
  Snapshot Count: 0
  Number of Bricks: 1 x 2 x 2 = 4
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick1
  Brick2: gluster-node1:/storage/brick2
  Brick3: gluster-node2:/storage/brick1
  Brick4: gluster-node2:/storage/brick2
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: on
  performance.client-io-threads: off
   
  [root@gluster-node1 ~]# gluster volume start gv1
  volume start: gv1: success

③挂载条带复制卷

  [root@gluster-node3 ~]# mount -t glusterfs 127.0.0.1:/gv1 /gv1
  [root@gluster-node3 ~]# df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda3 28G 1.7G 27G 6% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  127.0.0.1:/gv1 2.0G 85M 1.9G 5% /gv1

④touch创建文件测试

  [root@gluster-node3 ~]# cd /gv1/
  [root@gluster-node3 gv1]# touch file1 file2 file3
  [root@gluster-node3 gv1]# ll
  total 0
  -rw-r--r-- 1 root root 0 Aug 8 23:25 file1
  -rw-r--r-- 1 root root 0 Aug 8 23:25 file2
  -rw-r--r-- 1 root root 0 Aug 8 23:25 file3
   
   
  [root@gluster-node1 ~]# ll /storage/brick1/
  total 0
  -rw-r--r--. 2 root root 0 Aug 9 2018 file1
  -rw-r--r--. 2 root root 0 Aug 9 2018 file2
  -rw-r--r--. 2 root root 0 Aug 9 2018 file3
  [root@gluster-node1 ~]# ll /storage/brick2/
  total 0
  -rw-r--r--. 2 root root 0 Aug 9 2018 file1
  -rw-r--r--. 2 root root 0 Aug 9 2018 file2
  -rw-r--r--. 2 root root 0 Aug 9 2018 file3
   
  [root@gluster-node2 ~]# ll /storage/brick1/
  total 0
  -rw-r--r-- 2 root root 0 Aug 9 2018 file1
  -rw-r--r-- 2 root root 0 Aug 9 2018 file2
  -rw-r--r-- 2 root root 0 Aug 9 2018 file3
  [root@gluster-node2 ~]# ll /storage/brick2/
  total 0
  -rw-r--r-- 2 root root 0 Aug 9 2018 file1
  -rw-r--r-- 2 root root 0 Aug 9 2018 file2
  -rw-r--r-- 2 root root 0 Aug 9 2018 file3

从touch创建文件可以看到创建的文件,分别在gluster-node1和gluster-node2上的实际存储位置/storage/brick1、/storage/brick2上都存在这些文件,实现了复制功能

⑤dd创建文件测试

  [root@gluster-node3 gv1]# dd if=/dev/zero bs=1024 count=20000 of=/gv1/20M.file
  20000+0 records in
  20000+0 records out
  20480000 bytes (20 MB) copied, 1.08174 s, 18.9 MB/s
   
  [root@gluster-node1 ~]# ll /storage/brick1/ -h
  total 9.8M
  -rw-r--r--. 2 root root 9.8M Aug 7 00:50 20M.file
  -rw-r--r--. 2 root root 0 Aug 9 2018 file1
  -rw-r--r--. 2 root root 0 Aug 9 2018 file2
  -rw-r--r--. 2 root root 0 Aug 9 2018 file3
  [root@gluster-node1 ~]# ll /storage/brick2/ -h
  total 9.8M
  -rw-r--r--. 2 root root 9.8M Aug 7 00:50 20M.file
  -rw-r--r--. 2 root root 0 Aug 9 2018 file1
  -rw-r--r--. 2 root root 0 Aug 9 2018 file2
  -rw-r--r--. 2 root root 0 Aug 9 2018 file3
   
  [root@gluster-node2 ~]# ll /storage/brick1/ -h
  total 9.8M
  -rw-r--r-- 2 root root 9.8M Aug 6 19:24 20M.file
  -rw-r--r-- 2 root root 0 Aug 9 2018 file1
  -rw-r--r-- 2 root root 0 Aug 9 2018 file2
  -rw-r--r-- 2 root root 0 Aug 9 2018 file3
  [root@gluster-node2 ~]# ll /storage/brick2/ -h
  total 9.8M
  -rw-r--r-- 2 root root 9.8M Aug 6 19:24 20M.file
  -rw-r--r-- 2 root root 0 Aug 9 2018 file1
  -rw-r--r-- 2 root root 0 Aug 9 2018 file2
  -rw-r--r-- 2 root root 0 Aug 9 2018 file3

从dd创建的20M文件,在gluster-node1上可以看到20M的文件被切割成10M一份进行存储,并且在gluster-node2上也有这样的一份数据,实现了条带功能,也实现了复制功能。

*分布式+条带+复制卷架构示意图:*

20200831215004

三种模式混合, 至少需要8台 服务器或者2台服务器8块磁盘才能创建。 stripe 2 replica 2 , 每4个节点 组成一个 组。
gluster volume create test-volume stripe 2 replica 2 transport tcp server1:/exp1 server2:/exp2 server3:/exp3 server4:/exp4 server5:/exp5 server6:/exp6 server7:/exp7 server8:/exp8

①卸载前面的条带复制卷并删除

  [root@gluster-node1 ~]# umount /gv1
  [root@gluster-node1 ~]# gluster volume stop gv1
  Stopping volume will make its data inaccessible. Do you want to continue? (y/n) y
  volume stop: gv1: success
  [root@gluster-node1 ~]# gluster volume delete gv1
  Deleting volume will erase all information about the volume. Do you want to continue? (y/n) y
  volume delete: gv1: success

②创建分布式条带复制卷

  [root@gluster-node1 ~]# gluster volume create gv1 stripe 2 replica 2 gluster-node1:/storage/brick1 gluster-node1:/storage/brick2 gluster-node1:/storage/brick3 gluster-node2:/storage/brick1 gluster-node2:/storage/brick2 gluster-node2:/storage/brick3 gluster-node3:/storage/brick1 gluster-node3:/storage/brick2 force
  volume create: gv1: success: please start the volume to access data
   
   
  [root@gluster-node1 ~]# gluster volume info
   
  Volume Name: gv1
  Type: Distributed-Striped-Replicate
  Volume ID: fc16d0f2-52ec-4932-a21e-4226afc79ddf
  Status: Created
  Snapshot Count: 0
  Number of Bricks: 2 x 2 x 2 = 8
  Transport-type: tcp
  Bricks:
  Brick1: gluster-node1:/storage/brick1
  Brick2: gluster-node1:/storage/brick2
  Brick3: gluster-node1:/storage/brick3
  Brick4: gluster-node2:/storage/brick1
  Brick5: gluster-node2:/storage/brick2
  Brick6: gluster-node2:/storage/brick3
  Brick7: gluster-node3:/storage/brick1
  Brick8: gluster-node3:/storage/brick2
  Options Reconfigured:
  transport.address-family: inet
  nfs.disable: on
  performance.client-io-threads: off

③启动条带复制卷并挂载

  [root@gluster-node1 ~]# gluster volume start gv1
  volume start: gv1: success
  [root@gluster-node1 ~]# mount -t glusterfs 127.0.0.1:/gv1 /gv1
  [root@gluster-node1 ~]# df -h
  Filesystem Size Used Avail Use% Mounted on
  /dev/sda2 20G 1.5G 19G 8% /
  devtmpfs 483M 0 483M 0% /dev
  tmpfs 493M 0 493M 0% /dev/shm
  tmpfs 493M 6.8M 486M 2% /run
  tmpfs 493M 0 493M 0% /sys/fs/cgroup
  /dev/sda1 197M 97M 100M 50% /boot
  tmpfs 99M 0 99M 0% /run/user/0
  /dev/sdd 1014M 38M 977M 4% /storage/brick3
  /dev/sdb 1014M 33M 982M 4% /storage/brick1
  /dev/sdc 1014M 43M 972M 5% /storage/brick2
  127.0.0.1:/gv1 4.0G 200M 3.8G 5% /gv1
   
  观察存储容量,一共有4个G,原来的8块磁盘一共是8G大小。使用了分布式条带复制卷,存储空间会减半

④dd创建文件查看

  [root@gluster-node1 ~]# cd /gv1/
  [root@gluster-node1 gv1]# dd if=/dev/zero bs=1024 count=100000 of=/gv1/100M.file
  100000+0 records in
  100000+0 records out
  102400000 bytes (102 MB) copied, 29.2668 s, 3.5 MB/s
   
  [root@gluster-node1 gv1]# ll -h
  total 98M
  -rw-r--r--. 1 root root 98M Aug 7 04:30 100M.file
  [root@gluster-node1 gv1]# ll -h /storage/brick1/
  total 64M
  -rw-r--r--. 2 root root 49M Aug 7 04:30 100M.file
  [root@gluster-node1 gv1]# ll -h /storage/brick2/
  total 64M
  -rw-r--r--. 2 root root 49M Aug 7 04:30 100M.file
  [root@gluster-node1 gv1]# ll -h /storage/brick3/
  total 64M
  -rw-r--r--. 2 root root 49M Aug 7 04:30 100M.file
  [root@gluster-node2 ~]# ll /storage/brick1/ -h
  total 49M
  -rw-r--r-- 2 root root 49M Aug 6 23:04 100M.file

从创建的文件上看,创建了100M的文件,会进行分割49M文件作为一份数据进行存储,并通过hash算法随即存储到8块磁盘中的其中2块磁盘,并在另外2块磁盘中作了复制,以保证数据的备份。

本次的部署实验,使用了3台主机,8块磁盘,并不是生产使用方案,而是机子抗不住。在实际生产中,可以独立使用多台服务器进行磁盘整合,实现分布式文件系统。

glusterfs部署及卷类型使用 - 吕振江 - 博客园
https://www.cnblogs.com/lvzhenjiang/p/14199478.html

标签:部署,及卷,storage,gluster,volume,glusterfs,node1,gv1,root
来源: https://www.cnblogs.com/paul8339/p/15867873.html