03. Zookeeper 安装-伪集群环境

落日映苍穹つ 2022-03-18 02:24 295阅读 0赞

> 所谓伪集群, 就是指在一台物理机上部署多个Zookeeper 节点, 组成Zookeeper集群. 这种方式也就仅限于学习研究中, 在生产环境中并不常用. 需要注意是的是Zookeeper 集群在扩展时, 需要对每个zk 节点进行修改, 所以初期, 最好确定zk集群节点数量. Zookeeper集群环境和单实例环境主要有两点差别:

*  配置文件中新增集群节点配置, 配置格式: server.序号=ip/主机名:同步数据通信端口号:选主通信端口号
 *  Zookeeper节点数据目录下, 新增myid 文件.

## 1. Zookeeper伪集群环境搭建 ##

笔者始终认为, 对于任何一个软件或服务而言, 一个好的目录规划是非常重要的.

<table> 
 <thead> 
  <tr> 
   <th align="left">服务器ip</th> 
   <th align="left">占用端口</th> 
   <th align="left">ZK 家目录</th> 
   <th align="left">ZK 数据目录</th> 
   <th align="left">数据操作日志目录</th> 
   <th align="left">ZK 日志目录</th> 
  </tr> 
 </thead> 
 <tbody> 
  <tr> 
   <td align="left">localhost</td> 
   <td align="left">2181, 2878, 3878</td> 
   <td align="left">/opt/app/zookeeper/2181/zookeeper-3.4.11</td> 
   <td align="left">/data/zookeeper/2181</td> 
   <td align="left">/data/logs/zookeeper/2181</td> 
   <td align="left"></td> 
  </tr> 
  <tr> 
   <td align="left">localhost</td> 
   <td align="left">2182, 2888, 3888</td> 
   <td align="left">/opt/app/zookeeper/2182/zookeeper-3.4.11</td> 
   <td align="left">/data/zookeeper/2182</td> 
   <td align="left">/data/logs/zookeeper/2182</td> 
   <td align="left"></td> 
  </tr> 
  <tr> 
   <td align="left">localhost</td> 
   <td align="left">2183, 2898, 3898</td> 
   <td align="left">/opt/app/zookeeper/2183/zookeeper-3.4.11</td> 
   <td align="left">/data/zookeeper/2183</td> 
   <td align="left">/data/logs/zookeeper/2183</td> 
   <td align="left"></td> 
  </tr> 
 </tbody> 
</table>

### 1.1 创建相关目录 ###

# 创建目录
    $ mkdir -p /opt/app/zookeeper/2181 /opt/app/zookeeper/2182 /opt/app/zookeeper/2183
    $ mkdir -p /data/zookeeper/2181 /data/zookeeper/2182 /data/zookeeper/2183
    $ mkdir -p /data/logs/zookeeper/2181 /data/logs/zookeeper/2182 /data/logs/zookeeper/2183

### 1.2 解压Zookeeper 压缩包 ###

每个安装目录下都需要解压一个Zookeeper安装包, 不能共用一个.

$ tar -zxf zookeeper-3.4.11.tar.gz -C /opt/app/zookeeper/2181
    $ tar -zxf zookeeper-3.4.11.tar.gz -C /opt/app/zookeeper/2182
    $ tar -zxf zookeeper-3.4.11.tar.gz -C /opt/app/zookeeper/2183

### 1.3 修改日志 ###

修改bin/zkEnv.sh脚本, 第一非注释行设置日志路径. 注意不同节点设置不同路径.

#设置zk 启动日志目录
    ZOO_LOG_DIR=/data/logs/zookeeper/2181
    
    # 如需指定jdk, 添加以下配置
    # JAVA_HOME=/opt/app/jdk/jdk1.8.0_131

### 1.4 创建配置文件 ###

解压后, 每个conf 目录下新增一个zoo.cfg 配置文件, 配置几乎一致, 只需将clientPort 和 dataDir 修改为对应的端口号即可.

# 时间单位, 毫秒
    tickTime=2000
    
    #集群中follwer节点与leader 节点初始化连接超时时间,单位为tickTime
    initLimit=10
    
    #集群中follwer节点与leader节点之前请求-响应超时时间,单位为tickTime
    syncLimit=5
    
    #数据存放目录: 记录当前的版本号和数据快照
    dataDir=/data/zookeeper/2181
    
    #客户端连接端口
    clientPort=2181
    
    #格式: server.序号=ip/主机名:同步数据通信端口号:选主通信端口号
    server.1=localhost:2878:3878
    server.2=localhost:2888:3888
    server.3=localhost:2898:3898

### 1.5 创建myid 文件 ###

myid 文件是Zookeeper 节点在集群中的标识

$ echo 1 > /data/zookeeper/2181/myid
    $ echo 2 > /data/zookeeper/2182/myid
    $ echo 3 > /data/zookeeper/2183/myid

## 2. 集群管理脚本 ##

对于伪集群而已, 服务都是安装在同一服务器上, 因此, 笔者写了几个管理脚本, 方便测试. 脚本存放目录为: /opt/app/zookeeper/cluster-bins

### 2.1 集群启动管理 ###

用于操作整个集群的起停.

#!/bin/bash
    #Name zk-cluster.sh
    #Desc 本机zk 伪集群管理
    #Auth zongf
    #Date 2019-02-18
    
    if [ "statusx" == $1x -o "startx" == $1x -o "stopx" == $1x -o "restartx" == $1x ] ; then
        echo -e "\n $1 2181:"
        /opt/app/zookeeper/2181/zookeeper-3.4.11/bin/zkServer.sh $1
        
        echo -e "\n $1 2182:"
        /opt/app/zookeeper/2182/zookeeper-3.4.11/bin/zkServer.sh $1
        
        echo -e "\n $1 2183:"
        /opt/app/zookeeper/2183/zookeeper-3.4.11/bin/zkServer.sh $1
        
    elif [ "jpsx" == $1x ] ; then
        jps -m | grep QuorumPeerMain | grep zookeeper
    else
        echo "Command demo: $0 start|stop|restart|status|jps"
    fi

### 2.2 节点管理脚本 ###

对集群中单个节点进行起停

#!/bin/bash
    #Name zk-server.sh
    #Desc 本机zk 节点管理
    #Auth zongf
    #Date 2019-02-18
    
    if [ "1x" == $1x -o "2x" == $1x -o "3x" == $1x ] ; then
       if [ "startx" == $2x -o "stopx" == $2x -o "restartx" == $2x -o "statusx" == $2x ] ; then 
          echo -e "zookeeper 218$1 $2:"
          /opt/app/zookeeper/218$1/zookeeper-3.4.11/bin/zkServer.sh $2 
       else
          echo "Command demo: $0 1|2|3 start|stop|restart|status"
       fi
    else 
       echo "Command demo: $0 1|2|3 start|stop|restart|status"
    fi

### 2.3 客户端连接脚本 ###

客户端连接某个Zookeeper 节点.

#!/bin/bash
    #Name zk-cli.sh
    #Desc zk客户端连接
    #Auth zongf
    #Date 2019-02-18
    
    if [ "1x" == $1x -o "2x" == $1x -o "3x" == $1x ] ; then
        echo -e "connect zookeeper $1:"
        /opt/app/zookeeper/218$1/zookeeper-3.4.11/bin/zkCli.sh -server 127.0.0.1:218$1
    else
        echo "Command demo: $0 1|2|3"
    fi

## 3. 测试 ##

### 3.1 启动集群 ###

$ ./zk-cluster.sh start
    
     start 2181:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2181/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Starting zookeeper ... STARTED
    
     start 2182:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2182/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Starting zookeeper ... STARTED
    
     start 2183:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2183/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Starting zookeeper ... STARTED

### 3.2 查看节点状态 ###

$ ./zk-cluster.sh status
    
     status 2181:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2181/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Mode: follower
    
     status 2182:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2182/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Mode: leader
    
     status 2183:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2183/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Mode: follower

### 3.2 停止Leader 节点 ###

停止leader 节点之后, Zookeeper集群会迅速选取新的leader 节点.

# 停止leader 节点
    $ ./zk-server.sh 2 stop
    zookeeper 2182 stop:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2182/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Stopping zookeeper ... STOPPED
    
    # 查看集群状态
    $ ./zk-cluster.sh status
    
     status 2181:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2181/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Mode: follower
    
     status 2182:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2182/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Error contacting service. It is probably not running.
    
     status 2183:
    ZooKeeper JMX enabled by default
    Using config: /opt/app/zookeeper/2183/zookeeper-3.4.11/bin/../conf/zoo.cfg
    Mode: leader

### 3.3 连接客户端 ###

$ ./zk-cli.sh 1
    connect zookeeper 1:
    Connecting to 127.0.0.1:2181