linux中zookeeper-3.4.10安装，集群

一，安装

第一步：zookeeper安装

    1   下载   wget http://mirror.bit.edu.cn/apache/zookeeper/zookeeper-3.4.10/zookeeper-3.4.10.tar.gz

    2   解压 zookeeper-3.4.10.tar.gz

       进入  zookeeper-3.4.10/conf目录，把zoo_sample.cfg 配置文件改名为 zoo.cfg 

       因为zookeeper启动默认加载的配置文件为zoo.cfg

    3. 启动服务，与客户端

      进入zookeeper-3.4.10/bin 目录    

     [root@localhost bin]# sh zkServer.sh start   //启动

     [root@localhost bin]# sh zkServer.sh stop   //关闭

     [root@localhost bin]# sh zkCli.sh    //连接服务端

第二步，集群相关配置

 安装好2n+1台虚拟机. 如 分别在三台虚拟机上安装好，如192.168.1.128，192.168.1.129，192.168.1.130

     1.在/conf/zoo.cfg 添加，三台虚拟机上保持一致

           server.1=192.168.1.128:2888:3888
           server.2=192.168.1.129:2888:3888
           server.3=192.168.1.130:2888:3888

      2.zoo.cfg配置默认的数据存储在 dataDir=/tmp/zookeeper

        在/tmp/zookeeper 目录中新建文件 myid

        [root@localhost conf]# vim /tmp/zookeeper/myid

        192.168.1.128中 myid的内容 为 1， 192.168.1.129中 myid的内容 为 2， 192.168.1.130中 myid的内容 为 3

       与zoo.cfg 中保持一致

 

       3.重新启动三台虚拟机的zookeeper.

       [root@localhost bin]# tail -fn 100 zookeeper.out    //查看日记

        如果2888，3888两个端口没有打开的话会报错。

       [root@localhost bin]# systemctl stop firewalld   //  测试的话直接 关掉防火墙 

 

      4.查看主从，首次启动的时候，一般是先开启的为主，后开的为从.后面如果leader挂掉了，则通过leader选举机制重新选取

          [root@localhost bin]# sh zkServer.sh status
          ZooKeeper JMX enabled by default
          Using config: /data/program/zookeeper-3.4.10/bin/../conf/zoo.cfg
           Mode: leader

          [root@localhost bin]# sh zkServer.sh status
          ZooKeeper JMX enabled by default
          Using config: /data/program/zookeeper-3.4.10/bin/../conf/zoo.cfg
          Mode: follower
 

二，zookeeper的基本操作   

随便打开客户端 192.168.1.128，创建一个key，在其它几台集群机器上可以看到添加了相应的key

   [root@localhost bin]# sh zkCli.sh  

  [zk: localhost:2181(CONNECTED) 4] create /test1 a
  [zk: localhost:2181(CONNECTED) 5] ls /
  [zookeeper, test1]

 

1.创建节点  create [-s] [-e] path data acl

//同一节点下不能创建重复的节点
[zk: localhost:2181(CONNECTED) 19] create /lock v  
Created /lock
[zk: localhost:2181(CONNECTED) 20] create /lock v
Node already exists: /lock

2.创建有序节点

[zk: localhost:2181(CONNECTED) 28] create -s /lock/ v
Created /lock/0000000000
[zk: localhost:2181(CONNECTED) 29] create -s /lock/ v
Created /lock/0000000001
[zk: localhost:2181(CONNECTED) 30] create -s /lock/ v
Created /lock/0000000002
[zk: localhost:2181(CONNECTED) 31] create -s /lock/ v
Created /lock/0000000003
[zk: localhost:2181(CONNECTED) 32] create -s /lock/ v
Created /lock/0000000004
[zk: localhost:2181(CONNECTED) 34] ls /lock 
[0000000001, 0000000000, 0000000003, 0000000002, 0000000004]
[zk: localhost:2181(CONNECTED) 35] 

3.创建临时节点

[zk: localhost:2181(CONNECTED) 0] create -e /temp a
Created /temp

//退出重新进入，/temp节点被删除了
[zk: localhost:2181(CONNECTED) 3] quit
Quitting...
2019-09-29 15:30:44,083 [myid:] - INFO  [main:ZooKeeper@684] - Session: 0x16d7bc6e3490002 closed
2019-09-29 15:30:44,089 [myid:] - INFO  [main-EventThread:ClientCnxn$EventThread@519] - EventThread shut down for session: 0x16d7bc6e3490002
[root@localhost bin]# sh zkCli.sh 
[zk: localhost:2181(CONNECTED) 0] ls / 
[zookeeper, lock]

4.删除节点

//存在子节点时不能删除，须把子节点先删除
[zk: localhost:2181(CONNECTED) 47] delete /lock
Node not empty: /lock

[zk: localhost:2181(CONNECTED) 48] delete /lock/0000000004
[zk: localhost:2181(CONNECTED) 49] ls /lock               
[0000000001, 0000000000, 0000000003, 0000000002]

 5.修改

//修改
[zk: localhost:2181(CONNECTED) 13] set /tem a001
cZxid = 0x20
ctime = Sun Sep 29 15:44:08 CST 2019
mZxid = 0x21
mtime = Sun Sep 29 15:58:28 CST 2019
pZxid = 0x20
cversion = 0
dataVersion = 1
aclVersion = 0
ephemeralOwner = 0x0
dataLength = 4
numChildren = 0
//查看
[zk: localhost:2181(CONNECTED) 14] get /tem     
a001
cZxid = 0x20
ctime = Sun Sep 29 15:44:08 CST 2019
mZxid = 0x21
mtime = Sun Sep 29 15:58:28 CST 2019
pZxid = 0x20
cversion = 0
dataVersion = 1
aclVersion = 0
ephemeralOwner = 0x0
dataLength = 4
numChildren = 0

6.watch机制    get path [watch]，监听事件被触发一次后即失效

//watch /tem节点
[zk: localhost:2181(CONNECTED) 18] get /tem true 
a001
cZxid = 0x20
ctime = Sun Sep 29 15:44:08 CST 2019
mZxid = 0x21
mtime = Sun Sep 29 15:58:28 CST 2019
pZxid = 0x20
cversion = 0
dataVersion = 1
aclVersion = 0
ephemeralOwner = 0x0
dataLength = 4
numChildren = 0
[zk: localhost:2181(CONNECTED) 19] 
[zk: localhost:2181(CONNECTED) 19] 
[zk: localhost:2181(CONNECTED) 19] 
[zk: localhost:2181(CONNECTED) 19] 

//当/tmp节点被修改时，会触发
WATCHER::
WatchedEvent state:SyncConnected type:NodeDataChanged path:/tem

 

三，应用

1，负载均衡

•   拿到目标服务下的所有地址
•   动态感知

2，分布式锁（有序节点，watch机制）

3，全局ID（基于有序节点实现）