分布式管中窥豹之zookeeper小白学习（五）启动zookeeper仲裁模式

欢迎来到分布式管中窥豹之zookeeper小白学习系列，本系列会记录zookeeper以及分布式系统学习过程中的一些操作和细节，大饼果子非科班出身，如有错误，欢迎指出

1. 首先我们需要修改一下配置文件

节点1的配置文件z1.cfg

tickTime=2000
initLimit=10
syncLimit=5
dataDir=/Users/dabingguozi/Documents/zookeeper/data/z1/data
clientPort=2181
server.1=127.0.0.1:2222:2223
server.2=127.0.0.1:3333:3334
server.3=127.0.0.1:4444:4445

节点2的配置文件z2.cfg

tickTime=2000
initLimit=10
syncLimit=5
dataDir=/Users/dabingguozi/Documents/zookeeper/data/z2/data
clientPort=2182
server.1=127.0.0.1:2222:2223
server.2=127.0.0.1:3333:3334
server.3=127.0.0.1:4444:4445

节点3的配置文件z3.cfg

tickTime=2000
initLimit=10
syncLimit=5
dataDir=/Users/dabingguozi/Documents/zookeeper/data/z3/data
clientPort=2183
server.1=127.0.0.1:2222:2223
server.2=127.0.0.1:3333:3334
server.3=127.0.0.1:4444:4445

因为是本机的三个节点，所以需要不同的端口号

最后的三行是对节点的配置，server.n的n是服务的编号，我们会发现每个server配有两个端口号，前一个用于仲裁通信，后一个用于群首选举（涉及到分布式一致性算法，稍后会降到讲到）

2. 现在我们需要在dataDir的路径下指定节点的编号

echo 1 > /Users/dabingguozi/Documents/zookeeper/data/z1/data/myid
echo 2 > /Users/dabingguozi/Documents/zookeeper/data/z2/data/myid
echo 3 > /Users/dabingguozi/Documents/zookeeper/data/z3/data/myid

3. 启动服务

到z1的路径下启动

$ ./../../zookeeper-3.4.12/bin/zkServer.sh start ./z1.cfg

我们会发现他在尝试和其他节点通信，但是并没有成功（log在zookeeper.out文件中）

2019-01-29 10:40:14,968 [myid:1] - WARN  [QuorumPeer[myid=1]/0:0:0:0:0:0:0:0:2181:[email protected]] - Cannot open channel to 2
at election address /127.0.0.1:3334
java.net.ConnectException: Connection refused (Connection refused)
        at java.base/java.net.PlainSocketImpl.socketConnect(Native Method)
        at java.base/java.net.AbstractPlainSocketImpl.doConnect(AbstractPlainSocketImpl.java:400)
        at java.base/java.net.AbstractPlainSocketImpl.connectToAddress(AbstractPlainSocketImpl.java:243)
        at java.base/java.net.AbstractPlainSocketImpl.connect(AbstractPlainSocketImpl.java:225)
        at java.base/java.net.SocksSocketImpl.connect(SocksSocketImpl.java:402)
        at java.base/java.net.Socket.connect(Socket.java:591)
        at org.apache.zookeeper.server.quorum.QuorumCnxManager.connectOne(QuorumCnxManager.java:558)
        at org.apache.zookeeper.server.quorum.QuorumCnxManager.connectAll(QuorumCnxManager.java:610)
        at org.apache.zookeeper.server.quorum.FastLeaderElection.lookForLeader(FastLeaderElection.java:838)
        at org.apache.zookeeper.server.quorum.QuorumPeer.run(QuorumPeer.java:957)
2019-01-29 10:40:14,968 [myid:1] - INFO  [QuorumPeer[myid=1]/0:0:0:0:0:0:0:0:2181:[email protected]] - Resolved hostname: 127.0.0.1 to address: /127.0.0.1
2019-01-29 10:40:14,969 [myid:1] - WARN  [QuorumPeer[myid=1]/0:0:0:0:0:0:0:0:2181:[email protected]] - Cannot open channel to 3 at election address /127.0.0.1:4445
java.net.ConnectException: Connection refused (Connection refused)
        at java.base/java.net.PlainSocketImpl.socketConnect(Native Method)
        at java.base/java.net.AbstractPlainSocketImpl.doConnect(AbstractPlainSocketImpl.java:400)
        at java.base/java.net.AbstractPlainSocketImpl.connectToAddress(AbstractPlainSocketImpl.java:243)
        at java.base/java.net.AbstractPlainSocketImpl.connect(AbstractPlainSocketImpl.java:225)
        at java.base/java.net.SocksSocketImpl.connect(SocksSocketImpl.java:402)
        at java.base/java.net.Socket.connect(Socket.java:591)

再去到z2和z3的路径下分别启动服务

$ ./../../zookeeper-3.4.12/bin/zkServer.sh start ./z2.cfg
$ ./../../zookeeper-3.4.12/bin/zkServer.sh start ./z3.cfg

那么我们再次观察log

2019-01-29 10:41:55,925 [myid:1] - INFO  [QuorumPeer[myid=1]/0:0:0:0:0:0:0:0:2181:[email protected]] - FOLLOWING - LEADER ELECTION TOOK - 152007
2019-01-29 10:41:55,927 [myid:1] - INFO  [QuorumPeer[myid=1]/0:0:0:0:0:0:0:0:2181:QuorumPeer$Qu[email protected]] - Resolved hostname: 127.0.0.1 to address: /127.0.0.1
2019-01-29 10:41:55,973 [myid:1] - INFO  [QuorumPeer[myid=1]/0:0:0:0:0:0:0:0:2181:[email protected]] - Getting a diff from the leader 0x40000000a

z1已经成功的和z2还有z3完成通信了

4. 尝试连接

这次我们用z3的端口连接

$ ./zkCli.sh -server localhost:2183
Connecting to localhost:2183
2019-01-29 10:45:09,499 [myid:] - INFO  [main:[email protected]] - Client environment:zookeeper.version=3.4.12-e5259e437540f349646870ea94dc2658c4e44b3b, built on 03/27/2018 03:55 GMT
......Welcome to ZooKeeper!
2019-01-29 10:45:09,536 [myid:] - INFO  [main-SendThread(localhost:2183):[email protected]] - Opening socket connection to server localhost/0:0:0:0:0:0:0:1:2183. Will not attempt to authenticate using SASL (unknown error)
2019-01-29 10:45:09,550 [myid:] - INFO  [main-SendThread(localhost:2183):[email protected]] - Socket connection established to localhost/0:0:0:0:0:0:0:1:2183, initiating session
JLine support is enabled
2019-01-29 10:45:09,604 [myid:] - INFO  [main-SendThread(localhost:2183):[email protected]] - Session establishment complete on server localhost/0:0:0:0:0:0:0:1:2183, sessionid = 0x300322ca7c90000, negotiated timeout = 30000

WATCHER::

WatchedEvent state:SyncConnected type:None path:null
[zk: localhost:2183(CONNECTED) 0] 

我们成功连接上了

那么让我们去看一下这三个节点的状态信息

$ ./../../zookeeper-3.4.12/bin/zkServer.sh status ./z1.cfg
ZooKeeper JMX enabled by default
Using config: ./z1.cfg
Mode: follower

$ ./../../zookeeper-3.4.12/bin/zkServer.sh status ./z2.cfg
ZooKeeper JMX enabled by default
Using config: ./z2.cfg
Mode: leader

$ ./../../zookeeper-3.4.12/bin/zkServer.sh status ./z3.cfg
ZooKeeper JMX enabled by default
Using config: ./z3.cfg
Mode: follower

因为随机性，mode是leader的不一定是z2，但一定有一个节点是leader

这里就出现了这个概念，leader，follower，这是zookeeper仲裁模式下各个节点的角色（还有一个角色叫做observer）

之前的篇幅提到过zookeeper可以用来管理协同数据，那么这个管理的服务如果是单点的（像我们之前启动的独立模式），他的可用性就不高，也就是说，如果这个zookeeper节点宕机或者发生网络异常就没有办法再管理数据了。仲裁模式其实就是为解决这样的问题而提供的多节点协同服务，它也是一种分布式的服务。那么作为分布式的服务，为保证接入每一个节点拿到的数据都是一致的，就需要分布式的一致性协议去保证。

分布式的终极一致性协议paxos，本果子目前还没有甚解，但是zookeeper的ZAB（zookeeper atomic broadcast）协议，我们可以先一起学习一下

管中窥豹，可见一斑

那么下一篇就来研究一下这个ZAB协议吧

https://www.cnblogs.com/dabingguozi/p/10333110.html

原文地址：https://www.cnblogs.com/dabingguozi/p/10329411.html