NN1  NN2  DN  ZK  ZKFC  JNN
NODE01    *               *    *
NODE02        *    *   *    *    *
NODE03           *   *         *
NODE04           *   *

1、首先让两个NameNode能互相免密钥,在前一部分中node01已经能免密登陆node02了,所以再让node02免密登陆自己和node01就行

[ .ssh]# ssh-keygen -t dsa -P ‘‘ -f ~/.ssh/id_dsa
[ .ssh]# cat ~/.ssh/id_dsa.pub >> ~/.ssh/authorized_keys
[ .ssh]# scp id_dsa.pub node01:`pwd`/node02.pub

[ .ssh]# cat ~/.ssh/node02.pub >> ~/.ssh/authorized_keys

2、修改Hadoop配置文件,hdfs-site.xml

[ hadoop]# vi hdfs-site.xml
删除
    <property>
        <name>dfs.namenode.secondary.http-address</name>
        <value>node02:50090</value>
    </property>

添加
    <property>
        <name>dfs.nameservices</name>  //主节点服务名称
        <value>mycluster</value>
    </property>
    <property>
        <name>dfs.ha.namenodes.mycluster</name>  //namenode ID
        <value>nn1,nn2</value>
    </property>
    <property>
        <name>dfs.namenode.rpc-address.mycluster.nn1</name>  //namenode地址
        <value>node01:8020</value>
    </property>
    <property>
        <name>dfs.namenode.rpc-address.mycluster.nn2</name>
        <value>node02:8020</value>
    </property>
    <property>
        <name>dfs.namenode.http-address.mycluster.nn1</name>  //namenode的http服务地址
        <value>node01:50070</value>
    </property>
    <property>
        <name>dfs.namenode.http-address.mycluster.nn2</name>
        <value>node02:50070</value>
    </property>  <property>    <name>dfs.namenode.shared.edits.dir</name>  //JournalNode地址    <value>qjournal://node01:8485;node02:8485;node03:8485/mycluster</value>  </property>  <property>    <name>dfs.journalnode.edits.dir</name>    //JournalNode的edits保存路径    <value>/var/ycyz/hadoop/journal</value>  </property>  <property>    <name>dfs.client.failover.proxy.provider.mycluster</name>  //故障转移的代理模式    <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value>  </property>  <property>    <name>dfs.ha.fencing.methods</name>  //故障namenode状态隔离    <value>sshfence</value>  </property>  <property>    <name>dfs.ha.fencing.ssh.private-key-files</name>  //私钥路径    <value>/root/.ssh/id_dsa</value>  </property>

3、修改core-site.xml

[ hadoop]# vi core-site.xml    <property>
      <name>fs.defaultFS</name>
    <value>hdfs://mycluster</value>
  </property>

4、配置zookeeper,在Hadoop配置文件中设置自动故障转移,并配置zookeeper集群地址

[ hadoop]# vi hdfs-site.xml  <property>
    <name>dfs.ha.automatic-failover.enabled</name>
    <value>true</value>  </property>[ hadoop]# vi core-site.xml  <property>    <name>ha.zookeeper.quorum</name>    <value>node02:2181,node03:2181,node04:2181</value>  </property>

5、将hdfs-site.xml和core-site.xml分发给其他所有Hadoop节点

[ hadoop]# scp hdfs-site.xml  core-site.xml node02:`pwd`

6、安装zookeeper3.4.6,压缩包解压在其他路径下

[ software]# tar xf zookeeper-3.4.6.tar.gz -C /opt/ycyz/

7、进入zookeeper的conf目录,拷贝zoo_sample.cfg为zoo.cfg 并配置dataDir,集群节点

[ conf]# cp zoo_sample.cfg zoo.cfg
[ conf]# vi zoo.cfg 
    dataDir=/var/ycyz/zookeeper

    server.1=node02:2888:3888    //集群节点对应的id
    server.2=node03:2888:3888
    server.3=node04:2888:3888

8、将zookeeper安装文件分发到其他zookeeper节点

[ ycyz]# scp -r zookeeper-3.4.6/ node03:`pwd`

9、每个zookeeper节点都创建dataDir的目录

[ ycyz]# mkdir -p /var/ycyz/zookeeper

10、告知每个zookeeper节点自己对应的id

[ ycyz]# echo 1 > /var/ycyz/zookeeper/myid

[ ~]# echo 2 > /var/ycyz/zookeeper/myid

[ ~]# echo 3 > /var/ycyz/zookeeper/myid

11、配置zookeeper环境变量,并分发到其他节点

[ zookeeper-3.4.6]# vi /etc/profile
    export ZOOKEEPER_HOME=/opt/ycyz/zookeeper-3.4.6
    PATH=$PATH:$JAVA_HOME/bin:$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$ZOOKEEPER_HOME/bin
[ zookeeper-3.4.6]# scp /etc/profile node03:/etc/
[ zookeeper-3.4.6]# source /etc/profile

12、启动zookeeper集群

与Hadoop集群只需在一个节点上执行启动命令不同必须手动在所有zookeeper节点上执行 zkServer.sh start查看节点状态 zkServer.sh status

 13、启动所有JournalNode

hadoop-daemon.sh start journalnode

14、格式化namenode,如果两个namenode均未格式化,则先在其中一个namenode上格式化。

格式化完成后,启动已格式化的namenode,然后将格式化信息同步给未格式化的namenode节点。

[ ~]# hdfs namenode -format            格式化 
[ ~]# hadoop-daemon.sh start namenode    启动节点

[ ~]# hdfs namenode -bootstrapStandby    在未格式化节点上同步

15、在namenode上格式化zkfc

[ ~]# hdfs zkfc -formatZK

16、启动Hadoop集群,启动前请确认zookeeper集群已启动

[ ~]# start-dfs.sh