hadoop3.0高可用HA配置详解

hadoop3.0高可用HA大数据平台架构硬件和部署方案（一）

http://blog.csdn.net/lxb1022/article/details/78389836

hadoop3.0高可用HA大数据平台架构软件和部署方案（二)
http://blog.csdn.net/lxb1022/article/details/78399462

1.到官网下载hadoop-3.0.0-beta1.tar.gz

http://mirrors.tuna.tsinghua.edu.cn/apache/hadoop/common/hadoop-3.0.0-beta1/hadoop-3.0.0-beta1.tar.gz

2、解压到安装目录/opt/，并配置环境变量：vi  /etc/profile

tar -zxvf hadoop-3.0.0-beta1.tar.gz
#set java env
export JAVA_HOME=/opt/jdk1.8.0_144
export PATH=$PATH:$JAVA_HOME/bin
export CLASSPATH=.:$JAVA_HOME/lib/rt.jar

#set hadoop env
export HADOOP_HOME=/opt/hadoop-3.0.0-beta1
export PATH=$PATH:$HADOOP_HOME/bin
export PATH=$PATH:$HADOOP_HOME/sbin

3、配置：/opt/hadoop-3.0.0-beta1/etc/hadoop/hadoop-env.sh

export JAVA_HOME=/opt/jdk1.8.0_144
export HADOOP_HOME=/opt/hadoop-3.0.0-beta1

4.配置：/opt/hadoop-3.0.0-beta1/etc/hadoop/core-site.xml

     
        fs.defaultFS 
        hdfs://mycluster
        mycluster唯一标识hdfs实例的nameservice ID的名称
    
     
        hadoop.tmp.dir 
        file:///opt/hadoop-3.0.0-beta1/tmp
         指定hadoop临时目录
    
    
        ha.zookeeper.quorum
        zookeeper1:2181,zookeeper2:2181,zookeeper:2181
        指定zookeeper地址
    

5、配置：/opt/hadoop-3.0.0-beta1/etc/hadoop/hdfs-site.xml

    
        dfs.nameservices
        mycluster
        指定hdfs的nameservice为mycluster，需要和core-site.xml中的保持一致
    

    
        dfs.ha.namenodes.mycluster
        nn1,nn2
        mycluster下面有两个NameNode，分别是nn1，nn2
    


    
        dfs.namenode.rpc-address.mycluster.nn1
        namenode1:9000
        nn1的RPC通信地址
    


    
        dfs.namenode.rpc-address.mycluster.nn2
        namenode2:9000
        nn2的RPC通信地址
    

    
        dfs.namenode.http-address.mycluster.nn1
        namenode1:50070
        nn1的http通信地址
    
    
        dfs.namenode.http-address.mycluster.nn2
        namenode2:50070
        nn2的http通信地址
    

    
        dfs.namenode.shared.edits.dir
        qjournal://zookeeper1:8485;zookeeper2:8485;zookeeper:8485/mycluster
        指定NameNode的元数据在JournalNode上的存放位置
    

    
        dfs.journalnode.edits.dir
        /opt/hadoop-3.0.0-beta1/journal
        新建目录，用于设置journalnode节点保存本地状态的目录,指定journalnode日志文件存储的路径
    

    
        dfs.client.failover.proxy.provider.mycluster
        org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider
        配置失败自动切换实现方式,指定HDFS客户端连接active namenode的java类
    

    
        dfs.ha.fencing.methods
        sshfence
        配置隔离机制为ssh
    

    
        dfs.ha.fencing.ssh.private-key-files
        /root/.ssh/id_rsa
        使用隔离机制时需要ssh免密码登陆,指定秘钥的位置
    

    
        dfs.ha.automatic-failover.enabled
        true
        指定支持高可用自动切换机制,开启自动故障转移
    

    
        ha.zookeeper.quorum
        zookeeper1:2181,zookeeper2:2181,zookeeper:2181
        指定zookeeper地址
    

     
        dfs.replication
        3
        指定数据冗余份数 
    

     
        dfs.namenode.name.dir 
        file:///opt/hadoop-3.0.0-beta1/name
        新建name文件夹，指定namenode名称空间的存储地址 
    

     
        dfs.datanode.data.dir 
        file:///opt/hadoop-3.0.0-beta1/data
        新建data文件夹，指定datanode数据存储地址 
    

    
        dfs.webhdfs.enabled
        ture
        指定可以通过web访问hdfs目录
    

6、配置：/opt/hadoop-3.0.0-beta1/etc/hadoop/mapred-site.xml

    
        mapreduce.framework.name
        yarn
    

    
        mapreduce.application.classpath
        
         /opt/hadoop-3.0.0-beta1/etc/hadoop,
         /opt/hadoop-3.0.0-beta1/share/hadoop/common/*,
         /opt/hadoop-3.0.0-beta1/share/hadoop/common/lib/*,
         /opt/hadoop-3.0.0-beta1/share/hadoop/hdfs/*,
         /opt/hadoop-3.0.0-beta1/share/hadoop/hdfs/lib/*,
         /opt/hadoop-3.0.0-beta1/share/hadoop/mapreduce/*,
         /opt/hadoop-3.0.0-beta1/share/hadoop/mapreduce/lib/*,
         /opt/hadoop-3.0.0-beta1/share/hadoop/yarn/*,
         /opt/hadoop-3.0.0-beta1/share/hadoop/yarn/lib/*
        
    

7、配置：/opt/hadoop-3.0.0-beta1/etc/hadoop/yarn-site.xml

    
        yarn.nodemanager.aux-services
        mapreduce_shuffle
        NodeManager上运行的附属服务，需配置成mapreduce_shuffle才可运行MapReduce
    

    
        yarn.resourcemanager.ha.enabled
        true
        开启YARN HA
    

    
        yarn.resourcemanager.ha.automatic-failover.enabled
        true
        启用自动故障转移
    

    
        yarn.resourcemanager.cluster-id
        yarncluster
        指定YARN HA的名称
    

    
        yarn.resourcemanager.ha.rm-ids
        rm1,rm2
        指定两个resourcemanager的名称
    

    
        yarn.resourcemanager.hostname.rm1
        namenode1
        配置rm1的主机
    

    
        yarn.resourcemanager.hostname.rm2
        namenode2
        配置rm2的主机
    

    
        yarn.resourcemanager.ha.id
        rm1
        namenode1上配置rm1,在namenode上配置rm2
    

    
        yarn.resourcemanager.zk-address
        zookeeper1:2181,zookeeper2:2181,zookeeper:2181
        配置zookeeper的地址
    

    
        yarn.resourcemanager.store.class
        org.apache.hadoop.yarn.server.resourcemanager.recovery.ZKRMStateStore
        配置resourcemanager的状态存储到zookeeper中
    

    
        yarn.resourcemanager.webapp.address.rm1
        namenode1:8088
    

yarn.resourcemanager.webapp.address.rm2 namenode2:8088

7、配置：/opt/hadoop-3.0.0-beta1/etc/hadoop/workers

datanode1
datanode2
datanode3

8、将/opt/hadoop-3.0.0-beta1/  复制到各个节点

scp /opt/hadoop-3.0.0-beta1/ [email protected]:/opt/

9、格式化hdfs

1.三个ZK节点先启动Zookeeper
/opt/zookeeper-3.5.3-beta/bin/zkServer.sh start

2.接着三个ZK节点启动journalnode
/opt/hadoop-3.0.0-beta1/sbin/hadoop-daemon.sh start journalnode

3.在master节点上执行格式化
格式化namenode:/opt/hadoop-3.0.0-beta1/bin/hdfs namenode -format
格式化高可用：/opt/hadoop-3.0.0-beta1/bin/hdfs zkfc -formatZK

10、启动hadoop :

/opt/hadoop-3.0.0-beta1/sbin/start-all.sh

11.查看各节点进程jps

namenode1和namenode2如下图：

zk辅助节点进程如下图：

datanode节点进程如下图：

12、通过web页面查看：