Hadoop HA的安装

Hadoop集群的安装

在master1节点解压hadoop安装包
tar zxvf ./hadoop-2.6.5.tar.gz
修改hadoop-env.sh文件
cd /usr/local/hadoop-2.6.5/etc/hadoop
sudo nano hadoop-env.sh
添加JAVA_HOME
export JAVA_HOME=/usr/local/jdk1.7.0_80
配置core-site.xml文件
sudo nano core-site.xml
添加以下内容：

  
  
    fs.defaultFS
    hdfs://mycluster
  
    
  
  
    hadoop.tmp.dir
    /usr/local/hadoop-2.6.5/data/tmp
  

  
  
    fs.trash.interval
    1440
  
  
  
  
    ha.zookeeper.quorum
    slave1:2181,slave2:2181,slave3:2181
  

配置hdfs-site.xml文件
sudo nano hdfs-site.xml
添加以下内容：

  
  
    dfs.namenode.name.dir
    /usr/local/hadoop-2.6.5/data/namenode
  

  
  
    dfs.datanode.data.dir
    /usr/local/hadoop-2.6.5/data/datanode
  
  
  
  
    dfs.replication
    3
  

  
  
    dfs.permissions.enabled
    false
  
  
  
  
    dfs.webhdfs.enabled
    true
  
  
  
  
  
    dfs.nameservices
    mycluster
  

  
  
    dfs.ha.namenodes.mycluster
    nn1,nn2
  

  
  
    dfs.namenode.rpc-address.mycluster.nn1
    master1:8020
  
  
    dfs.namenode.rpc-address.mycluster.nn2
    master2:8020
  

  
  
    dfs.namenode.http-address.mycluster.nn1
    master1:50070
  
  
    dfs.namenode.http-address.mycluster.nn2
    master2:50070
  

  
  
    dfs.namenode.shared.edits.dir
    qjournal://slave1:8485;slave2:8485;slave3:8485/mycluster
  
  
  
  
    dfs.journalnode.edits.dir
    /usr/local/hadoop-2.6.5/data/journal
  

  
  
    dfs.client.failover.proxy.provider.mycluster
    org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider
  

  
  
    dfs.ha.fencing.methods
    sshfence
  

  
  
    dfs.ha.fencing.ssh.private-key-files
    /home/hadoop-sna/.ssh/id_rsa
  
  
  
  
    dfs.ha.automatic-failover.enabled
    true
  

配置mapred-site.xml文件
sudo nano mapred-site.xml
添加以下内容：

  
  
    mapreduce.framework.name
    yarn
  

  
  
    mapreduce.jobhistory.address
    master1:10020
  

  
  
    mapreduce.jobhistory.webapp.address
    master1:19888
  

  
  
    mapreduce.job.ubertask.enable
    true
  

  
  
    mapreduce.job.ubertask.maxmaps
    9
  

  
  
    mapreduce.job.ubertask.maxreduces
    1
  

配置yarn-site.xml文件
sudo nano yarn-site.xml
添加以下内容：

  
  
    yarn.nodemanager.aux-services
    mapreduce_shuffle
  

  
  
    yarn.web-proxy.address
    master2:8888
  
  
  
  
    yarn.log-aggregation-enable
    true
  

  
  
    yarn.log-aggregation.retain-seconds
    604800
  

  
  
    yarn.nodemanager.remote-app-log-dir
    /logs
  

  
  
    yarn.nodemanager.resource.memory-mb
    2048
  

  
  
    yarn.nodemanager.resource.cpu-vcores
    2
  
  
  
  
  
    yarn.resourcemanager.ha.enabled
    true
  

  
  
    yarn.resourcemanager.ha.automatic-failover.enabled
    true
  

  
  
    yarn.resourcemanager.cluster-id
    yarncluster
  

  
  
    yarn.resourcemanager.ha.rm-ids
    rm1,rm2
  

  
  
    yarn.resourcemanager.hostname.rm1
    master1
  
  
    yarn.resourcemanager.hostname.rm2
    master2
  

  
  
    yarn.resourcemanager.webapp.address.rm1
    master1:8088
     
  
    yarn.resourcemanager.webapp.address.rm2
    master2:8088
  

  
  
    yarn.resourcemanager.zk-address
    slave1:2181,slave2:2181,slave3:2181
  

  
  
    yarn.resourcemanager.zk-state-store.parent-path
    /rmstore
  

  
  
    yarn.resourcemanager.recovery.enabled
    true
  

  
  
    yarn.resourcemanager.store.class
    org.apache.hadoop.yarn.server.resourcemanager.recovery.ZKRMStateStore
  

  
  
    yarn.nodemanager.recovery.enabled
    true
  

  
  
    yarn.nodemanager.address
    0.0.0.0:45454
  

配置slaves文件
sudo nano slaves
添加以下内容：

slave1
slave2
slave3

创建配置文件中涉及的目录

cd /usr/local/hadoop-2.6.5/
mkdir-p data/tmp
mkdir-p data/journal
mkdir-p data/namenode
mkdir-p data/datanode

将hadoop工作目录同步到集群其它节点

scp -r /usr/local/hadoop-2.6.5/ master2:/usr/local/
scp -r /usr/local/hadoop-2.6.5/ slave1:/usr/local/
scp -r /usr/local/hadoop-2.6.5/ slave2:/usr/local/
scp -r /usr/local/hadoop-2.6.5/ slave3:/usr/local/

在集群各节点上修改环境变量
sudo nano /etc/profile
添加以下几行：

export HADOOP_HOME=/usr/local/hadoop-2.6.5
export LD_LIBRARY_PATH=$HADOOP_HOME/lib/native
export PATH=$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$PATH

使修改的环境变量生效
source /etc/profile
方便起见，还是将hadoop-2.6.5的目录权限改一下比较好

sudo chown -R hadoop-sna hadoop-2.6.5
sudo chgrp -R hadoop-sna hadoop-2.6.5

这样，hadoop集群的安装就完成了，接下来是启动集群服务。

Hadoop集群的初始化

启动zookeeper集群（分别在slave1、slave2和slave3上执行）
zkServer.sh start
格式化ZKFC（在master1上执行）
hdfs zkfc -formatZK
启动journalnode（分别在slave1、slave2和slave3上执行）
hadoop-daemon.sh start journalnode
格式化HDFS（在master1上执行）
hdfs namenode -format
将格式化后master1节点hadoop工作目录中的元数据目录复制到master2节点
scp -r /usr/local/hadoop-2.6.5/data/namenode/* master2:/usr/local/hadoop-2.6.5/data/namenode/
初始化完毕后可关闭journalnode（分别在slave1、slave2和slave3上执行）
hadoop-daemon.sh stop journalnode

遇到的问题：

1. hdfs namenode –format的时候出现connection refused
   原因：连接datanode的时候出现拒绝连接的现象，原因是journalnode没有启动，可以查看datanode上hadoop文件夹里面的logs里面的hadoop-hadoop-journalnode-data.out，里面有写journalnode启动失败的原因，我的原因是没有写入权限
   解决方法：

   sudo chown –R hadoop-sna hadoop-2.6.5
   sudo chgrp –R hadoop-sna hadoop-2.6.5