hadoop2.7.3 HA高可用集群安装


    1. 环境准备

    2. Hadoop安装配置


    4. 修改配置

    1. 所有节点安装jdk1.8，ssh，配置hosts文件，配置centos680和centos681之间的互相免登陆，centos680到其他所有机器的免登陆，关闭防火墙。

    2. 在centos682，centos683，centos684上安装Zookeeper3.4.9。


    4. 复制hadoop-2.7.3.tar.gz到centos680的/opt目录下，以后所有的操作都在centos680上进行。

    5. 解压hadoop-2.7.3.tar.gz到/opt/bigdata文件夹下：tar –zxvf hadoop-2.7.3.tar.gz –C /opt/bigdata/。




    1. ResourceManager借助zk实现热备，当某个节点失效之后另一个节点能够被通知，并接管任务。


    2. NN必须实现共享元数据才能保证无缝切换；共享元数据可以采用Linux提供的NFS服务，也可以使用Hadoop提供的JournalNode；JournalNode采用多数派理论，保证半数以上的节点写成功即认为成功。

    3. NN要实现自动切换（不需要手工切换），就必须实现实时监控每个节点的状态；这里采用DFSFailoverController进行监控，如果某个NN失败，其他NN能够通过ZK被通知到，并接替NN的任务。

    1. HDFS高可用性（NameNode）

    3. RM高可用性（ResourceManager）

    5. Hadoop2.7.3 HA搭建步骤



    2. hadoop-env.sh



  3. 修改JAVA_HOME



    2. core-site.xml





    2. hdfs-site.xml



  7. [html] view plain copy

    1. <property>   

    2.        <name>dfs.nameservicesname>   

    3.        <value>nsvalue>   

    4.    property>   

    5.   

    6.    <property>   

    7.        <name>dfs.ha.namenodes.nsname>   

    8.        <value>nn1,nn2value>   

    9.    property>   

    10.   

    11.    <property>   

    12.        <name>dfs.namenode.rpc-address.ns.nn1name>   

    13.        <value>centos680:9000value>   

    14.    property>   

    15.   

    16.    <property>   

    17.        <name>dfs.namenode.http-address.ns.nn1name>   

    18.        <value>centos680:50070value>   

    19.    property>   

    20.   

    21.    <property>   

    22.        <name>dfs.namenode.rpc-address.ns.nn2name>   

    23.        <value>centos681:9000value>   

    24.    property>   

    25.   

    26.    <property>   

    27.        <name>dfs.namenode.http-address.ns.nn2name>   

    28.        <value>centos681:50070value>   

    29.    property>   

    30.   

    31.    <property>   

    32.        <name>dfs.namenode.shared.edits.dirname>   

    33.        <value>qjournal://zk1:8485;zk2:8485;zk3:8485/nsvalue>   

    34.    property>   

    35.   

    36.    <property>   

    37.        <name>dfs.journalnode.edits.dirname>   

    38.        <value>/opt/big/hadoop-2.7.3/journaldatavalue>   

    39.    property>   

    40.   

    41.    <property>   

    42.        <name>dfs.ha.automatic-failover.enabledname>   

    43.        <value>truevalue>   

    44.    property>   

    45.   

    46.    <property>   

    47.        <name>dfs.client.failover.proxy.provider.nsname>   

    48.        <value>  

    49.                org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider  

    50.        value>   

    51.    property>   

    52.    <property>   

    53.        <name>dfs.ha.fencing.methodsname>   

    54.        <value>   

    55.            sshfence   

    56.            shell(/bin/true)   

    57.        value>   

    58.    property>   

    59.    <property>   

    60.        <name>dfs.ha.fencing.ssh.private-key-filesname>   

    61.        <value>/root/.ssh/id_rsavalue>   

    62.    property>   

    63.    <property>   

    64.        <name>dfs.ha.fencing.ssh.connect-timeoutname>   

    65.        <value>30000value>   

    66.    property>  



    2. mapred-side.xml



  9. [html] view plain copy



    1. <span style="white-space:pre">  span><property>  

    2.                 <name>mapreduce.framework.namename>  

    3.                 <value>yarnvalue>  

    4.         property>  



    2. yarn-site.xml

    3. [html] view plain copy



    4. <configuration>  

    5.                                             

    6.   

    7.                                           <property>  

    8.   

    9.                                              <name>yarn.resourcemanager.ha.enabledname>  

    10.   

    11.                                              <value>truevalue>  

    12.   

    13.                                           property>  

    14.   

    15.                                             

    16.   

    17.                                           <property>  

    18.   

    19.                                              <name>yarn.resourcemanager.cluster-idname>  

    20.   

    21.                                              <value>yrcvalue>  

    22.   

    23.                                           property>  

    24.   

    25.                                             

    26.   

    27.                                           <property>  

    28.   

    29.                                              <name>yarn.resourcemanager.ha.rm-idsname>  

    30.   

    31.                                              <value>rm1,rm2value>  

    32.   

    33.                                           property>  

    34.   

    35.                                             

    36.   

    37.                                           <property>  

    38.   

    39.                                              <name>yarn.resourcemanager.hostname.rm1name>  

    40.   

    41.                                              <value>h0value>  

    42.   

    43.                                           property>  

    44.   

    45.                                           <property>  

    46.   

    47.                                              <name>yarn.resourcemanager.hostname.rm2name>  

    48.   

    49.                                              <value>h1value>  

    50.   

    51.                                           property>  

    52.   

    53.                                             

    54.   

    55.                                           <property>  

    56.   

    57.                                              <name>yarn.resourcemanager.zk-addressname>  

    58.   

    59.                                              <value>h2:2181,h3:2181,h4:2181value>  

    60.   

    61.                                           property>  

    62.   

    63.                                           <property>  

    64.   

    65.                                              <name>yarn.nodemanager.aux-servicesname>  

    66.   

    67.                                              <value>mapreduce_shufflevalue>  

    68.   

    69.                                           property>  

    70.   

    71.                             configuration>  





    2. slaves





    2. 分发到其他节点（在centos680上操作）


  14.   scp -r /opt/bigdata/hadoop-2.7.3/ h1:/opt/bigdata/
      scp -r /opt/bigdata/hadoop-2.7.3/ h2:/opt/bigdata/
      scp -r /opt/bigdata/hadoop-2.7.3/ h3:/opt/bigdata/
      scp -r /opt/bigdata/hadoop-2.7.3/ h4:/opt/bigdata/



    2. 到h2,h3,h4启动所有的Zookeeper。

    3. 到h2,h3,h4启动JournalNode（hadoop-daemon.sh start journalnode）

    4. 到centos680上执行格式化namenode（hdfs namenode -format）,并将格式化后的元数据内容复制到另外一个namenode节点中（h1）:scp -r tmp/ h1:/opt/bigdata/hadoop-2.7.3/(tmp是在core-site中配置的namenode元数据应该所处的位置)

    5. 格式化zk：bin/hdfs zkfc –formatZK。该操作在Zookeeper中创建了数据节点：

    6. 初始化和启动




    2. 启动dfs和yarn



  18. sbin/start-dfs.sh

    sbin/start-yarn.sh



    2. 验证


  20.   通过杀死active的namenode进程或者关闭namenode所在机器来验证切换情况。