转自：http://blog.csdn.net/dr_guo/article/details/50975851
1.集群规划：
主机名 IP 安装的软件 运行的进程 drguo1 192.168.80.149 jdk、hadoop NameNode、DFSZKFailoverController(zkfc)、ResourceManager drguo2 192.168.80.150 jdk、hadoop NameNode、DFSZKFailoverController(zkfc)、ResourceManager drguo3 192.168.80.151 jdk、hadoop、zookeeper DataNode、NodeManager、JournalNode、QuorumPeerMain drguo4 192.168.80.152 jdk、hadoop、zookeeper DataNode、NodeManager、JournalNode、QuorumPeerMaindrguo5 192.168.80.153 jdk、Hadoop、zookeeper DataNode、NodeManager、JournalNode、QuorumPeerMain
排的好好的，显示出来就乱了！！！
2.前期准备：
准备五台机器，修改静态IP、主机名、主机名与IP的映射，关闭防火墙，安装JDK并配置环境变量（不会请看这http://blog.csdn.net/dr_guo/article/details/50886667），创建用户：用户组，SSH免密码登录SSH免密码登录（报错请看这http://blog.csdn.net/dr_guo/article/details/50967442）。
注意：要把127.0.1.1那一行注释掉，要不然会出现jps显示有datanode，但网页显示live nodes为0；
注释之后就正常了，好像有人没注释也正常，我也不知道为什么0.0
3.搭建zookeeper集群（drguo3/drguo4/drguo5）
见：ZooKeeper完全分布式集群搭建
4.正式开始搭建Hadoop HA集群
去官网下最新的Hadoop（http://apache.opencas.org/hadoop/common/stable/），目前最新的是2.7.2，下载完之后把它放到/opt/Hadoop下
[plain] view plain copy guo@guo:~/下载$ mv ./hadoop-2.7.2.tar.gz /opt/Hadoop/ mv: 无法创建普通文件"/opt/Hadoop/hadoop-2.7.2.tar.gz": 权限不够 guo@guo:~/下载$ su root 密码： root@guo:/home/guo/下载# mv ./hadoop-2.7.2.tar.gz /opt/Hadoop/ 解压 [plain] view plain copy guo@guo:/opt/Hadoop$ sudo tar -zxf hadoop-2.7.2.tar.gz [sudo] guo 的密码： 解压jdk的时候我用的是tar -zxvf，其中的v呢就是看一下解压的过程，不想看你可以不写。修改opt目录所有者(用户：用户组)直接把opt目录的所有者/组换成了guo。具体情况在ZooKeeper完全分布式集群搭建说过。
[plain] view plain copy root@guo:/opt/Hadoop# chown -R guo:guo /opt
设置环境变量
[plain] view plain copy guo@guo:/opt/Hadoop$ sudo gedit /etc/profile 在最后加上（这样设置在执行bin/sbin目录下的脚本时就不用进入该目录用了） [plain] view plain copy #hadoop export HADOOP_HOME=/opt/Hadoop/hadoop-2.7.2 export PATH=$PATH:$HADOOP_HOME/sbin export PATH=$PATH:$HADOOP_HOME/bin 然后更新配置 [plain] view plain copy guo@guo:/opt/Hadoop$ source /etc/profile修改/opt/Hadoop/hadoop-2.7.2/etc/hadoop下的hadoop-env.sh
[plain] view plain copy guo@guo:/opt/Hadoop$ cd hadoop-2.7.2 guo@guo:/opt/Hadoop/hadoop-2.7.2$ cd etc/hadoop/ guo@guo:/opt/Hadoop/hadoop-2.7.2/etc/hadoop$ sudo gedit ./hadoop-env.sh 进入文件后 [plain] view plain copy export JAVA_HOME=${JAVA_HOME}#将这个改成JDK路径，如下 export JAVA_HOME=/opt/Java/jdk1.8.0_73 然后更新文件配置 [plain] view plain copy guo@guo:/opt/Hadoop/hadoop-2.7.2/etc/hadoop$ source ./hadoop-env.sh 前面配置和单机模式一样，我就直接复制了。 注意：汉语注释是给你看的，复制粘贴的时候都删了！！！ 修改core-site.xml [html] view plain copy <configuration> <!-- 指定hdfs的nameservice为ns1 --> <property> <name>fs.defaultFS</name> <value>hdfs://ns1/</value> </property> <!-- 指定hadoop临时目录 --> <property> <name>hadoop.tmp.dir</name> <value>/opt/Hadoop/hadoop-2.7.2/tmp</value> </property> <!-- 指定zookeeper地址 --> <property> <name>ha.zookeeper.quorum</name> <value>drguo3:2181,drguo4:2181,drguo5:2181</value> </property> </configuration> 修改hdfs-site.xml [html] view plain copy <configuration> <!--指定hdfs的nameservice为ns1，需要和core-site.xml中的保持一致 --> <property> <name>dfs.nameservices</name> <value>ns1</value> </property> <!-- ns1下面有两个NameNode，分别是nn1，nn2 --> <property> <name>dfs.ha.namenodes.ns1</name> <value>nn1,nn2</value> </property> <!-- nn1的RPC通信地址 --> <property> <name>dfs.namenode.rpc-address.ns1.nn1</name> <value>drguo1:9000</value> </property> <!-- nn1的http通信地址 --> <property> <name>dfs.namenode.http-address.ns1.nn1</name> <value>drguo1:50070</value> </property> <!-- nn2的RPC通信地址 --> <property> <name>dfs.namenode.rpc-address.ns1.nn2</name> <value>drguo2:9000</value> </property> <!-- nn2的http通信地址 --> <property> <name>dfs.namenode.http-address.ns1.nn2</name> <value>drguo2:50070</value> </property> <!-- 指定NameNode的元数据在JournalNode上的存放位置 --> <property> <name>dfs.namenode.shared.edits.dir</name> <value>qjournal://drguo3:8485;drguo4:8485;drguo5:8485/ns1</value> </property> <!-- 指定JournalNode在本地磁盘存放数据的位置 --> <property> <name>dfs.journalnode.edits.dir</name> <value>/opt/Hadoop/hadoop-2.7.2/journaldata</value> </property> <!-- 开启NameNode失败自动切换 --> <property> <name>dfs.ha.automatic-failover.enabled</name> <value>true</value> </property> <!-- 配置失败自动切换实现方式 --> <property> <name>dfs.client.failover.proxy.provider.ns1</name> <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value> </property> <!-- 配置隔离机制方法，多个机制用换行分割，即每个机制暂用一行--> <property> <name>dfs.ha.fencing.methods</name> <value> sshfence shell(/bin/true) </value> </property> <!-- 使用sshfence隔离机制时需要ssh免登陆 --> <property> <name>dfs.ha.fencing.ssh.private-key-files</name> <value>/home/guo/.ssh/id_rsa</value> </property> <!-- 配置sshfence隔离机制超时时间 --> <property> <name>dfs.ha.fencing.ssh.connect-timeout</name> <value>30000</value> </property> </configuration> 先将mapred-site.xml.template改名为mapred-site.xml然后修改mapred-site.xml [html] view plain copy <configuration> <!-- 指定mr框架为yarn方式 --> <property> <name>mapreduce.framework.name</name> <value>yarn</value> </property> </configuration> 修改yarn-site.xml [html] view plain copy <configuration> <!-- 开启RM高可用 --> <property> <name>yarn.resourcemanager.ha.enabled</name> <value>true</value> </property> <!-- 指定RM的cluster id --> <property> <name>yarn.resourcemanager.cluster-id</name> <value>yrc</value> </property> <!-- 指定RM的名字 --> <property> <name>yarn.resourcemanager.ha.rm-ids</name> <value>rm1,rm2</value> </property> <!-- 分别指定RM的地址 --> <property> <name>yarn.resourcemanager.hostname.rm1</name> <value>drguo1</value> </property> <property> <name>yarn.resourcemanager.hostname.rm2</name> <value>drguo2</value> </property> <!-- 指定zk集群地址 --> <property> <name>yarn.resourcemanager.zk-address</name> <value>drguo3:2181,drguo4:2181,drguo5:2181</value> </property> <property> <name>yarn.nodemanager.aux-services</name> <value>mapreduce_shuffle</value> </property> </configuration> 修改slaves [html] view plain copy drguo3 drguo4 drguo5 把Hadoop整个目录拷贝到drguo2/3/4/5，拷之前把share下doc删了（文档不用拷），这样会快点。 5.启动zookeeper集群（分别在drguo3、drguo4、drguo5上启动zookeeper） [plain] view plain copy guo@drguo3:~$ zkServer.sh start ZooKeeper JMX enabled by default Using config: /opt/zookeeper-3.4.8/bin/../conf/zoo.cfg Starting zookeeper ... STARTED guo@drguo3:~$ jps 2005 Jps 1994 QuorumPeerMain guo@drguo3:~$ ssh drguo4 Welcome to Ubuntu 15.10 (GNU/Linux 4.2.0-16-generic x86_64) * Documentation: https://help.ubuntu.com/ Last login: Fri Mar 25 14:04:43 2016 from 192.168.80.151 guo@drguo4:~$ zkServer.sh start ZooKeeper JMX enabled by default Using config: /opt/zookeeper-3.4.8/bin/../conf/zoo.cfg Starting zookeeper ... STARTED guo@drguo4:~$ jps 1977 Jps 1966 QuorumPeerMain guo@drguo4:~$ exit 注销 Connection to drguo4 closed. guo@drguo3:~$ ssh drguo5 Welcome to Ubuntu 15.10 (GNU/Linux 4.2.0-16-generic x86_64) * Documentation: https://help.ubuntu.com/ Last login: Fri Mar 25 14:04:56 2016 from 192.168.80.151 guo@drguo5:~$ zkServer.sh start ZooKeeper JMX enabled by default Using config: /opt/zookeeper-3.4.8/bin/../conf/zoo.cfg Starting zookeeper ... STARTED guo@drguo5:~$ jps 2041 Jps 2030 QuorumPeerMain guo@drguo5:~$ exit 注销 Connection to drguo5 closed. guo@drguo3:~$ zkServer.sh status ZooKeeper JMX enabled by default Using config: /opt/zookeeper-3.4.8/bin/../conf/zoo.cfg Mode: leader 6.启动journalnode（分别在drguo3、drguo4、drguo5上启动journalnode）注意只有第一次需要这么启动，之后启动hdfs会包含journalnode。 [plain] view plain copy guo@drguo3:~$ hadoop-daemon.sh start journalnode starting journalnode, logging to /opt/Hadoop/hadoop-2.7.2/logs/hadoop-guo-journalnode-drguo3.out guo@drguo3:~$ jps 2052 Jps 2020 JournalNode 1963 QuorumPeerMain guo@drguo3:~$ ssh drguo4 Welcome to Ubuntu 15.10 (GNU/Linux 4.2.0-16-generic x86_64) * Documentation: https://help.ubuntu.com/ Last login: Fri Mar 25 00:09:08 2016 from 192.168.80.149 guo@drguo4:~$ hadoop-daemon.sh start journalnode starting journalnode, logging to /opt/Hadoop/hadoop-2.7.2/logs/hadoop-guo-journalnode-drguo4.out guo@drguo4:~$ jps 2103 Jps 2071 JournalNode 1928 QuorumPeerMain guo@drguo4:~$ exit 注销 Connection to drguo4 closed. guo@drguo3:~$ ssh drguo5 Welcome to Ubuntu 15.10 (GNU/Linux 4.2.0-16-generic x86_64) * Documentation: https://help.ubuntu.com/ Last login: Thu Mar 24 23:52:17 2016 from 192.168.80.152 guo@drguo5:~$ hadoop-daemon.sh start journalnode starting journalnode, logging to /opt/Hadoop/hadoop-2.7.2/logs/hadoop-guo-journalnode-drguo5.out guo@drguo5:~$ jps 2276 JournalNode 2308 Jps 1959 QuorumPeerMain guo@drguo5:~$ exit 注销 Connection to drguo5 closed. 在drguo4/5启动时发现了问题，没有journalnode，查看日志发现是因为汉语注释造成的，drguo4/5全删了问题解决。drguo4/5的拼音输入法也不能用，我很蛋疼。。镜像都是复制的，咋还变异了呢。 7.格式化HDFS(在drguo1上执行) [plain] view plain copy guo@drguo1:/opt$ hdfs namenode -format 这回又出问题了，还是汉语注释闹得，drguo1/2/3也全删了，问题解决。 注意：格式化之后需要把tmp目录拷给drguo2（不然drguo2的namenode起不来） [plain] view plain copy guo@drguo1:/opt/Hadoop/hadoop-2.7.2$ scp -r tmp/ drguo2:/opt/Hadoop/hadoop-2.7.2/ 8.格式化ZKFC(在drguo1上执行) [plain] view plain copy guo@drguo1:/opt$ hdfs zkfc -formatZK 9.启动HDFS(在drguo1上执行) [plain] view plain copy guo@drguo1:/opt$ start-dfs.sh 10.启动YARN(在drguo1上执行) [plain] view plain copy guo@drguo1:/opt$ start-yarn.sh PS: 1.drguo2的resourcemanager需要手动单独启动： yarn-daemon.sh start resourcemanager 2.namenode、datanode也可以单独启动： hadoop-daemon.sh start namenode hadoop-daemon.sh start datanode 3.NN 由standby转化成active hdfs haadmin -transitionToActive nn1 --forcemanual 大功告成！！！ 是不是和之前规划的一样0.0