基于ZK自动切换模式的实现（六）

配置自动切换模式

（1）配置zookeeper集群

（2）开启自动切换模式

在hdfs-site.xml中配置dfs.ha.automatic-failover.enabled参数

（3）配置zookeeper实例

在core-site.xml中配置ha.zookeeper.quorum

（4）初始化zookeeper

（5）启动journalnode，namenode和datanode

（6）启动zkfc

根据主机名可以判断每一条命令是在那个节点执行，是用哪个用户执行的

例子：

[hadoop@master opt]$ 

mater主机

hadoop用户

1） 上传zookeeper-3.4.6.tar.gz

2） 解压

[hadoop@master opt]$ mv zookeeper-3.4.6.tar.gz /opt

[hadoop@master opt]$ tar -zxvf zookeeper-3.4.6.tar.gz

3） 修改配置文件

[hadoop@master conf]$ cd /opt/zookeeper-3.4.6/conf

[hadoop@master conf]$ cp zoo_sample.cfg zoo.cfg

[hadoop@master conf]$ vi zoo.cfg

dataDir=/opt/zookeeper/data

server.1=hadoop04:8001:8002

server.2=slave1:8001:8002

server.3=slave2:8001:8002

4）将zookeeper软件copy到规划好的节点

[hadoop@master opt]$ scp -r zookeeper-3.4.6 hadoop04:/opt

[hadoop@master opt]$ scp -r zookeeper-3.4.6 slave1:/opt

[hadoop@master opt]$ scp -r zookeeper-3.4.6 slave2:/opt

5）在hadoop04、slave1、slave2上创建链接

[hadoop@hadoop04 opt]$ ln -s zookeeper-3.4.6 zookeeper

[hadoop@slave1 opt]$ ln -s zookeeper-3.4.6 zookeeper

[hadoop@slave2 opt]$ ln -s zookeeper-3.4.6 zookeeper

6）修改环境变量

On master

[hadoop@master ~]$ vi .bash_profile

[hadoop@master ~]$ vi .bash_profile 

        . ~/.bashrc

fi

# User specific environment and startup programs

PATH=$PATH:$HOME/bin

export PATH
export JAVA_HOME=/usr/java/default
export PATH=$JAVA_HOME/bin:$PATH
export CLASSPATH=$CLASSPATH:$JAVA_HOME/lib
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib

#Hadoop1.0
#export HADOOP1_HOME=/home/hadoop/hadoop
#export PATH=$HADOOP1_HOME/bin:$PATH
#export HADOOP_CONF_DIR=${HADOOP1_HOME}/conf

#Hadoop2.0
export HADOOP2_HOME=/opt/hadoop
export HADOOP_CONF_DIR=${HADOOP2_HOME}/etc/hadoop
export HADOOP_MAPRED_HOME=${HADOOP2_HOME}
export YARN_CONF_DIR=${HADOOP2_HOME}/etc/hadoop
export HADOOP_YARN_HOME=${HADOOP2_HOME}
export HADOOP_COMMON_HOME=${HADOOP2_HOME}
export HADOOP_HDFS_HOME=${HADOOP2_HOME}
export HDFS_CONF_DIR=${HADOOP2_HOME}/etc/hadoop
export PATH=$HADOOP2_HOME/bin:$HADOOP2_HOME/sbin:$PATH

#Ant
export ANT_HOME=/home/hadoop/apache-ant-1.8.0
export PATH=$ANT_HOME/bin:$PATH

#zookeeper
export ZOOKEEPER_HOME=/opt/zookeeper
export PATH=$ZOOKEEPER_HOME/bin:$PATH

7）同步环境变量

[hadoop@master ~]$ scp -r .bash_profile hadoop04:~   

[hadoop@master ~]$ scp -r .bash_profile slave1:~

[hadoop@master ~]$ scp -r .bash_profile slave2:~

8）在hadoop04、slave1、slave2创建目录

[hadoop@hadoop04 ~]$ mkdir -p /opt/zookeeper/data

[hadoop@master ~]$ slaves.sh mkdir -p /opt/ zookeeper /data    《===相当于在slave1和slave2节点执行了mkdir命令

小技巧：使用slaves.sh 可以简单的在slave节点同意执行命令

9）创建myid文件


[hadoop@hadoop04 ~]$ cd /opt/ zookeeper /data/

[hadoop@hadoop04 data]$ echo 1 > myid

[hadoop@slave1 ~]$ cd /opt/ zookeeper /data/

[hadoop@slave1 data]$ echo 2 > myid

[hadoop@slave2 hadoop]$ cd /opt/ zookeeper /data/

[hadoop@slave2 data]$ echo 3 > myid

10） 启动zookeeper

[hadoop@hadoop04 ~]$ zkServer.sh start

[hadoop@slave1 ~]$ zkServer.sh start

[hadoop@slave2 ~]$ zkServer.sh start

11）修改zoo.cfg，添加一条日志记录   《===只是为了报错有日志生成，可以不加

[hadoop@master conf]$ vi zoo.cfg

dataLogDir=/opt/zookeeper/log

12）停掉所有其他服务

On master

[hadoop@master hadoop]$ hadoop-daemons.sh --hostnames 'slave1 slave2 hadoop04' stop journalnode

[hadoop@master hadoop]$ hadoop-daemons.sh stop datanode

[hadoop@master hadoop]$ hadoop-daemons.sh --hostnames 'hadoop04 master'stop namenode

13）打开自动切换模式

[hadoop@master ~]$ cd /opt/hadoop/etc/hadoop/

[hadoop@master hadoop]$ vi hdfs-site.xml

<property>

  <name>dfs.ha.automatic-failover.enabled</name>
  <value>true</value>

 </property>

[hadoop@master hadoop]$ vi core-site.xml

<property>
  <name>ha.zookeeper.quorum</name>
  <value>hadoop04:2181,slave1:2181,slave2:2181</value>
 </property>


将这两个文件拷贝到hadoop04服务器

[hadoop@master hadoop]$ scp -r hdfs-site.xml hadoop04:/opt/hadoop/etc/hadoop/

[hadoop@master hadoop]$ scp -r core-site.xml hadoop04:/opt/hadoop/etc/hadoop/

14）Zookeeper的初始化

[hadoop@master hadoop]$ hdfs zkfc –formatZK

[hadoop@hadoop04 data]$ zkCli.sh

WatchedEvent state:SyncConnected type:None path:null

[zk: localhost:2181(CONNECTED) 0] ls /

[hadoop-ha, zookeeper]

[zk: localhost:2181(CONNECTED) 1] ls /hadoop-ha

[ns1]
[zk: localhost:2181(CONNECTED) 2] get /hadoop-ha/ns1

cZxid = 0x100000003
ctime = Tue Mar 21 02:40:13 EDT 2017
mZxid = 0x100000003
mtime = Tue Mar 21 02:40:13 EDT 2017
pZxid = 0x100000003
cversion = 0
dataVersion = 0
aclVersion = 0
ephemeralOwner = 0x0
dataLength = 0
numChildren = 0
[zk: localhost:2181(CONNECTED) 3]

15）启动所有服务

[hadoop@master hadoop]$ hadoop-daemons.sh --hostnames 'slave1 slave2 hadoop04' start journalnode

[hadoop@master hadoop]$ hadoop-daemons.sh --hostnames 'hadoop04 master' start namenode

[hadoop@master hadoop]$ hadoop-daemons.sh start datanode

16）在两个namenode节点启动zkfc

[hadoop@master hadoop]$ hadoop-daemons.sh --hostnames 'hadoop04 master' start zkfc

17）

on hadoop04

on master

18）模拟测试将当前active状态的namenode节点kill掉


[hadoop@master hadoop]$ jps
3436 Jps
2760 NameNode
2952 DFSZKFailoverController
[hadoop@master hadoop]$ kill -9 2760
[hadoop@master hadoop]$

19）

on hadoop04

on master

切换成功

20）创建目录

[hadoop@master hadoop]$ hdfs dfs -mkdir hdfs://ns1/test
[hadoop@master ~]$ hdfs dfs -put a.txt hdfs://ns1/test
[hadoop@master ~]$ hdfs dfs -ls hdfs://ns1/test
Found 1 items
-rw-r--r--   2 hadoop supergroup          6 2017-03-21 03:32 hdfs://ns1/test/a.txt
[hadoop@master ~]$

由于我们在core-site.xml配置了
<configuration>
 <property>
  <name>fs.defaultFS</name>
  <value>hdfs://ns1</value>
 </property>

因此，我们可以直接指定目录，不需要加hdfs://ns1这个前缀了

[hadoop@master ~]$ hdfs dfs -ls /test
Found 1 items
-rw-r--r--   2 hadoop supergroup          6 2017-03-21 03:32 /test/a.txt

基于zk自动切换，环境搭建完成