HBase HA的分布式集群部署（适合3、5节点）

本博文的主要内容有：

　　.HBase的分布模式（3、5节点）安装

 　  .HBase的分布模式（3、5节点）的启动

　　.HBase HA的分布式集群的安装

　　.HBase HA的分布式集群的启动

 　  .HBase HA的切换

 

 

 

 

 

 HBase HA分布式集群搭建———集群架构

 

 

 

HBase HA分布式集群搭建———安装步骤

 

 

 

 

 

 

 

 

 

 

HBase的分布模式（3、5节点）安装

1、分别对djt11、djt12、djt13、djt14、djt15的启动进程恢复到没有任何启动进程的状态。

[hadoop@djt11 hadoop]$ pwd

[hadoop@djt11 hadoop]$ jps

 

 

[hadoop@djt12 hadoop]$ jps

 

[hadoop@djt13 hadoop]$ jps

 

[hadoop@djt14 hadoop]$ jps

 

[hadoop@djt15 hadoop]$ jps

 

 

 

2、切换到app安装目录

下载HBase压缩包

 

 

[hadoop@djt11 hadoop]$ pwd

[hadoop@djt11 hadoop]$ cd ..

[hadoop@djt11 app]$ pwd

[hadoop@djt11 app]$ ls

[hadoop@djt11 app]$ rz

[hadoop@djt11 app]$ ls

 

[hadoop@djt11]$ tar -zxvf hbase-0.98.19-hadoop2-bin.tar.gz

 

[hadoop@djt11 app]$ pwd

[hadoop@djt11 app]$ ls

[hadoop@djt11 app]$ mv hbase-0.98.19-hadoop2 hbase

[hadoop@djt11 app]$ rm -rf hbase-0.98.19-hadoop2-bin.tar.gz

[hadoop@djt11 app]$ ls

[hadoop@djt11 app]$ pwd

[hadoop@djt11 app]$

 

[hadoop@djt11 app]$ ls

[hadoop@djt11 app]$ cd hbase/

[hadoop@djt11 hbase]$ ls

[hadoop@djt11 hbase]$ cd conf/

[hadoop@djt11 conf]$ pwd

[hadoop@djt11 conf]$ ls

[hadoop@djt11 conf]$ vi regionservers

 

djt13

djt14

djt15

 

[hadoop@djt11 conf]$ vi backup-masters

 

djt12

 

[hadoop@djt11 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/core-site.xml ./
[hadoop@djt11 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/hdfs-site.xml ./

 

[hadoop@djt11 conf]$ vi hbase-site.xml

 

<configuration>

        <property>

                <name>hbase.zookeeper.quorum</name>

                <value>djt11,djt12,djt13,djt14,djt15</value>

        </property>

        <property>

                <name>hbase.zookeeper.property。dataDir</name>

                <value>/home/hadoop/data/zookeeper/zkdata</value>

        </property>

        <property>

                <name>hbase.zookeeper.property.clientPort</name>

                <value>2181</value>

        </property>

        <property>

                <name>hbase.rootdir</name>

                <value>hdfs://cluster1/hbase</value>

        </property>

        <property>

                <name>hbase.cluster.distributed</name>

                <value>true</value>

        </property>

        <property>

<name>hbase.tmp.dir</name>

<value>/home/hadoop/data/tmp/hbase</value>

         </property>

 

        <property>

                <name>hbase.master</name>

                <value>hdfs://djt11:60000</value>

        </property>

</configuration>

 

 

vi hbase-env.sh

 

#export JAVA_HOME=/usr/java/jdk1.6.0/

 

修改为，

export JAVA_HOME=/home/hadoop/app/jdk1.7.0_79

export HBASE_MANAGES_ZK=true

 

这里，有一个知识点。

进程HQuorumPeer，设HBASE_MANAGES_ZK=true，在启动HBase时，HBase把Zookeeper作为自身的一部分运行。

进程QuorumPeerMain，设HBASE_MANAGES_ZK=false，先手动启动Zookeeper，再启动HBase。

 

[hadoop@djt11 conf]$ pwd

[hadoop@djt11 conf]$ su root

[root@djt11 conf]# pwd

[root@djt11 conf]# vi /etc/profile

 

 

JAVA_HOME=/home/hadoop/app/jdk1.7.0_79

ZOOKEEPER_HOME=/home/hadoop/app/zookeeper

HADOOP_HOME=/home/hadoop/app/hadoop

HIVE_HOME=/home/hadoop/app/hive

CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar

PATH=$JAVA_HOME/bin:$ZOOKEEPER_HOME/bin:$HADOOP_HOME/bin:$HIVE_HOME/bin:/home/hadoop/tools:$PATH

export JAVA_HOME CLASSPATH PATH ZOOKEEPER_HOME HADOOP_HOME HIVE_HOME

 

 

JAVA_HOME=/home/hadoop/app/jdk1.7.0_79

ZOOKEEPER_HOME=/home/hadoop/app/zookeeper

HADOOP_HOME=/home/hadoop/app/hadoop

HIVE_HOME=/home/hadoop/app/hive

HBASE_HOME=/home/hadoop/app/hbase

CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar

PATH=$JAVA_HOME/bin:$ZOOKEEPER_HOME/bin:$HADOOP_HOME/bin:$HIVE_HOME/bin:$HBASE_HOME/bin:/home/hadoop/tools:$PATH

export JAVA_HOME CLASSPATH PATH ZOOKEEPER_HOME HADOOP_HOME HIVE_HOME HBASE_HOME

 

[root@djt11 conf]# source /etc/profile

[root@djt11 conf]# su hadoop

 

这个脚本，我们之前在搭建hadoop的5节点时，已经写好并可用的。这里，我们查看下并温习。不作修改

 

 

 

将djt11的hbase分发到slave，即djt12、djt13、djt14、djt15

[hadoop@djt11 tools]$ pwd

[hadoop@djt11 tools]$ cd /home/hadoop/app/

[hadoop@djt11 app]$ ls

[hadoop@djt11 app]$ deploy.sh hbase /home/hadoop/app/ slave

 

查看分发后的结果情况

表明，分发成功！

接下来，分别也跟djt11亿元，进行djt12、djt13、djt14、djt15的配置。

 

 

djt12的配置:

[hadoop@djt12 hbase]$ pwd

[hadoop@djt12 hbase]$ ls

[hadoop@djt12 hbase]$ cd conf/

[hadoop@djt12 conf]$ pwd

[hadoop@djt12 conf]$ ls

 

[hadoop@djt12 conf]$ vi regionservers

 

都是已经配置好了的

 

[hadoop@djt12 conf]$ vi backup-masters

 

都是之前已经配置好了的

 

[hadoop@djt12 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/core-site.xml ./
[hadoop@djt12 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/hdfs-site.xml ./ 

 

[hadoop@djt12 conf]$ vi hbase-site.xml

 

<configuration>

        <property>

                <name>hbase.zookeeper.quorum</name>

                <value>djt11,djt12,djt13,djt14,djt15</value>

        </property>

        <property>

                <name>hbase.zookeeper.property</name>

                <value>/home/hadoop/data/zookeeper/zkdata</value>

        </property>

        <property>

                <name>hbase.zookeeper.property.clientPort</name>

                <value>2181</value>

        </property>

        <property>

                <name>hbase.rootdir</name>

                <value>hdfs://cluster1/hbase</value>

        </property>

        <property>

                <name>hbase.cluster.distributed</name>

                <value>true</value>

        </property>

        <property>

                <name>hbase.master</name>

                <value>hdfs://djt11:60000</value>

        </property>

</configuration>

 

 

[hadoop@djt12 conf]$ vi hbase-env.sh

 

[hadoop@djt12 conf]$ pwd

[hadoop@djt12 conf]$ su root

[root@djt12 conf]# pwd

[root@djt12 conf]# vi /etc/profile

 

[root@djt12 conf]# cd ..

[root@djt12 hbase]# pwd

[root@djt12 hbase]# su hadoop

[hadoop@djt12 hbase]$ pwd

[hadoop@djt12 hbase]$ ls

[hadoop@djt12 hbase]$

 

 

djt13的配置

[hadoop@djt13 app]$ pwd

[hadoop@djt13 app]$ ls

[hadoop@djt13 app]$ cd hbase/

[hadoop@djt13 hbase]$ ls

[hadoop@djt13 hbase]$ cd conf/

[hadoop@djt13 conf]$ ls

[hadoop@djt13 conf]$ vi regionservers

 

[hadoop@djt13 conf]$ vi backup-masters

 

[hadoop@djt13 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/core-site.xml ./
[hadoop@djt13 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/hdfs-site.xml ./

 

[hadoop@djt13 conf]$ vi hbase-site.xml

 

<configuration>

        <property>

                <name>hbase.zookeeper.quorum</name>

                <value>djt11,djt12,djt13,djt14,djt15</value>

        </property>

        <property>

                <name>hbase.zookeeper.property</name>

                <value>/home/hadoop/data/zookeeper/zkdata</value>

        </property>

        <property>

                <name>hbase.zookeeper.property.clientPort</name>

                <value>2181</value>

        </property>

        <property>

                <name>hbase.rootdir</name>

                <value>hdfs://cluster1/hbase</value>

        </property>

        <property>

                <name>hbase.cluster.distributed</name>

                <value>true</value>

        </property>

        <property>

                <name>hbase.master</name>

                <value>hdfs://cluster1:60000</value>

        </property>

　　（注意，我的图片里是错误的！）  因为是做了高可用，是cluster1而不是单独的djt11。cluster1包括了djt11和djt12

</configuration>

 

[hadoop@djt13 conf]$ pwd

[hadoop@djt13 conf]$ su root

[root@djt13 conf]# pwd

[root@djt13 conf]# vi /etc/profile

 

JAVA_HOME=/home/hadoop/app/jdk1.7.0_79

ZOOKEEPER_HOME=/home/hadoop/app/zookeeper

HADOOP_HOME=/home/hadoop/app/hadoop

HIVE_HOME=/home/hadoop/app/hive

HBASE_HOME=/home/hadoop/app/hbase

CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar

PATH=$JAVA_HOME/bin:$ZOOKEEPER_HOME/bin:$HADOOP_HOME/bin:$HIVE_HOME/bin:$HBASE_HOME/bin:/home/hadoop/tools:$PATH

export JAVA_HOME CLASSPATH PATH ZOOKEEPER_HOME HADOOP_HOME HIVE_HOME HBASE_HOME

 

[hadoop@djt13 conf]$ pwd

[hadoop@djt13 conf]$ su root

[root@djt13 conf]# pwd

[root@djt13 conf]# vi /etc/profile

[root@djt13 conf]# source /etc/profile

[root@djt13 conf]# cd ..

[root@djt13 hbase]# pwd

[root@djt13 hbase]# su hadoop

[hadoop@djt13 hbase]$ pwd

[hadoop@djt13 hbase]$ ls

[hadoop@djt13 hbase]$

 

djt14的配置

[hadoop@djt14 app]$ pwd

[hadoop@djt14 app]$ ls

[hadoop@djt14 app]$ cd hbase/

[hadoop@djt14 hbase]$ ls

[hadoop@djt14 hbase]$ cd conf/

[hadoop@djt14 conf]$ ls

[hadoop@djt14 conf]$ vi regionservers

 

[hadoop@djt14 conf]$ vi backup-masters

 

[hadoop@djt14 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/core-site.xml ./
[hadoop@djt14 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/hdfs-site.xml ./ 

 

[hadoop@djt14 conf]$ vi hbase-site.xml

 

<configuration>

        <property>

                <name>hbase.zookeeper.quorum</name>

                <value>djt11,djt12,djt13,djt14,djt15</value>

        </property>

        <property>

                <name>hbase.zookeeper.property</name>

                <value>/home/hadoop/data/zookeeper/zkdata</value>

        </property>

        <property>

                <name>hbase.zookeeper.property.clientPort</name>

                <value>2181</value>

        </property>

        <property>

                <name>hbase.rootdir</name>

                <value>hdfs://cluster1/hbase</value>

        </property>

        <property>

                <name>hbase.cluster.distributed</name>

                <value>true</value>

        </property>

        <property>

                <name>hbase.master</name>

                <value>hdfs://djt11:60000</value>

        </property>

</configuration>

 

 

[hadoop@djt14 conf]$ vi hbase-env.sh

 

[hadoop@djt14 conf]$ pwd

[hadoop@djt14 conf]$ su root

[root@djt14 conf]# pwd

[root@djt14 conf]# vi /etc/profile

 

[hadoop@djt14 conf]$ pwd

[hadoop@djt14 conf]$ su root

[root@djt14 conf]# pwd

[root@djt14 conf]# vi /etc/profile

[root@djt14 conf]# source /etc/profile

[root@djt14 conf]# cd ..

[root@djt14 hbase]# pwd

[root@djt14 hbase]# su hadoop

[hadoop@djt14 hbase]$ pwd

[hadoop@djt14 hbase]$ ls

[hadoop@djt14 hbase]$

 

 

 

djt15的配置

 

[hadoop@djt15 app]$ pwd

[hadoop@djt15 app]$ ls

[hadoop@djt15 app]$ cd hbase/

[hadoop@djt15 hbase]$ ls

[hadoop@djt15 hbase]$ cd conf/

[hadoop@djt15 conf]$ ls

[hadoop@djt15 conf]$ vi regionservers

 

[hadoop@djt15 conf]$ vi backup-masters

 

[hadoop@djt15 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/core-site.xml ./
[hadoop@djt15 conf]$ cp /home/hadoop/app/hadoop/etc/hadoop/hdfs-site.xml ./ 

 

[hadoop@djt15 conf]$ vi hbase-site.xml

 

<configuration>

        <property>

                <name>hbase.zookeeper.quorum</name>

                <value>djt11,djt12,djt13,djt14,djt15</value>

        </property>

        <property>

                <name>hbase.zookeeper.property</name>

                <value>/home/hadoop/data/zookeeper/zkdata</value>

        </property>

        <property>

                <name>hbase.zookeeper.property.clientPort</name>

                <value>2181</value>

        </property>

        <property>

                <name>hbase.rootdir</name>

                <value>hdfs://cluster1/hbase</value>

        </property>

        <property>

                <name>hbase.cluster.distributed</name>

                <value>true</value>

        </property>

        <property>

                <name>hbase.master</name>

                <value>hdfs://djt11:60000</value>

        </property>

</configuration>

 

[hadoop@djt15 conf]$ vi hbase-env.sh

 

[hadoop@djt15 conf]$ pwd

[hadoop@djt15 conf]$ su root

[root@djt15 conf]# pwd

[root@djt15 conf]# vi /etc/profile

 

[hadoop@djt15 conf]$ pwd

[hadoop@djt15 conf]$ su root

[root@djt15 conf]# pwd

[root@djt15 conf]# vi /etc/profile

[root@djt15 conf]# source /etc/profile

[root@djt15 conf]# cd ..

[root@djt15 hbase]# pwd

[root@djt15 hbase]# su hadoop

[hadoop@djt15 hbase]$ pwd

[hadoop@djt15 hbase]$ ls

[hadoop@djt15 hbase]$

 

 

 　  .HBase的分布模式（3、5节点）的启动

 

 

 

这里，只需启动sbin/start-dfs.sh即可。

不需sbin/start-all.sh （它包括sbin/start-dfs.sh和sbin/start-yarn.sh）

启动zookeeper，是因为，hbase是建立在zookeeper之上的。数据是保存在hdfs。

 

 

[hadoop@djt11 app]$ jps

[hadoop@djt11 app]$ ls

[hadoop@djt11 app]$ cd hadoop/

[hadoop@djt11 hadoop]$ ls

[hadoop@djt11 hadoop]$ sbin/start-dfs.sh

 

[hadoop@djt11 hadoop]$ jps

 

[hadoop@djt12 hadoop]$ jps

 

[hadoop@djt13 app]$ cd hadoop/

[hadoop@djt13 hadoop]$ jps

 

[hadoop@djt14 app]$ cd hadoop/

[hadoop@djt14 hadoop]$ pwd

[hadoop@djt14 hadoop]$ jps

 

 

[hadoop@djt15 app]$ cd hadoop/

[hadoop@djt15 hadoop]$ pwd

[hadoop@djt15 hadoop]$ jps

 

[hadoop@djt11 hbase]$ bin/start-hbase.sh

[hadoop@djt11 hbase]$ jps

 

[hadoop@djt12 hbase]$ cd ..

[hadoop@djt12 app]$ ls

[hadoop@djt12 app]$ cd hbase/

[hadoop@djt12 hbase]$ pwd

[hadoop@djt12 hbase]$ jps

 

[hadoop@djt13 hbase]$ cd ..

[hadoop@djt13 app]$ ls

[hadoop@djt13 app]$ cd hbase/

[hadoop@djt13 hbase]$ pwd

[hadoop@djt13 hbase]$ jps

 

[hadoop@djt14 hbase]$ cd ..

[hadoop@djt14 app]$ ls

[hadoop@djt14 app]$ cd hbase/

[hadoop@djt14 hbase]$ pwd

[hadoop@djt14 hbase]$ jps

 

[hadoop@djt15 hbase]$ cd ..

[hadoop@djt15 app]$ ls

[hadoop@djt15 app]$ cd hbase/

[hadoop@djt15 hbase]$ pwd

[hadoop@djt15 hbase]$ jps

 

 

 

 

 

 

那么，djt11的master被杀死掉，则访问不到了。

 

 

 

 

然后，我们再把djt11的master启起来，

 

 

 

 

则，djt11由不可访问，变成备用的master了。djt12依然还是主用的master

 　　成功！