Cluster deployment of Apache HBase 1.2.12 under CentOS

1, Dependent environment

Pre dependency and installation

• CentOS 7. X, JDK, Hadoop 2.7.7 cluster, zookeeper-3.4.14 cluster
• The three node zookeeper is deployed this time. The corresponding machine hostname s are bigdata02, bigdata03 and bigdata04

The cluster planning is shown in the following table:

machine	Zookeeper	HMaster	HRegionServer
bigdata02	√	√
bigdata03	√		√
bigdata04	√	√	√

Download Apache HBase

• HBase official website: http://hbase.apache.org/ Find the official website and download the HBase installation package hbase-1.2.12-bin.tar.gz
• Recommended images for other websites: http://mirrors.hust.edu.cn/apache/hbase/
• Official documents of corresponding versions: http://hbase.apache.org/1.2/book.html
• If you want to download another version: http://archive.apache.org/dist/hbase/

2, Installing Apache Hbase

Create installation path

• Create folder hbase under / path

  [root@bigdata02 /]# mkdir hbase
  

• Put the downloaded hbase-1.2.12-bin.tar.gz into the HBase path and unzip it

  [root@bigdata02 hbase]# tar -zxvf hbase-1.2.12-bin.tar.gz
  

Modify profile

• Modify hbase-env.sh to enter the conf path under the decompression path

  [root@bigdata02 conf]# vim hbase-env.sh
  

• Modify the jdk path and the zk that is not applicable

  export JAVA_HOME=/usr/local/src/jdk8
  export HBASE_MANAGES_ZK=false
  

• Modify hbase-site.xml

  [root@bigdata02 conf]# vim hbase-site.xml 
  

• Add the following configuration:

  <property>
      <!-- appoint hbase Table data in HDFS Root path stored on -->
      <name>hbase.rootdir</name>
      <value>hdfs://bigdata02:9000/hbase</value>
  </property>
  <property>
      <!-- appoint hbase It is a distributed cluster mode -->
      <name>hbase.cluster.distributed</name>
      <value>true</value>
  </property>
  <property>
      <!-- appoint zookeeper Address, multiple","division -->
      <name>hbase.zookeeper.quorum</name>
      <value>bigdata02:2181,bigdata03:2181,bigdata04:2181</value>
  </property>
  

• Here, you need to create the root path of hbase table data stored on HDFS under the HDFS path

  [root@bigdata02 conf]# hdfs dfs -mkdir /hbase
  

• Modify regionservers

  [root@bigdata02 conf]# vim regionservers  
  # The contents are as follows:
  bigdata03
  bigdata04
  

• Create backup masters under conf

  [root@bigdata02 conf]# touch backup-masters
  #The contents are as follows: configure the backup master node of the cluster
  bigdata04
  

• Copy the core configuration files hdfs-site.xml and core-site.xml of hadoop cluster to the conf path

Distribute the installation package to each node

• Go back to the HBase path and distribute the installation package to each node. Note that the installation path should be the same; Distribute the hbase-1.2.12 folder to the other two nodes bigdata03 and bigdata04 (first establish the HBase file path at bigdata03 and bigdata04 nodes)

  [root@bigdata02 hbase] # scp -r hbase-1.2.12 root@bigdata03:/hbase
  [root@bigdata02 hbase] # scp -r hbase-1.2.12 root@bigdata04:/hbase
  

Synchronize server time

• HBase cluster has stricter time synchronization requirements than HDFS, and the difference is required to be no more than 30s

  sudo yum install -y ntpdate
  sudo ntpdate -u ntp.api.bz
  

Configure environment variables for Hbase

• Configure the environment variables of Hbase on the three machines

  # vim /etc/profile
  export HBASE_HOME=/hbase/hbase-1.2.12
  export PATH=$JAVA_HOME/bin:$PATH:$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$SCALA_HOME/bin:$SPARK_HOME/bin:$HIVE_HOME/bin:$ZOOKEEPER_HOME/bin:$HBASE_HOME/bin
  

• Make it effective

  # source /etc/profile
  

3, Start service

• Start the service in the following order: 1.Zookeeper cluster, 2.Hadoop cluster, 3.Hive cluster (if any) and 4.Hbase cluster;

• Start hbase here at the bigdata02 node

  [root@bigdata02 hbase] # start-hbase.sh
  

4, View startup

• Check whether the startup is normal and successful. jps each node in turn, as follows:

• Note: both the primary node and the standby node start the hmaster process; Each slave node starts the hregionserver process

• By accessing the browser page: http://bigdata02:16010

5, Turn off the Hbase service

• Execute under the bin path of the HMaster node:

  [root@bigdata04 bin]# ./hbase-daemon.sh stop master
  

• Execute under the bin path of the HRegionServer node:

  [root@bigdata04 bin]# ./hbase-daemon.sh stop regionserver RegionServer