1. HBase installation mode

　　①Single machine installation: does not depend on Hadoop HDFS, configuration is complete Ready to use, the advantage is that it is easy to test, but the disadvantage is that it does not have the ability to store distributed data.

　　②Pseudo-distributed installation: a single host simulates the real environment.

　　③Fully distributed installation: multiple hosts (virtual machines) to build

Second, preparations for building a fully distributed HBase cluster

< p>　　①Build Hadoop+JDK+ZooKeeper (3 zookeeper cluster modes, installed in other blog posts by the blogger< /span>)

　　②Prepare the HBase installation package

　　③Prepare three nodes:

　　　　192.168.144.130(The pseudo-distributed blogger of Hadoop has been installed in other blog posts, JDK+zookeeper+HBase)

　　　　192.168.144.132 (JDK+zookeeper+HBase)

　　　　192.168.144.134 (JDK+zookeeper+HBase)

3. Construction steps instructions

①Turn off the firewall and modify the hostname , Hosts file mapping, password-free login (3 nodes must do this)

Turn off the firewall

service iptables stop Temporary shutdown

　　chkconfig iptables off Permanent shutdown

Modify hostname

　　vim /etc/sysconfig/network　

NETWORKING=yes
HOSTNAME=hadoopalone //Here means the host name , and the names of the three nodes are configured as hadoopalone, hadoop02, and hadoop03

hosts file mapping

　　vim /etc/hosts

127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4

::1 localhost localhost.localdomain localhost6 localhost6.localdomain6

192.168.144.133 hadoopalone //Configure the corresponding ip and hostname mapping

192.168.144.131 hadoop02

192.168.144.132 hadoop03

Password-free login

　　　　#ssh-keygen, just press Enter all the way.

　　　　#ssh-copy-id [emailprotected]hadoopalone(Here select 3 in turn Hostnames of nodes)

②Get and decompress the HBase installation package

　　　　 download address: http://hbase.apache.org/downloads.html

p>

　　　　 I take 0.98.17 as an example here: tar -xvf　hbase-xxxxx　

③Modify the configuration file—hbase-env.sh

　　　　vim conf/hbase -env.sh (remember after the modification: source hbase-env.sh)

　　　　#modify JAVA_HOME=XXXX

< p>　　　　#Modify the coordination mode of zookeeper and hbase: HBase uses its own zookeeper by default. If you need to use an external zookeeper, you need to close it first: export HBASE_MANAGES_ZK=false

③Modify the configuration file—hbase-site .xml

　　　　Configure to enable fully distributed mode

 <property>

 <name>hbase.rootdirname>

 <value>hdfs://hadoopalone:9000/hbasevalue< /span>>

 property>

 <property>

 <name>hbase.cluster.distributedname< span style="color: #0000ff;">>

 <value>truevalue>

 property>

 

 <property >

 <name>hbase.zookeeper.quorumname< span style="color: #0000ff;">>

 <value>hadoopalone:2181,hadoop02:2181,hadoop03:2181 value>

property>

④Configure the region server

　　vim conf/regionservers

　　 each hostname has its own line

< p>　　Enter the following:

　　hadoopalone

　　hadoop02

　　hadoop03

⑤Remote copy the hbase installation package to the other two nodes

h3>

⑥Start zookeeper, Hadoop, Hbase in turn

Four. Formal establishment of HBase cluster

①Turn off the firewall, modify the host name, hosts File mapping, password-free login (3 nodes must do this)

Turn off the firewall

Edit host name

　　vim /etc/sysconfig/network (after the modification, remember to restart)

The host name of the first node is hadoopalone

　　The host name of the second node is hadoop02

　　 The host name of the third node is hadoop03

hosts file ip mapping

　　 select a node: vim /etc/hosts

Password-free login

　　　　①ssh-keygen, then press Enter all the way.

　　　　②ssh-copy-id [email protected], ssh-copy-id [email protected], ssh-copy-id [email protected]

　　　　 each node executes the above commands in turn :

②Upload the zookeeper and hbase installation packages to 3 nodes, and decompress them

　　Upload via rz command

　　Unzip: tar -xvf zookeeper.xxxxx tar -xvf hbase-. …..

　　 Each node has hbase and zookeeper

③Modify the configuration file—hbase-env.sh

　　 modify java_home

　　 modify hbase and Zookeeper coordination mode (modified: source hbase-env.sh takes effect)

③Modify configuration file—hbase -site.xml

④Configure region server

　　vim regionservers

⑤ Remotely copy the hbase installation package to the other two nodes

　　 make sure the jdk installation path is the same

⑥Start zookeeper, Hadoop, Hbase in turn

Start zookeeper

< p>　　　　sh zkServer.sh start start zookeeper, check the status through sh zkServer.sh status

　　　　 hadoopalone node zookeeper is: follower

　　　　hadoop02 node zookeeper is: leader

　　　　hadoop03 node zookeeper is: follower

Start the pseudo-distribution of Hadoop

　　　　 I have built the pseudo-distribution of hadoop in hadoopalone before , On the hadoopalone node: start-all.sh to start hadoop

　　　　 through jps to check whether the pseudo-distributed process of hadoop started successfully.

Start HBase

Enter the HBase bin directory of the Hadoopalone node: sh start-hbase.sh to start the server, through jps Check whether there is an HMaster process.

　　 On another node, through jps to view, the following process appears, indicating that the HBase cluster is built. It can be executed on this node: sh start-hbase.sh, it will become the slave of HMaster. The HMaster that is started first becomes the master.

So far, the HBase cluster has been set up , If there is a problem, we comment and discuss it, thank you.

NETWORKING=yes
HOSTNAME=hadoopalone //Here means the host name , and the names of the 3 nodes are configured as hadoopalone, hadoop02, hadoop03

127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4

::1 localhost localhost.localdomain localhost6 localhost6.localdomain6

192.168.144.133 hadoopalone //Configure the corresponding ip and hostname mapping

192.168.144.131 hadoop02

192.168.144.132 hadoop03

 <property>

 <name>hbase.rootdirname>

 <value>hdfs://hadoopalone:9000/hbasevalue< /span>>

 property>

 <property>

 <name>hbase.cluster.distributedname< span style="color: #0000ff;">>

 <value>truevalue>

 property>

 

 <property >

 <name>hbase.zookeeper.quorumname< span style="color: #0000ff;">>

 <value>hadoopalone:2181,hadoop02:2181,hadoop03:2181 value>

property>