zoukankan      html  css  js  c++  java
  • Hbase-2.0.0_01_安装部署

    该文章是基于 Hadoop2.7.6_01_部署 进行的

    1. 主机规划

    主机名称

    IP信息

    内网IP

    操作系统

    安装软件

    备注:运行程序

    mini01

    10.0.0.11

    172.16.1.11

    CentOS 7.4

    jdk、hadoop、zookeeper、hbase

    QuorumPeerMain、NameNode、HMaster

    mini02

    10.0.0.12

    172.16.1.12

    CentOS 7.4

    jdk、hadoop、zookeeper、hbase

    QuorumPeerMain、ResourceManager、HMaster

    mini03

    10.0.0.13

    172.16.1.13

    CentOS 7.4

    jdk、hadoop、zookeeper、hbase

    QuorumPeerMain、DataNode、NodeManager、HRegionServer

    mini04

    10.0.0.14

    172.16.1.14

    CentOS 7.4

    jdk、hadoop、zookeeper、hbase

    QuorumPeerMain、DataNode、NodeManager、HRegionServer

    mini05

    10.0.0.15

    172.16.1.15

    CentOS 7.4

    jdk、hadoop、zookeeper、hbase

    QuorumPeerMain、DataNode、NodeManager、HRegionServer

    2. Zookeeper部署

           共部署5台,所以在mini01~mini05都得部署

    2.1. 配置信息

     1 [yun@mini01 conf]$ pwd
     2 /app/zookeeper/conf
     3 [yun@mini01 conf]$ cat zoo.cfg 
     4 #单个客户端与单台服务器之间的连接数的限制,是ip级别的,默认是60,如果设置为0,那么表明不作任何限制。
     5 maxClientCnxns=1500
     6 # The number of milliseconds of each tick
     7 tickTime=2000
     8 # The number of ticks that the initial 
     9 # synchronization phase can take
    10 initLimit=10
    11 # The number of ticks that can pass between 
    12 # sending a request and getting an acknowledgement
    13 syncLimit=5
    14 # the directory where the snapshot is stored.
    15 # do not use /tmp for storage, /tmp here is just 
    16 # example sakes.
    17 # dataDir=/tmp/zookeeper
    18 dataDir=/app/bigdata/zookeeper/data
    19 # the port at which the clients will connect
    20 clientPort=2181
    21 #
    22 # Be sure to read the maintenance section of the 
    23 # administrator guide before turning on autopurge.
    24 #
    25 # http://zookeeper.apache.org/doc/current/zookeeperAdmin.html#sc_maintenance
    26 #
    27 # The number of snapshots to retain in dataDir
    28 #autopurge.snapRetainCount=3
    29 # Purge task interval in hours
    30 # Set to "0" to disable auto purge feature
    31 #autopurge.purgeInterval=1
    32 
    33 # leader和follow通信端口和投票选举端口
    34 server.1=mini01:2888:3888
    35 server.2=mini02:2888:3888
    36 server.3=mini03:2888:3888
    37 server.4=mini04:2888:3888
    38 server.5=mini05:2888:3888

    2.2. 添加myid文件

    1 [yun@mini01 data]$ pwd
    2 /app/bigdata/zookeeper/data
    3 [yun@mini01 data]$ ll
    4 total 4
    5 -rw-r--r-- 1 yun yun   2 May 26 14:36 myid
    6 drwxr-xr-x 2 yun yun 232 Jun 27 14:54 version-2
    7 [yun@mini01 data]$ cat myid  # 其中mini01的myid 为1;mini02的myid 为2;mini03的myid 为3;mini04的myid 为4;mini05的myid 为5  
    8 1

    2.3. 环境变量

    1 [root@mini01 profile.d]# pwd
    2 /etc/profile.d
    3 [root@mini01 profile.d]# cat zk.sh 
    4 export ZK_HOME="/app/zookeeper"
    5 export PATH=$ZK_HOME/bin:$PATH
    6 
    7 [root@mini01 profile.d]# logout
    8 [yun@mini01 conf]$ source /etc/profile # 重新加载环境变量  

    2.4. 启动zk服务

    1 # 依次在启动mini01、mini02、mini03、mini04、mini05  zk服务
    2 [yun@mini01 zookeeper]$ pwd
    3 /app/zookeeper
    4 [yun@mini01 zookeeper]$ zkServer.sh start
    5 JMX enabled by default
    6 Using config: /app/zookeeper/bin/../conf/zoo.cfg
    7 Starting zookeeper ... STARTED

           建议在 /app/zookeeper,因为在启动时,会有相关日志产生

    1 [yun@mini01 zookeeper]$ pwd
    2 /app/zookeeper
    3 [yun@mini01 zookeeper]$ ll zookeeper.out 
    4 -rw-rw-r-- 1 yun yun 41981 Aug  6 17:24 zookeeper.out

    2.5. 查询运行状态

     1 # 其中mini01、mini02、mini04、mini05状态如下
     2 [yun@mini01 zookeeper]$ zkServer.sh status
     3 JMX enabled by default
     4 Using config: /app/zookeeper/bin/../conf/zoo.cfg
     5 Mode: follower
     6 
     7 # 其中mini03 状态如下
     8 [yun@mini03 zookeeper]$ zkServer.sh status
     9 JMX enabled by default
    10 Using config: /app/zookeeper/bin/../conf/zoo.cfg
    11 Mode: leader

    3. Hbase部署与配置修改

    3.1. 软件部署

    1 [yun@mini01 software]$ pwd
    2 /app/software
    3 [yun@mini01 software]$ tar xf hbase-2.0.0-bin.tar.gz 
    4 [yun@mini01 software]$ mv hbase-2.0.0 /app/
    5 [yun@mini01 software]$ cd 
    6 [yun@mini01 ~]$ ln -s hbase-2.0.0/ hbase

    3.2. 环境变量

           注意所有部署hbase的机器【mini01、mini02、mini03、mini04、mini05】都需要该环境变量

    1 [root@mini01 profile.d]# pwd
    2 /etc/profile.d
    3 [root@mini01 profile.d]# cat hbase.sh  # 也可以直接写在 /etc/profile 文件中  
    4 export HBASE_HOME="/app/hbase"
    5 export PATH=$HBASE_HOME/bin:$PATH
    6 
    7 [root@mini01 profile.d]# logout
    8 [yun@mini01 hbase]$ source /etc/profile  # 使用yun用户,并重新加载环境变量 

    3.3. hbase-env.sh 修改

     1 [yun@mini01 conf]$ pwd
     2 /app/hbase/conf
     3 [yun@mini01 conf]$ cat hbase-env.sh 
     4 #!/usr/bin/env bash
     5 ………………
     6 # The java implementation to use.  Java 1.8+ required.
     7 # export JAVA_HOME=/usr/java/jdk1.8.0/
     8 export JAVA_HOME=${JAVA_HOME}
     9 
    10 # Extra Java CLASSPATH elements.  Optional.  # hadoop配置文件的位置
    11 # export HBASE_CLASSPATH=
    12 export HBASE_CLASSPATH=${HADOOP_HOME}/etc/hadoop/
    13 ………………
    14 # Tell HBase whether it should manage it's own instance of ZooKeeper or not.
    15 # 如果使用独立安装的zookeeper这个地方就是false
    16 # export HBASE_MANAGES_ZK=true
    17 export HBASE_MANAGES_ZK=false
    18 ………………

    3.4. hbase-site.xml 修改

     1 [yun@mini01 conf]$ pwd
     2 /app/hbase/conf
     3 [yun@mini01 conf]$ cat hbase-site.xml
     4 <?xml version="1.0"?>
     5 <?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
     6 <!--
     7 /**
     8  *
     9 ………………
    10  */
    11 -->
    12 <configuration>
    13   <property>
    14     <name> hbase.master.port</name>  <!-- hbasemaster的端口 -->
    15     <value>16000</value>
    16   </property>
    17   <property>
    18     <name>hbase.tmp.dir</name>        <!-- hbase 临时存储 -->
    19     <value>/app/hbase/tmp</value>
    20   </property>  
    21   <property>
    22     <name>hbase.master.maxclockskew</name>   <!-- 时间同步允许的时间差 单位毫秒 -->
    23     <value>180000</value>
    24   </property>
    25   <property>
    26     <name>hbase.rootdir</name>
    27     <value>hdfs://mini01:9000/hbase</value>  <!-- hbase共享目录,持久化hbase数据  存放在对应的HDFS上 -->
    28   </property>
    29   <property>
    30     <name>hbase.cluster.distributed</name>  <!-- 是否分布式运行,false即为单机 -->
    31     <value>true</value>
    32   </property>
    33   <property>  
    34     <name>hbase.zookeeper.property.clientPort</name> <!-- zookeeper端口 -->
    35     <value>2181</value>  
    36   </property> 
    37   <property>
    38     <name>hbase.zookeeper.quorum</name>  <!-- zookeeper地址 -->
    39     <value>mini01,mini02,mini03,mini04,mini05</value>
    40   </property>
    41   <property>
    42     <name>hbase.zookeeper.property.dataDir</name>  <!-- zookeeper配置信息快照的位置 -->
    43     <value>/app/hbase/tmp/zookeeper</value>
    44   </property>
    45 </configuration>

    3.5. regionservers 修改

    1 [yun@mini01 conf]$ pwd
    2 /app/hbase/conf
    3 [yun@mini01 conf]$ cat regionservers  # 从机器的域名 
    4 mini03
    5 mini04
    6 mini05

    4. Hbase的分发与启动

           注意:启动hbase之前,必须保证hadoop集群和zookeeper集群是可用的。

    4.1. hbase分发到其他机器

           将 /app/hbase-2.0.0 从mini01 分发到mini02【用于HA】、mini03、mini04、mini05

           其中配置不需要任何修改

    1 scp hbase-2.0.0 yun@mini02:/app
    2 scp hbase-2.0.0 yun@mini03:/app
    3 scp hbase-2.0.0 yun@mini04:/app
    4 scp hbase-2.0.0 yun@mini05:/app

           分发完毕后,记得登录不同的主机然后创建软连接

    1 [yun@mini02 ~]$ pwd
    2 /app
    3 [yun@mini02 ~]$ ln -s hbase-2.0.0/ hbase

    4.2. 启动程序

     1 [yun@mini01 ~]$ start-hbase.sh 
     2 SLF4J: Class path contains multiple SLF4J bindings.
     3 SLF4J: Found binding in [jar:file:/app/hbase-2.0.0/lib/slf4j-log4j12-1.7.25.jar!/org/slf4j/impl/StaticLoggerBinder.class]
     4 SLF4J: Found binding in [jar:file:/app/hadoop-2.7.6/share/hadoop/common/lib/slf4j-log4j12-1.7.10.jar!/org/slf4j/impl/StaticLoggerBinder.class]
     5 SLF4J: See http://www.slf4j.org/codes.html#multiple_bindings for an explanation.
     6 SLF4J: Actual binding is of type [org.slf4j.impl.Log4jLoggerFactory]
     7 running master, logging to /app/hbase/logs/hbase-yun-master-mini01.out
     8 SLF4J: Class path contains multiple SLF4J bindings.
     9 SLF4J: Found binding in [jar:file:/app/hbase-2.0.0/lib/slf4j-log4j12-1.7.25.jar!/org/slf4j/impl/StaticLoggerBinder.class]
    10 SLF4J: Found binding in [jar:file:/app/hadoop-2.7.6/share/hadoop/common/lib/slf4j-log4j12-1.7.10.jar!/org/slf4j/impl/StaticLoggerBinder.class]
    11 SLF4J: See http://www.slf4j.org/codes.html#multiple_bindings for an explanation.
    12 SLF4J: Actual binding is of type [org.slf4j.impl.Log4jLoggerFactory]
    13 mini05: running regionserver, logging to /app/hbase/bin/../logs/hbase-yun-regionserver-mini05.out
    14 mini04: running regionserver, logging to /app/hbase/bin/../logs/hbase-yun-regionserver-mini04.out
    15 mini03: running regionserver, logging to /app/hbase/bin/../logs/hbase-yun-regionserver-mini03.out

    master进程

    1 [yun@mini01 ~]$ jps 
    2 1808 SecondaryNameNode
    3 1592 NameNode
    4 6540 HMaster
    5 4158 QuorumPeerMain
    6 7998 Jps

    slave进程

    1 [yun@mini04 ~]$ jps 
    2 2960 Jps
    3 1921 QuorumPeerMain
    4 1477 NodeManager
    5 1547 DataNode
    6 2333 HRegionServer

    4.3. zk中的信息

    1 [zk: localhost:2181(CONNECTED) 9] ls /hbase
    2 [replication, meta-region-server, rs, splitWAL, backup-masters, table-lock, flush-table-proc, master-maintenance, online-snapshot, switch, master, running, draining, namespace, hbaseid, table]

    1 [zk: localhost:2181(CONNECTED) 7] ls /hbase/rs
    2 [mini03,16020,1533633829928, mini05,16020,1533633829675, mini04,16020,1533633829711]

    4.4. 浏览器访问

    1 http://mini01:16010    

    5. Hbase的HA

           根据规划,mini01和mini02为Hmaster,其中mini01的Hmaster已经起来了

    5.1. 启动另一个HMaster

     1 [yun@mini02 ~]$ hbase-daemon.sh start master  # 在mini02起一个 HMaster 
     2 running master, logging to /app/hbase/logs/hbase-yun-master-mini02.out
     3 SLF4J: Class path contains multiple SLF4J bindings.
     4 SLF4J: Found binding in [jar:file:/app/hbase-2.0.0/lib/slf4j-log4j12-1.7.25.jar!/org/slf4j/impl/StaticLoggerBinder.class]
     5 SLF4J: Found binding in [jar:file:/app/hadoop-2.7.6/share/hadoop/common/lib/slf4j-log4j12-1.7.10.jar!/org/slf4j/impl/StaticLoggerBinder.class]
     6 SLF4J: See http://www.slf4j.org/codes.html#multiple_bindings for an explanation.
     7 SLF4J: Actual binding is of type [org.slf4j.impl.Log4jLoggerFactory]
     8 [yun@mini02 ~]$ 
     9 [yun@mini02 ~]$ jps 
    10 3746 Jps
    11 1735 ResourceManager
    12 2265 QuorumPeerMain
    13 3643 HMaster

    5.2. zk中的信息

    1 [zk: localhost:2181(CONNECTED) 3] ls /hbase/backup-masters
    2 [mini02,16000,1533635137064]

    5.3. 浏览器访问

    1 http://mini02:16010    

  • 相关阅读:
    override new virtual 的比较
    c#页面无内容解决方案
    插入排序
    排序算法(转)
    treenode遍历文件夹
    案例篇(1)
    索引器(转)
    迭代器的实现
    抽象类和接口的区别
    索引器与迭代器,属性的区别
  • 原文地址:https://www.cnblogs.com/zhanglianghhh/p/9491335.html
Copyright © 2011-2022 走看看