Hadoop2.6.4、zookeeper3.4.6、HBase1.2.2、Hive1.2.1、sqoop1.99.7、spark1.6.2安装

一、准备

1. 安装虚拟机与编译Hadoop

 

创建3个虚拟机，分别为hsm01, hss01, hss02

hostname	ip
hsm01	192.168.99.145
hss01	192.168.99.151
hss02	192.168.99.152

2. 配置服务器

2.1 关闭防火墙

# 执行命令
service iptables stop
# 验证
service iptables status
# 关闭防火墙的自动运行
chkconfig iptables off
# 验证
chkconfig --list | grep iptables

2.2 设置主机名

hostname hss01
vim /etc/sysconfig/network

# ip 与 hostname 绑定
vim /etc/hosts

2.3 免密码登录

# 设置 ssh 免密码登录（在三个节点分别执行以下命令）
ssh-keygen -t rsa
# ~/.ssh/id_rsa.pub就是生成的公钥，把三个id_rsa.pub的内容合并，写入以下文件
cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys
# 复制到其他节点
scp ~/.ssh/authorized_keys zkpk@hss01:~/.ssh/
scp ~/.ssh/authorized_keys zkpk@hss02:~/.ssh/

3. 安装JDK

# root用户（也可以其他用户安装）
vim /etc/profile

export JAVA_HOME=/opt/jdk1.8.0_45
export PATH=$PATH:$JAVA_HOME/bin
export CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar

source /etc/profile

4. 版本

程序	版本
JDK	1.8.0_45
Hadoop	2.6.4
zookeeper	3.4.6
hbase	1.2.2
hive	1.2.1
mysql	5.7.14
sqoop	1.99.7
spark	1.6.2

 

5. 规划

节点	安装软件	进程
hsm01	jdk, hadoop, zookeeper, hbase, hive, sqoop, spark	NameNode, ResourceManager, JournalNode, QuorumPeerMain, DFSZKFailoverController, HMaster, Worker, Master
hss01	jdk, hadoop, zookeeper, hbase, spark	NameNode, ResourceManager(需单独启动), JournalNode, QuorumPeerMain, DataNode, NodeManager, DFSZKFailoverController, Worker
hss02	jdk, hadoop, zookeeper, hbase, mysql, spark	DataNode, NodeManager, JournalNode, QuorumPeerMain, Worker

二、安装

hadoop相关程序都是用zkpk用户进行操作，并安装在/home/zkpk目录下

1. zookeeper

1.1 解压

tar -xf zookeeper-3.4.6.tar.gz

1.2 配置

cd ~/zookeeper-3.4.6/conf
cp zoo_sample.cfg zoo.cfg
vim zoo.cfg

# 修改
dataDir=/home/zkpk/zookeeper-3.4.6/data
# 添加
dataLogDir=/home/zkpk/zookeeper-3.4.6/logs

# 在最后添加
server.1=hsm01:2888:3888
server.2=hss01:2888:3888
server.3=hss02:2888:3888

1.3 创建目录与myid文件

# zookeeper根目录执行
mkdir data
mkdir logs

# 在dataDir目录下创建myid文件写入1
vim data/myid

1.4 复制ZooKeeper到其他节点

scp -r ~/zookeeper-3.4.6/ zkpk@hss01:~/
scp -r ~/zookeeper-3.4.6/ zkpk@hss02:~/

# 将hss01中的myid改为2，hss02中的myid改为3
vim ~/zookeeper-3.4.6/data/myid

1.5 配置环境变量

vim ~/.bash_profile

export ZOOKEEPER_HOME=/home/zkpk/zookeeper-3.4.6
export PATH=$PATH:$ZOOKEEPER_HOME/bin

source ~/.bash_profile

1.6 逐个启动验证

zkServer.sh start
zkServer.sh status

1.7 问题

zookeeper环境搭建中的几个坑[Error contacting service. It is probably not running]的分析及解决
http://www.paymoon.com/index.php/2015/06/04/zookeeper-building/

安装zookeeper时候，可以查看进程启动，但是状态显示报错：Error contacting service. It is probably not running
http://www.cnblogs.com/xiaohua92/p/5460515.html

所有节点的系统时间要同步

# root用户

date -s "yyyyMMdd HH:mm:ss"
clock -w

Zookeeper 日志输出到指定文件夹
http://www.tuicool.com/articles/MbUb63n

2. Hadoop

2.1 解压(/home/zkpk)

tar -xf hadoop-2.6.4.tar.gz

2.2 创建相应目录

cd hadoop-2.6.4
# namenode信息存放目录
mkdir name
# datanode信息存放目录
mkdir data

2.3 修改JAVA_HOME

cd etc/hadoop
vim yarn-env.sh
vim hadoop-env.sh
vim mapred-env.sh

export JAVA_HOME=/opt/jdk1.8.0_45

2.4 配置core-site.xml

vim core-site.xml


  
     fs.defaultFS
    hdfs://ns1
  
  
     hadoop.tmp.dir
     /home/zkpk/hadoop-2.6.4/tmp
  
  
     ha.zookeeper.quorum
     hsm01:2181,hss01:2181,hss02:2181
  

注：不要忘了创建tmp目录

2.5 配置hdfs-site.xml

vim hdfs-site.xml


  
  
    dfs.replication
    1
  
  
  
    dfs.permissions
    false
  
  
  
    dfs.namenode.name.dir
    /home/zkpk/hadoop-2.6.4/name
    true
  
  
  
    dfs.datanode.data.dir
    /home/zkpk/hadoop-2.6.4/data
    true
  
  
    dfs.nameservices
    ns1
  
  
    dfs.ha.namenodes.ns1
    nn1,nn2
  
  
    dfs.namenode.rpc-address.ns1.nn1
    hsm01:9000
  
  
    dfs.namenode.http-address.ns1.nn1
    hsm01:50070
  
  
    dfs.namenode.rpc-address.ns1.nn2
    hss01:9000
  
  
    dfs.namenode.http-address.ns1.nn2
    hss01:50070
  
  
    dfs.namenode.shared.edits.dir
    qjournal://hsm01:8485;hss01:8485;hss02:8485/ns1
  
  
    dfs.journalnode.edits.dir
    /home/zkpk/hadoop-2.6.4/journal
  
  
    dfs.ha.automatic-failover.enabled
    true
  
  
    dfs.client.failover.proxy.provider.ns1
    org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider
  
  
    dfs.ha.fencing.methods
    
          sshfence
          shell(/bin/true)
    
  
  
    dfs.ha.fencing.ssh.private-key-files
    /home/zkpk/.ssh/id_rsa
  
  
    dfs.ha.fencing.ssh.connect-timeout
    30000
  

2.6 编辑mapred-site.xml

cp mapred-site.xml.template mapred-site.xml
vim mapred-site.xml


  
    mapreduce.framework.name
    yarn
  

2.7 编辑yarn-site.xml

vim yarn-site.xml


  
    yarn.nodemanager.aux-services
    mapreduce_shuffle
  
  
    yarn.nodemanager.aux-services.mapreduce.shuffle.class
    org.apache.hadoop.mapred.ShuffleHandler
  
  
    yarn.resourcemanager.ha.enabled
    true
  
  
    yarn.resourcemanager.ha.automatic-failover.enabled
    true
  
  
    yarn.resourcemanager.ha.id
    rm1
  
  
    yarn.resourcemanager.cluster-id
    yrc
  
  
    yarn.resourcemanager.ha.rm-ids
    rm1,rm2
  
  
    yarn.resourcemanager.hostname.rm1
    hsm01
  
  
    yarn.resourcemanager.hostname.rm2
    hss01
  
  
    yarn.resourcemanager.zk-address
    hsm01:2181,hss01:2181,hss02:2181
  
  
    yarn.resourcemanager.recovery.enabled
    true
  
  
    yarn.resourcemanager.store.class
    org.apache.hadoop.yarn.server.resourcemanager.recovery.ZKRMStateStore
  

2.8 编辑slaves

vim slaves

hss01
hss02

2.9 复制到其他节点

scp -r ~/hadoop-2.6.4 hss01:~/
scp -r ~/hadoop-2.6.4 hss02:~/

2.10 配置各节点环境变量

打开：
vim ~/.bash_profile
添加：
export HADOOP_HOME=/home/zkpk/hadoop-2.6.4
export PATH=$PATH:$HADOOP_HOME/bin:$HADOOP_HOME/sbin
刷新：
source ~/.bash_profile
验证：（输入以下命令，如果出现hadoop对应的版本，则hadoop配置成功。）
hadoop version

2.11 集群启动（严格按照下面的步骤）

a. 启动zookeeper集群（分别在hsm01、hss01、hss02上启动zk）

zkServer.sh start
# 查看状态：一个leader，两个follower
zkServer.sh status

b. 启动journalnode（分别在hsm01、hss01、hss02上启动journalnode）

hadoop-daemon.sh start journalnode

# 运行jps命令检验，hsm01、hss01、hss02上多了JournalNode进程

c. 格式化HDFS

# hsm01上执行
hdfs namenode -format

d. 将tmp拷到其他节点

scp -r ~/hadoop-2.6.4/name hss01:~/hadoop-2.6.4/
scp -r ~/hadoop-2.6.4/name hss02:~/hadoop-2.6.4/

e. 格式化ZK

# hsm01上执行
hdfs zkfc -formatZK

f. 启动HDFS

# 启动
zkServer.sh start
start-dfs.sh
start-yarn.sh

# 关闭
stop-dfs.sh
stop-yarn.sh
zkServer.sh stop

2.13 问题

待续

3. Hive安装

3.1 MySQL安装

http://blog.csdn.net/u013980127/article/details/52261400

# 创建hadoop用户
grant all on *.* to hadoop@'%' identified by 'hadoop';
grant all on *.* to hadoop@'localhost' identified by 'hadoop';
grant all on *.* to hadoop@'hsm01' identified by 'hadoop';
flush privileges;

# 创建数据库
create database hive_121;

3.2 解压

tar -xf apache-hive-1.2.1-bin.tar.gz

# 文件名修改为hive-1.2.1
mv apache-hive-1.2.1-bin/ hive-1.2.1

3.3 修改文件名

# 在hive-1.2.1/conf下，修改文件名
mv hive-default.xml.template hive-site.xml
mv hive-log4j.properties.template hive-log4j.properties
mv hive-exec-log4j.properties.template hive-exec-log4j.properties
mv hive-env.sh.template hive-env.sh

3.4 hive-env.sh

export HADOOP_HOME=/home/zkpk/hadoop-2.6.4
export HIVE_CONF_DIR=/home/zkpk/hive-1.2.1/conf

3.5 hive-log4j.properties

hive.log.dir=/home/zkpk/hive-1.2.1/logs

# 创建日志目录
mkdir /home/zkpk/hive-1.2.1/logs

3.6 hive-site.xml

删除所有内容，添加如下内容：

  
    hive.metastore.warehouse.dir
    hdfs://ns1/hive/warehouse
  
  
    hive.exec.scratchdir
    hdfs://ns1/hive/scratchdir
  
  
    hive.querylog.location
    /home/zkpk/hive-1.2.1/logs
  
  
    javax.jdo.option.ConnectionURL
    jdbc:mysql://hss02:3306/hive_121?characterEncoding=UTF-8
  
  
    javax.jdo.option.ConnectionDriverName
    com.mysql.jdbc.Driver
  
  
    javax.jdo.option.ConnectionUserName
    hadoop
  
  
    javax.jdo.option.ConnectionPassword
    hadoop
  

3.7 环境变量

vim ~/.bash_profile

export HIVE_HOME=/home/zkpk/hive-1.2.1
export PATH=$PATH:$HIVE_HOME/bin

source ~/.bash_profile

在hive/lib下有个jline的jar，将hadoop内的这个jar包换成一致的，否则会启动hive会报错。

将mysql-connector-java-5.1.29.jar连接jar拷贝到hive-1.2.1/lib目录下

# 运行下面命令
hive

# http://hsm01:50070，查看是否多了hive目录。

3.8 问题与参考

Hive配置项的含义详解

Hive 使用陷阱（Lock table） 排查过程

Hive、Spark SQL、Impala比较

4. Sqoop安装

4.1 解压

tar -xf sqoop-1.99.7-bin-hadoop200.tar.gz

# 修改目录名
mv sqoop-1.99.7-bin-hadoop200/ sqoop-1.99.7

4.2 配置Hadoop代理访问

# 配置代理
vim $HADOOP_HOME/etc/hadoop/core-site.xml

# zkpk是运行server的用户

  hadoop.proxyuser.zkpk.hosts
  *


  hadoop.proxyuser.zkpk.groups
  *


# 由于用户id小于1000（可用id命令查看），设置此项
vim $HADOOP_HOME/etc/hadoop/container-executor.cfg

allowed.system.users=zkpk

4.3 sqoop.properties

# @LOGDIR@修改为/home/zkpk/sqoop-1.99.7/logs
# @BASEDIR@修改为/home/zkpk/sqoop-1.99.7

# hadoop配置文件路径
org.apache.sqoop.submission.engine.mapreduce.configuration.directory=/home/zkpk/hadoop-2.6.4/etc/hadoop/

# 设置验证机制（去掉注释）
org.apache.sqoop.security.authentication.type=SIMPLE
org.apache.sqoop.security.authentication.handler=org.apache.sqoop.security.authentication.SimpleAuthenticationHandler
org.apache.sqoop.security.authentication.anonymous=true

4.4 配置第三方jar引用路径

复制mysql驱动jar文件到$SQOOP_HOME/extra（创建extra目录）

export SQOOP_SERVER_EXTRA_LIB=$SQOOP_HOME/extra

4.5 环境变量

vim ~/.bash_profile

export SQOOP_HOME=/home/zkpk/sqoop-1.99.7
export SQOOP_SERVER_EXTRA_LIB=$SQOOP_HOME/extra
export PATH=$PATH:$SQOOP_HOME/bin

source ~/.bash_profile

4.6 启动验证

# 验证配置是否有效
sqoop2-tool verify

# 开启服务器
sqoop2-server start

# 客户端验证
sqoop2-shell

show connector

4. 问题与参考

Sqoop1.99.7安装、配置和使用（一）

Sqoop1.99.7安装、配置和使用（二）

Sqoop2的安装与使用

Sqoop1.X 和 Sqoop2架构区别

Hadoop数据收集与入库系统Flume与Sqoop

5. HBase安装

5.1 解压

tar -xf hbase-1.2.2-bin.tar.gz

5.2 lib更新

cd hbase-1.2.2/lib

cp ~/hadoop-2.6.4/share/hadoop/mapreduce/lib/hadoop-annotations-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/tools/lib/hadoop-auth-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/common/hadoop-common-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/hdfs/hadoop-hdfs-2.6.4.jar .

cp ~/hadoop-2.6.4/share/hadoop/mapreduce/hadoop-mapreduce-client-app-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/mapreduce/hadoop-mapreduce-client-common-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/mapreduce/hadoop-mapreduce-client-core-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/mapreduce/hadoop-mapreduce-client-jobclient-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/mapreduce/hadoop-mapreduce-client-shuffle-2.6.4.jar .

cp ~/hadoop-2.6.4/share/hadoop/yarn/hadoop-yarn-api-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/yarn/hadoop-yarn-client-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/yarn/hadoop-yarn-common-2.6.4.jar .
cp ~/hadoop-2.6.4/share/hadoop/yarn/hadoop-yarn-server-common-2.6.4.jar .

# 解决java.lang.NoClassDefFoundError: org/htrace/Trace
cp ~/hadoop-2.6.4/share/hadoop/common/lib/htrace-core-3.0.4.jar .

# 删除老版的jar
rm *-2.5.1.jar

5.2 hbase-env.sh

export JAVA_HOME=/opt/jdk1.8.0_45
export HBASE_MANAGES_ZK=false
export HBASE_CLASSPATH=/home/zkpk/hadoop-2.6.4/etc/hadoop

# 注释掉下面的配置（因为1.8JDK没有这个选项）
#export HBASE_MASTER_OPTS="$HBASE_MASTER_OPTS -XX:PermSize=128m -XX:MaxPermSize=128m"
#export HBASE_REGIONSERVER_OPTS="$HBASE_REGIONSERVER_OPTS -XX:PermSize=128m -XX:MaxPermSize=128m"

5.3 hbase-site.xml

  
    hbase.cluster.distributed
    true
  
  
    hbase.tmp.dir
    /home/zkpk/hbase-1.2.2/tmp
  
  
    hbase.rootdir
    hdfs://ns1/hbase
  
  
    zookeeper.session.timeout
    120000
  
  
    hbase.zookeeper.property.tickTime
    6000
  
  
    hbase.zookeeper.property.clientPort
    2181
  
  
    hbase.zookeeper.quorum
    hsm01,hss01,hss02
  
  
    hbase.zookeeper.property.dataDir 
    /home/zkpk/zookeeper-3.4.6/data
  
  
    dfs.replication
    1
  
   
    hbase.master.maxclockskew 
    180000
   

5.4 regionservers

hss01
hss02

5.5 拷贝hbase到其他节点

把hadoop的hdfs-site.xml和core-site.xml 放到hbase/conf下

cp hadoop-2.6.4/etc/hadoop/hdfs-site.xml hbase-1.2.2/conf/
cp hadoop-2.6.4/etc/hadoop/core-site.xml hbase-1.2.2/conf/

scp -r /home/zkpk/hbase-1.2.2  hss01:~/
scp -r /home/zkpk/hbase-1.2.2  hss02:~/

5.6 配置环境变量

# 各节点分别配置
vim ~/.bash_profile
export HBASE_HOME=/home/zkpk/hbase-1.2.2
export PATH=$PATH:$HBASE_HOME/bin
source ~/.bash_profile

5.7 启动验证

# 启动
start-hbase.sh

# 通过浏览器访问hbase HMaster Web页面
http://hsm01:16010

# HRegionServer Web页面
http://hss01:16030
http://hss02:16030

# shell验证
hbase shell
# list验证
list
# 建表验证
create 'user','name','sex' 

5.8 问题与参考

Hbase与hadoop有版本兼容要求,一般的解决方式都是把Hbase中与hadoop相关的jar包,替换成hadoop版本的jar包。

集群时间记得要同步,同步方式界面操作调整时区和格式。

date -s "yyyyMMdd HH:mm:dd"
clock -w

aspx">hbase启动时报错：java.lang.NoClassDefFoundError: org/htrace/Trace

或者用ntp设置
Linux NTP配置详解 (Network Time Protocol)

6. Spark安装

6.1 安装 Scala

# root安装（其他用户也可以）
tar -xf scala-2.11.7.tgz
mv scala-2.11.7/ /opt/

# 环境变量
vim /etc/profile

export SCALA_HOME=/opt/scala-2.11.7
export PATH=$PATH:$SCALA_HOME/bin

source /etc/profile

# 验证
scala -version

# 将scala复制到其他节点，并配置环境变量
scp -r scala-2.11.7 root@hss01:/opt
scp -r scala-2.11.7 root@hss02:/opt

6.2 解压spark

tar -xf spark-1.6.2-bin-hadoop2.6.tgz
mv spark-1.6.2-bin-hadoop2.6/ spark-1.6.2

6.3 spark-env.sh

# conf目录
cp spark-env.sh.template spark-env.sh

vim spark-env.sh

export JAVA_HOME=/opt/jdk1.8.0_45
export SCALA_HOME=/opt/scala-2.11.7
export SPARK_MASTER_IP=hsm01
export SPARK_WORKER_MEMORY=1g
export HADOOP_CONF_DIR=/home/zkpk/hadoop-2.6.4/etc/hadoop
export HADOOP_OPTS="-Djava.library.path=$HADOOP_HOME/lib:$HADOOP_HOME/lib/native"

6.4 slaves

cp slaves.template slaves

hsm01
hss01
hss02

6.5 复制spark到其他节点

scp -r spark-1.6.2/ hss01:~/
scp -r spark-1.6.2/ hss02:~/

6.6 环境变量

vim ~/.bash_profile

export SPARK_HOME=/home/zkpk/spark-1.6.2
export PATH=$PATH:$SPARK_HOME/bin

source ~/.bash_profile

6.7 启动验证

# 启动（由于和hadoop的启动shell名字一样，需要注意）
$SPARK_HOME/sbin/start-all.sh

# 查看集群状态
http://hsm01:8080/

# 命令行交互验证
./bin/spark-shell

scala> val textFile = sc.textFile("file:///home/zkpk/spark-1.6.2/README.md")
textFile: org.apache.spark.rdd.RDD[String] = file:///home/zkpk/spark-1.6.2/README.md MapPartitionsRDD[1] at textFile at :27

scala> textFile.count()
res0: Long = 95

scala> textFile.first()
res1: String = # Apache Spark