一、首先说一下,Hadoop有三种运行模式。
第一个是独立(或本地)运行模式:无需运行任何守护进程,所有程序都在一个同一个JVM上执行。
第二个是伪分布模式:Hadoop守护进程运行在本地机器上,模拟一个小规模的集群。
第三个是全分布模式,也就是通常所说的完全分布式模式:Hadoop守护进程运行在一个集群上。
下面简单介绍下完全分布式的配置:
首先准备三台机器,系统公共配置部分:
系统信息:
[root@hadoop1 ~]# cat /etc/redhat-release
CentOS release 6.7 (Final)
[centos@hadoop1 ~]$ uname -r
2.6.32-573.el6.x86_64
[root@hadoop1 ~]# uname -m
x86_64
公共操作用户:centos
centos用户在/etc/sudoers中的公共配置:
[root@hadoop1 ~]# grep centos /etc/sudoers
centos ALL=(ALL) NOPASSWD: ALL
环境变量如下:
[root@hadoop1 ~]# tail -16 /etc/profile
#java
JAVA_HOME=/home/centos/download/jdk
PATH=$JAVA_HOME/bin:$PATH
CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar
export JAVA_HOME
export CLASSPATH
#hadoop
HADOOP_HOME=/home/centos/download/hadoop
PATH=$PATH:$HADOOP_HOME/bin:$HADOOP_HOME/sbin
export PATH
主机IP域名映射如下:
[root@hadoop1 ~]# cat /etc/hosts
10.0.0.101 hadoop1
10.0.0.102 hadoop2
10.0.0.103 hadoop3
主节点IP为:
[root@hadoop1 ~]# ifconfig | awk -F '[ :]+' 'NR==2{print $4}'
10.0.0.101
从节点IP分别为:
[root@hadoop2 ~]# ifconfig | awk -F '[ :]+' 'NR==2{print $4}'
10.0.0.102
[root@hadoop3 ~]# ifconfig | awk -F '[ :]+' 'NR==2{print $4}'
10.0.0.103
Java版本
[root@hadoop1 ~]# java -version
java version "1.8.0_65"
Java(TM) SE Runtime Environment (build1.8.0_65-b17)
Java HotSpot(TM) 64-Bit Server VM (build25.65-b01, mixed mode)
Hadoop版本
[root@hadoop1 ~]# hadoop version
Hadoop 2.7.3
Subversionhttps://git-wip-us.apache.org/repos/asf/hadoop.git -r baa91f7c6bc9cb92be5982de4719c1c8af91ccff
Compiled by root on 2016-08-18T01:41Z
Compiled with protoc 2.5.0
From source with checksum2e4ce5f957ea4db193bce3734ff29ff4
This command was run using/home/centos/download/hadoop-2.7.3/share/hadoop/common/hadoop-common-2.7.3.jar
二、配置ssh
首先在三台机器上的家目录分别执行下述生成密钥对脚本,这样的话,三台机器的centos之间就可以免密码登录、免密码传输文件等操作。
[centos@hadoop1 ~]$ ll createssh.sh
-rwxr-xr-x 1 centos centos 51 May 14 11:05createssh.sh
[centos@hadoop1 ~]$ cat createssh.sh
#!/bin/sh
ssh-keygen -t rsa -P '' -f ~/.ssh/id_rsa
然后在三台机器上的家目录分别执行下述分发公钥
[centos@hadoop1 ~]$ ll fen.sh
-rwxr-xr-x 1 centos centos 142 May 14 12:14fen.sh
[centos@hadoop1 ~]$ cat fen.sh
#!/bin/sh
ssh-copy-id -i ~/.ssh/id_rsa.pub centos@hadoop1
ssh-copy-id -i ~/.ssh/id_rsa.pub centos@hadoop2
ssh-copy-id -i ~/.ssh/id_rsa.pub centos@hadoop3
#上述只是在执行的时候会输入yes和系统用户密码,其他在执行的时候就不会再次输入密码了。执行SSH的目的是,比如在配置好完全分布式的时候,主节点会自动吊起从节点的日志并且往里面写入数据,此时SSH的密钥对就起作用了。
#经常碰到的坑,要配置三台机器家目录中.ssh的权限为700,而且要配置家目录中.ssh目录中的authorized_keys权限为644.
[centos@hadoop1 ~]$ chmod 700 .ssh
[centos@hadoop1 ~]$ chmod 644 .ssh/authorized_keys
三、重点的配置文件内容
#三台机器共同配置
[centos@hadoop1 ~]$ grep JAVA_HOME download/hadoop/etc/hadoop/hadoop-env.sh
# The only required environment variable isJAVA_HOME. All others are
# set JAVA_HOME in this file, so that it iscorrectly defined on
export JAVA_HOME=/home/centos/download/jdk #JAVA环境变量要在这再配置一下
[centos@hadoop1 ~]$ cat download/hadoop/etc/hadoop/core-site.xml
<?xml version="1.0"?>
<configuration>
<property>
<name>fs.defaultFS</name>
<value>hdfs://hadoop1</value>
</property>
</configuration>
[centos@hadoop1 ~]$ cat download/hadoop/etc/hadoop/hdfs-site.xml
<?xml version="1.0"?>
<configuration>
<property>
<name>dfs.replication</name>
<value>2</value>#副本数,对应从节点数
</property>
</configuration>
[centos@hadoop1 ~]$ cat download/hadoop/etc/hadoop/mapred-site.xml
<?xml version="1.0"?>
<configuration>
<property>
<name>mapreduce.framework.name</name>
<value>yarn</value>
</property>
</configuration>
[centos@hadoop1 ~]$ cat download/hadoop/etc/hadoop/yarn-site.xml
<?xml version="1.0"?>
<configuration>
<property>
<name>yarn.resourcemanager.hostname</name>
<value>hadoop1</value>
</property>
<property>
<name>yarn.nodemanager.aux-services</name>
<value>mapreduce_shuffle</value>
</property>
</configuration>
#注意这里只配置从节点的域名
[centos@hadoop1 ~]$ cat download/hadoop/etc/hadoop/slaves
hadoop2
hadoop3
四、启动hadoop
#下述命令只在主节点hadoop1中执行!
#首先执行格式化命令
[centos@hadoop1 ~]$ hadoop namenode -format
#然后启动hadoop
[centos@hadoop1 ~]$ start-all.sh
#查看启动进程
#主节点执行,名称节点只会存在于主节点中,包括文件名称等内容。
[centos@hadoop1 ~]$ jps
2736 SecondaryNameNode
2961 Jps
2889 ResourceManager
2539 NameNode
#从节点执行,数据节点只会存在于从节点,是真正存放数据的地方。
[centos@hadoop2 ~]$ jps
2322 NodeManager
2234 DataNode
2380 Jps
[centos@hadoop3 ~]$ jps
2481 DataNode
2646 Jps
2569 NodeManager
#临时文件都生成在了根目录下的/tmp目录下
[centos@hadoop1 ~]$ ll /tmp/
total 32
drwxrwxr-x 3 centos centos 4096 May 1414:05 hadoop-centos
-rw-rw-r-- 1 centos centos 5 May 14 14:06 hadoop-centos-namenode.pid
-rw-rw-r-- 1 centos centos 5 May 14 14:06hadoop-centos-secondarynamenode.pid
drwxr-xr-x 2 centos centos 4096 May 1414:06 hsperfdata_centos
drwxrwxr-x 3 centos centos 4096 May 1414:06 Jetty_0_0_0_0_50070_hdfs____w2cu08
drwxrwxr-x 3 centos centos 4096 May 1414:06 Jetty_0_0_0_0_50090_secondary____y6aanv
drwxrwxr-x 4 centos centos 4096 May 1414:06 Jetty_hadoop1_8088_cluster____.jtly5k
-rw-rw-r-- 1 centos centos 5 May 14 14:06yarn-centos-resourcemanager.pid