hadoop分布式安装

1.hadoop-2.7.4.tar.gz跟jdk-8u121-linux-x64.tar.gz自行下载并解压，我全部放置在/usr/local/下并命名为hadoop与jdk1.8.0_121，当然只要是1.8的jdk都可以没必要完全一样

2.所有机器做免密认证，并且对主机名写hosts

1 [root@hadoop1 hadoop]# cat /etc/hosts
2 127.0.0.1 localhost
3 ::1 localhost localhost.localdomain localhost6 localhost6.localdomain6
4 192.168.56.11 hadoop1
5 192.168.56.12 hadoop2
6 192.168.56.13 hadoop3

3.配置两个环境变量后生效下

 1 [root@hadoop1 hadoop]# mkdir /usr/local/hadoop/tmp/    # 所有机器
 2 [root@hadoop1 hadoop]# tail /etc/profile
 3 ############# hadoop env #################
 4 export HADOOP_HOME=/usr/local/hadoop
 5 export PATH=$PATH:$HADOOP_HOME/bin:$HADOOP_HOME/sbin
 6 
 7 ############# jdk env #################
 8 export JAVA_HOME=/usr/local/jdk1.8.0_121
 9 export CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar
10 export PATH=$JAVA_HOME/bin:$PATH

4.配置hadoop的个配置文件，只在第一台机器上操作即可

 1 [root@hadoop1 ~]# vim /usr/local/hadoop/etc/hadoop/core-site.xml            # 指定hdfs的主
 2 <configuration>
 3     <property>
 4         <name>fs.defaultFS</name>
 5         <value>hdfs://hadoop1:9000</value>
 6     </property>
 7     <property>
 8         <name>hadoop.tmp.dir</name>
 9         <value>/usr/local/hadoop/tmp</value>
10     </property>
11 </configuration>
12     
13 [root@hadoop1 ~]# vim /usr/local/hadoop/etc/hadoop/hdfs-site.xml            # 指定hdfs备份副本数量
14 <configuration>
15     <property>
16         <name>dfs.replication</name>
17         <value>3</value>
18     </property>
19 </configuration>

20 [root@hadoop1 ~]# cp /usr/local/hadoop/etc/hadoop/lmapred-site.xml.template /usr/local/hadoop/etc/hadoop/mapred-site.xml
21 [root@hadoop1 ~]# vim /usr/local/hadoop/etc/hadoop/mapred-site.xml            # 指定mr运行在yarn上
22 <configuration>
23     <property>
24         <name>mapreduce.framework.name</name>
25         <value>yarn</value>
26     </property>
27 </configuration>
28 
29 [root@hadoop1 ~]# vim /usr/local/hadoop/etc/hadoop/yarn-site.xml            # 指定yarn的主
30 <configuration>
31 
32 <!-- Site specific YARN configuration properties -->
33     <property>
34         <name>yarn.nodemanager.aux-services</name>
35         <value>mapreduce_shuffle</value>
36     </property>
37     <property>
38         <name>yarn.resourcemanager.hostname</name>
39         <value>hadoop1</value>
40     </property>
41     <property>
42         <name>yarn.nodemanager.resource.memory-mb</name>
43         <value>1024</value>                                                    # 占用的内存，默认是8g，不可以小于1024
44     </property>
45     <property>
46         <name>yarn.nodemanager.resource.cpu-vcores</name>
47         <value>1</value>                                                    # 使用cpu的核数，物理机可以多分配
48     </property>
49 </configuration>
50 
51 [root@hadoop1 ~]# vim /usr/local/hadoop/etc/hadoop/slaves                    # 指定hadoop集群机器
52 hadoop1
53 hadoop2
54 hadoop3
55 
56 [root@hadoop1 ~]# scp -r /usr/local/hadoop/etc/hadoop/* hadoop2:/usr/local/hadoop/etc/hadoop/
57 [root@hadoop1 ~]# scp -r /usr/local/hadoop/etc/hadoop/* hadoop3:/usr/local/hadoop/etc/hadoop/
58 [root@hadoop1 ~]# hdfs namenode -format
59 [root@hadoop1 ~]# /usr/local/hadoop/etc/hadoop/start-dfs.sh
60 [root@hadoop1 ~]# /usr/local/hadoop/etc/hadoop/start-yarn.sh

5.验证

 1 [root@hadoop1 ~]# jps    # 主机
 2 17641 NodeManager
 3 17946 Jps
 4 14331 DataNode
 5 14204 NameNode
 6 14541 SecondaryNameNode
 7 17519 ResourceManager
 8 
 9 [root@hadoop2 java]# jps    # 从机
10 14784 NodeManager
11 13797 DataNode
12 14918 Jps

http://ip:50070可查看hdfs信息