Hadoop部署的三种模式：
- 单机模式（standalone）
一般不建议安装，网络上很少这方面资料
单机模式是Hadoop的默认模式。当首次解压Hadoop的源码包时，Hadoop无法了解硬件安装环境，便保守地选择了最小配置。
在这种默认模式下所有3个XML文件均为空。当配置文件为空时，Hadoop会完全运行在本地。
因为不需要与其他节点交互，单机模式就不使用HDFS，也不加载任何Hadoop的守护进程。
该模式主要用于开发调试MapReduce程序的应用逻辑。
伪分布模式（Pseudo-Distributed Mode）
伪分布模式在“单节点集群”上运行Hadoop，其中所有的守护进程都运行在同一台机器上。 该模式在单机模式之上增加了代码调试功能，允许你检查内存使用情况，HDFS输入输出，以及其他的守护进程交互。 比如namenode，datanode，secondarynamenode，jobtracer，tasktracer这5个进程，都能在集群上看到。全分布模式（Fully Distributed Mode）
Hadoop守护进程运行在一个集群上。 意思是说master上看到namenode,jobtracer，secondarynamenode可以安装在master节点，也可以单独安装。slave节点能看到datanode和tasktracer
Hadoop伪分布模式部署
环境要求java、ssh
添加hadoop用户
[root@hadoop-01 ~]# useradd hadoop [root@hadoop-01 ~]# vi /etc/sudoers # 找到root ALL=(ALL) ALL，添加 hadoop ALL=(ALL) NOPASSWD:ALL上传并解压
[root@hadoop-01 software]# rz #上传hadoop-2.8.1.tar.gz [root@hadoop-01 software]# tar -xzvf hadoop-2.8.1.tar.gz软连接
[root@hadoop-01 software]# ln -s /opt/software/hadoop-2.8.1 /opt/software/hadoop设置环境变量
[root@hadoop-01 software]# vi /etc/profile export HADOOP_HOME=/opt/software/hadoop export PATH=$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$PROTOC_HOME/bin:$FINDBUGS_HOME/bin:$MAVEN_HOME/bin:$JAVA_HOME/bin:$PATH [root@hadoop-01 software]# source /etc/profile设置用户用户组
[root@hadoop-01 software]# chown -R hadoop:hadoop hadoop [root@hadoop-01 software]# chown -R hadoop:hadoop hadoop/* [root@hadoop-01 software]# chown -R hadoop:hadoop hadoop-2.8.1 [root@hadoop-01 software]# cd hadoop [root@hadoop-01 hadoop]# rm -f *.txt切换用户hadoop
[root@hadoop-01 software]# su - hadoop [root@hadoop-01 hadoop]# ll total 32 drwxrwxr-x. 2 hadoop hadoop 4096 Jun 2 14:24 bin drwxrwxr-x. 3 hadoop hadoop 4096 Jun 2 14:24 etc drwxrwxr-x. 2 hadoop hadoop 4096 Jun 2 14:24 include drwxrwxr-x. 3 hadoop hadoop 4096 Jun 2 14:24 lib drwxrwxr-x. 2 hadoop hadoop 4096 Aug 20 13:59 libexec drwxr-xr-x. 2 hadoop hadoop 4096 Aug 20 13:59 logs drwxrwxr-x. 2 hadoop hadoop 4096 Jun 2 14:24 sbin drwxrwxr-x. 4 hadoop hadoop 4096 Jun 2 14:24 share # bin: 可执行文件 # etc: 配置文件 # sbin: shell脚本，启动关闭hdfs,yarn等配置文件
[hadoop@hadoop-01 ~]# cd /opt/software/hadoop [hadoop@hadoop-01 hadoop]# vi etc/hadoop/core-site.xml <configuration> <property> <name>fs.defaultFS</name> <value>hdfs://192.168.137.130:9000</value> # 配置自己机器的IP </property> </configuration> [hadoop@hadoop-01 hadoop]# vi etc/hadoop/hdfs-site.xml <configuration> <property> <name>dfs.replication</name> <value>1</value> </property> </configuration>配置hadoop用户的ssh信任关系
# 公钥/密钥 配置无密码登录 [hadoop@hadoop-01 ~]# ssh-keygen -t rsa -P '' -f ~/.ssh/id_rsa [hadoop@hadoop-01 ~]# cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys [hadoop@hadoop-01 ~]# chmod 0600 ~/.ssh/authorized_keys # 查看日期，看是否配置成功 [hadoop@hadoop-01 ~]# ssh hadoop-01 date The authenticity of host 'hadoop-01 (192.168.137.130)' can't be established. RSA key fingerprint is 09:f6:4a:f1:a0:bd:79:fd:34:e7:75:94:0b:3c:83:5a. Are you sure you want to continue connecting (yes/no)? yes # 第一次回车输入yes Warning: Permanently added 'hadoop-01,192.168.137.130' (RSA) to the list of known hosts. Sun Aug 20 14:22:28 CST 2017 [hadoop@hadoop-01 ~]# ssh hadoop-01 date #不需要回车输入yes,即OK Sun Aug 20 14:22:29 CST 2017 [hadoop@hadoop-01 ~]# ssh localhost date The authenticity of host 'hadoop-01 (192.168.137.130)' can't be established. RSA key fingerprint is 09:f6:4a:f1:a0:bd:79:fd:34:e7:75:94:0b:3c:83:5a. Are you sure you want to continue connecting (yes/no)? yes # 第一次回车输入yes Warning: Permanently added 'hadoop-01,192.168.137.130' (RSA) to the list of known hosts. Sun Aug 20 14:22:28 CST 2017 [hadoop@hadoop-01 ~]# ssh localhost date #不需要回车输入yes,即OK Sun Aug 20 14:22:29 CST 2017格式化和启动
[hadoop@hadoop-01 hadoop]# bin/hdfs namenode -format [hadoop@hadoop-01 hadoop]# sbin/start-dfs.sh ERROR: hadoop-01: Error: JAVA_HOME is not set and could not be found. localhost: Error: JAVA_HOME is not set and could not be found. 解决方法:添加环境变量 [hadoop@hadoop-01 hadoop]# vi etc/hadoop/hadoop-env.sh # 将export JAVA_HOME=${JAVA_HOME}改为 export JAVA_HOME=/usr/java/jdk1.8.0_45 [hadoop@hadoop-01 hadoop]# sbin/start-dfs.sh ERROR: mkdir: cannot create directory `/opt/software/hadoop-2.8.1/logs': Permission denied 解决方法: [hadoop@hadoop-01 hadoop]# exit [root@hadoop-01 hadoop]# cd ../ [root@hadoop-01 software]# chown -R hadoop:hadoop hadoop-2.8.1 [root@hadoop-01 software]# su - hadoop [root@hadoop-01 ~]# cd /opt/software/hadoop # 继续启动 [hadoop@hadoop-01 hadoop]# sbin/start-dfs.sh检查是否成功
[hadoop@hadoop-01 hadoop]# jps 19536 DataNode 19440 NameNode 19876 Jps 19740 SecondaryNameNode
访问： http://192.168.137.130:50070
修改dfs启动的进程，以hadoop-01启动
启动的三个进程： namenode: hadoop-01 bin/hdfs getconf -namenodes datanode: localhost datanodes (using default slaves file) etc/hadoop/slaves secondarynamenode: 0.0.0.0 [hadoop@hadoop-01 ~]# cd /opt/software/hadoop [hadoop@hadoop-01 hadoop]# echo "hadoop-01" > ./etc/hadoop/slaves [hadoop@hadoop-01 hadoop]# cat ./etc/hadoop/slaves hadoop-01 [hadoop@hadoop-01 hadoop]# vi ./etc/hadoop/hdfs-site.xml <property> <name>dfs.namenode.secondary.http-address</name> <value>hadoop-01:50090</value> </property> <property> <name>dfs.namenode.secondary.https-address</name> <value>hadoop-01:50091</value> </property> # 重启 [hadoop@hadoop-01 hadoop]# sbin/stop-dfs.sh [hadoop@hadoop-01 hadoop]# sbin/start-dfs.sh
