散仙在上篇博客,已经讲了如何在编译hadoop2.2.0的64位的版本,如果有不清楚怎么编译,可以参考散仙上篇博文,在这里就不涉及了。
下面,散仙以表格的形式描述下伪分布式安装
系统环境如下:
序号 | 名称 | 描述 |
1 | 系统环境 | Centos6.5 |
2 | JAVA环境 | JDK1.7 |
3 | Haoop版本 | hadoop2.2.0 |
安装步骤如下:
序号 | 步骤 |
1 | 解压hadoop,并配置环境变量 |
2 | 使用which hadoop命令,测试是否成功 |
3 | 配置core-site.xml |
4 | 配置hdfs-site.xml |
5 | 配置yarn-site.xml(非必须,使用默认也可) |
6 | 配置mapred-site.xml |
7 | 配置mapred-env.sh里面的JAVA路径 |
8 | 如上的配置完成后,需要新建几个文件夹,来提供HDFS的格式化底层一个是hadoop的tmp文件夹,另外的是namenode和datanode的文件夹,来分别存储各自的信息 |
9 | 上面一切完成后,即可执行hadoop namenode -format 进行格式化 |
10 | 然后启动伪分布式集群,执行命令sbin/start-all.sh 最后使用JPS查看JAVA进程 |
11 | 配置本机的hosts文件,映射主机名信息 |
下面来详细说下,各个步骤的具体操作:
在这之前,最好配置好本机的SSH无密码登陆操作,避免hadoop进程之间通信时,需要验证。
下面是各个xml文件的配置
- @H_502_221@<!--core-site.xml配置-->@H_502_221@
- @H_502_221@<configuration>@H_502_221@
- @H_502_221@<property>@H_502_221@
- @H_502_221@<name>@H_502_221@fs.default.name</name>@H_502_221@
- @H_502_221@<value>@H_502_221@hdfs://192.168.46.28:9000</value>@H_502_221@
- @H_502_221@</property>@H_502_221@
- @H_502_221@<property>@H_502_221@
- @H_502_221@<name>@H_502_221@hadoop.tmp.dir</name>@H_502_221@
- @H_502_221@<value>@H_502_221@/root/hadoop/tmp</value>@H_502_221@
- @H_502_221@</property>@H_502_221@
- @H_502_221@
- @H_502_221@</configuration>@H_502_221@
- @H_502_221@<!--hdfs-site.xml配置-->@H_502_221@
- @H_502_221@
- @H_502_221@<configuration>@H_502_221@
- @H_502_221@<property>@H_502_221@
- @H_502_221@<name>@H_502_221@dfs.replication</name>@H_502_221@
- @H_502_221@<value>@H_502_221@1</value>@H_502_221@
- @H_502_221@</property>@H_502_221@
- @H_502_221@
- @H_502_221@<property>@H_502_221@
- @H_502_221@<name>@H_502_221@dfs.namenode.name.dir</name>@H_502_221@
- @H_502_221@<value>@H_502_221@/root/hadoop/nddir</value>@H_502_221@
- @H_502_221@</property>@H_502_221@
- @H_502_221@
- @H_502_221@
- @H_502_221@<property>@H_502_221@
- @H_502_221@<name>@H_502_221@dfs.datanode.data.dir</name>@H_502_221@
- @H_502_221@<value>@H_502_221@/root/hadoop/dddir</value>@H_502_221@
- @H_502_221@</property>@H_502_221@
- @H_502_221@
- @H_502_221@<property>@H_502_221@
- @H_502_221@<name>@H_502_221@dfs.permissions</name>@H_502_221@
- @H_502_221@<value>@H_502_221@false</value>@H_502_221@
- @H_502_221@</property>@H_502_221@
- @H_502_221@
- @H_502_221@</configuration>@H_502_221@
- @H_502_221@<!--mapred-site.xml的配置-->@H_502_221@
- @H_502_221@<configuration>@H_502_221@
- @H_502_221@<property>@H_502_221@
- @H_502_221@<name>@H_502_221@mapreduce.cluster.temp.dir</name>@H_502_221@
- @H_502_221@<value></value>@H_502_221@
- @H_502_221@<description>@H_502_221@Nodescription</description>@H_502_221@
- @H_502_221@<final>@H_502_221@true</final>@H_502_221@
- @H_502_221@</property>@H_502_221@
- @H_502_221@
- @H_502_221@<property>@H_502_221@
- @H_502_221@<name>@H_502_221@mapreduce.cluster.local.dir</name>@H_502_221@
- @H_502_221@<value></value>@H_502_221@
- @H_502_221@<description>@H_502_221@Nodescription</description>@H_502_221@
- @H_502_221@<final>@H_502_221@true</final>@H_502_221@
- @H_502_221@</property>@H_502_221@
- @H_502_221@</configuration>@H_502_221@
mapred-env.sh里面的配置
- @H_502_221@@H_502_221@#LicensedtotheApacheSoftwareFoundation(ASF)underoneormore
- @H_502_221@#contributorlicenseagreements.SeetheNOTICEfiledistributedwith
- @H_502_221@#this@H_502_221@workfor@H_502_221@additionalinformationregardingcopyrightownership.
- @H_502_221@#TheASFlicensesthis@H_502_221@filetoYouundertheApacheLicense,Version2.0@H_502_221@
- @H_502_221@#(the"License"@H_502_221@);youmaynotusethis@H_502_221@fileexceptincompliancewith
- @H_502_221@#theLicense.YoumayobtainacopyoftheLicenseat
- @H_502_221@#
- @H_502_221@#http://www.apache.org/licenses/LICENSE-2.0@H_502_221@
- @H_502_221@#
- @H_502_221@#Unlessrequiredbyapplicablelaworagreedtoinwriting,software
- @H_502_221@#distributedundertheLicenseisdistributedonan"ASIS"@H_502_221@BASIS,
- @H_502_221@#WITHOUTWARRANTIESORCONDITIONSOFANYKIND,eitherexpressorimplied.
- @H_502_221@#SeetheLicensefor@H_502_221@thespecificlanguagegoverningpermissionsand
- @H_502_221@#limitationsundertheLicense.
- @H_502_221@
- @H_502_221@exportJAVA_HOME=/usr/local/jdk
- @H_502_221@
- @H_502_221@exportHADOOP_JOB_HISTORYSERVER_HEAPSIZE=1000@H_502_221@
- @H_502_221@
- @H_502_221@exportHADOOP_MAPRED_ROOT_LOGGER=INFO,RFA
- @H_502_221@
- @H_502_221@#exportHADOOP_JOB_HISTORYSERVER_OPTS=
- @H_502_221@#exportHADOOP_MAPRED_LOG_DIR=""@H_502_221@#Wherelogfilesarestored.$HADOOP_MAPRED_HOME/logsbydefault@H_502_221@.
- @H_502_221@#exportHADOOP_JHS_LOGGER=INFO,RFA#HadoopJobSummarylogger.
- @H_502_221@#exportHADOOP_MAPRED_PID_DIR=#Thepidfilesarestored./tmpbydefault@H_502_221@.
- @H_502_221@#exportHADOOP_MAPRED_IDENT_STRING=#Astringrepresentingthis@H_502_221@instanceofhadoop.$USERbydefault@H_502_221@
- @H_502_221@#exportHADOOP_MAPRED_NICENESS=#Theschedulingpriorityfor@H_502_221@daemons.Defaultsto0@H_502_221@.
然后,建对应的几个文件夹即可,路径都在hadoop2.2的根目录下即可,
然后执行格式化命令bin/hadoop namenode -format即可;
最后使用jps命令,查看进程,如果有以下几个进程名称,就代表部署伪分布式成功
然后,访问界面端口,注意与之前的配置文件里一致,namenode的端口号仍为50070,原来1.x的50030端口,已经没有了,可以访问8088的yarn的端口,来查看,具体的截图如下:
OK,此时,我们已经成功的完成伪分布式的部署,下一步我们要做的,就是跑一个经典的MR程序Hellow World,来测试我们的集群了。