【Hadoop】YARN 完全分布式配置

承接 https://www.cnblogs.com/jzsg/p/12622214.html

mv mapred-site.xml.template  mapred-site.xml

// 基于YARN的资源调度配置
vi etc/hadoop/mapred-site.xml:



    
        mapreduce.framework.name
        yarn
    




vi etc/hadoop/yarn-site.xml:





    
        yarn.nodemanager.aux-services
        mapreduce_shuffle
    

    
        yarn.resourcemanager.resource-tracker.address
        172.16.152.130:8031
    
    
            yarn.resourcemanager.address
            172.16.152.130:8032
    
    
            yarn.resourcemanager.scheduler.address
            172.16.152.130:8034
    
    
            yarn.resourcemanager.webapp.address
            172.16.152.130:8088
    

    
            yarn.log-aggregation-enable
            true
    
    
          yarn.log.server.url
          http://master:19888/jobhistory/logs/
    


  


修改后拷贝至从节点（slave）
scp -r yarn-site.xml root@slave:/usr/local/software/hadoop-2.5.1/etc/hadoop/


$ sbin/start-yarn.sh

$ sbin/stop-yarn.sh


在master上jsp 发现 20906 ResourceManager，从节点发现20808 NodeManager 即成功，web 访问YARN ：http://master:8088/cluster