Yarn Installation Deployment

The installation of yarn is based on HDFs HA (http://www.cnblogs.com/yinchengzhe/p/5140117.html).

1, Configuration Yarn-site.xml

Parameter Details Reference http://www.cnblogs.com/yinchengzhe/p/5142659.html

The configuration is as follows:

　　

　　

2, Configuration Mapred-site.xml

Under ${hadoop_home}/etc/hadoop/, rename the Mapred-site.xml.templat to Mapred-site.xml

The configuration is as follows:

　　

Compared to Hadoop1.0, the user no longer has to configure Mapred.job.tracker, because Jobtracker has become a library of the client, he may be randomly dispatched to any slave, that is, its location is dynamically generated

3, start off

　　Start yarn on the H1, execute

> sbin/start-yarn.sh

Start Jobhistory Server on H2, execute

> sbin/mr-jobhistory-daemon.sh start historyserver

　　　Note: Hadoop launches jobhistoryserver to implement the historical operation of the Web view job, because the Jobhistoryserver process does not start after the HDFs and yarn processes are started and requires manual startup

　　　　　Jobhistory configuration can be referenced http://www.cnblogs.com/yinchengzhe/p/5142659.html

　　Close sbin/stop-yarn.sh

　　　　　

Yarn Installation Deployment