Hadoop JobHistory

hadoop jobhistory记录下已运行完的MapReduce做业信息并存放在指定的HDFS目录下，默认状况下是没有启动的，须要配置完后手工启动服务。

mapred-site.xml添加以下配置

<property>
  <name>mapreduce.jobhistory.address</name>
  <value>hadoop000:10020</value>
  <description>MapReduce JobHistory Server IPC host:port</description>
</property>

<property>
  <name>mapreduce.jobhistory.webapp.address</name>
  <value>hadoop000:19888</value>
  <description>MapReduce JobHistory Server Web UI host:port</description>
</property>

<property>
    <name>mapreduce.jobhistory.done-dir</name>
    <value>/history/done</value>
</property>

<property>
    <name>mapreduce.jobhistory.intermediate-done-dir</name>
    <value>/history/done_intermediate</value></property>

启动history-server：

$HADOOP_HOME/sbin/mr-jobhistory-daemon.sh start historyserver

中止history-server：

$HADOOP_HOME/sbin/mr-jobhistory-daemon.sh stop historyserver

history-server启动以后，能够经过浏览器访问WEBUI: hadoop000:19888

在hdfs上会生成两个目录

hadoop fs -ls /history

drwxrwx---   - spark supergroup          0 2014-10-11 15:11 /history/done
drwxrwxrwt   - spark supergroup          0 2014-10-11 15:16 /history/done_intermediate

mapreduce.jobhistory.done-dir(/history/done): Directory where history files are managed by the MR JobHistory Server(已完成做业信息)
mapreduce.jobhistory.intermediate-done-dir(/history/done_intermediate): Directory where history files are written by MapReduce jobs.(正在运行做业信息)

测试：

经过hive查询city表观察hdfs文件目录和hadoop000:19888

hive> select id, name from city;

观察hdfs文件目录：

1）历史做业记录是按照年/月/日的形式分别存放在相应的目录（/history/done/2014/10/11/000000）；

2）每一个做业有2个不一样的后缀名的记录：jhist和xml

hadoop fs -ls /history/done/2014/10/11/000000

-rwxrwx---   1 spark supergroup      22572 2014-10-11 15:23 /history/done/2014/10/11/000000/job_1413011730351_0002-1413012208648-spark-select+id%2C+name+from+city%28Stage%2D1%29-1413012224777-1-0-SUCCEEDED-root.spark-1413012216261.jhist -rwxrwx---   1 spark supergroup     160149 2014-10-11 15:23 /history/done/2014/10/11/000000/job_1413011730351_0002_conf.xml

观察WEBUI: hadoop000:19888

在WEBUI中展示了每一个job使用的Map/Reduce的数量、做业提交时间、做业启动时间、做业完成时间、Job ID、提交人User、队列等信息；

点击【job_1413011730351_0002】弹出页面显示相似信息：Aggregation is not enabled. Try the nodemanager at ......

解决方法： yarn-site.xml添加以下配置

<property>  
    <name>yarn.log-aggregation-enable</name>  
    <value>true</value>  
</property> 

重启yarn便可。

 

 

参考CDH文档：http://archive.cloudera.com/cdh5/cdh/5/hadoop-2.3.0-cdh5.0.0/hadoop-project-dist/hadoop-common/ClusterSetup.html