hadoop常用技巧

hadoop输入多个文件时获取输入文件名，目的是可以对多个输入文件进行区分

filepath = os.environ["map_input_file"]
filename = os.path.split(filepath)[-1]

• https://blog.csdn.net/catchertherye/article/details/44410645
• https://blog.csdn.net/modiziri/article/details/47661091

hadoop多路输出

• http://wiki.baidu.com/pages/viewpage.action?pageId=41884974

hadoop touchz命令：创建一个空文件

• hadoop fs -touchz xx.done[check文件完成小技巧]

hadoop text命令：将源文件输出为文本格式

• hadoop fs -text

hadoop grep命令，grep on Hadoop

   hadoop jar hadoop-streaming-1.2.1.jar \
    -D mapreduce.job.name="grep" \
    -D stream.non.zero.exit.is.failure=false \
    -mapper "grep $1" \
    -numReduceTasks 0 \
    -input /INPUT \
    -output /OUTPUT

hadoop动态修改任务优先级

• hadoop job -set-priority job_name HIGH

hadoop数据去重的例子

• https://www.mmmmmcclxxvii.cn/2017/06/09/dedup/ 【待实现】

hadoop join的方法【待实现】

• http://www.crazyant.net/1112.html
• https://blog.csdn.net/levy_cui/article/details/68945542
• https://www.jianshu.com/p/9375de2411db