You can implement such logic using the advanced MapRunner (Stable API) class, or by overriding the Mapper#run method (New, unstable API).
MapRunner: http://hadoop.apache.org/common/docs/current/api/org/apache/hadoop/mapred/MapRunner.html#run(org.apache.hadoop.mapred.RecordReader, org.apache.hadoop.mapred.OutputCollector, org.apache.hadoop.mapred.Reporter) and http://hadoop.apache.org/common/docs/current/api/org/apache/hadoop/mapred/JobConf.html#setMapRunnerClass(java.lang.Class) to set it. Mapper#run: http://hadoop.apache.org/common/docs/current/api/org/apache/hadoop/mapreduce/Mapper.html#run(org.apache.hadoop.mapreduce.Mapper.Context) 2012/2/4 Radim Kolar <[email protected]>: > Is there way to signal mapreduce framework from mapper or reducer that I am > not interested in any more input data? > > Currently i read rest of data but ignore them. -- Harsh J Customer Ops. Engineer Cloudera | http://tiny.cloudera.com/about
