[ https://issues.apache.org/jira/browse/MAPREDUCE-5051?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13597403#comment-13597403 ]
Robert Joseph Evans commented on MAPREDUCE-5051: ------------------------------------------------ In trunk it is the MergeManagerImpl that calls the combiner ./hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/task/reduce/MergeManagerImpl.java In branch-1 MapTask.java itself handles the combiner code. You can look at Task.java for some of the implementations of it. > Combiner not used when NUM_REDUCES=0 > ------------------------------------ > > Key: MAPREDUCE-5051 > URL: https://issues.apache.org/jira/browse/MAPREDUCE-5051 > Project: Hadoop Map/Reduce > Issue Type: Bug > Components: mrv1 > Affects Versions: 2.0.2-alpha > Environment: CDH4.1.2 MR1 > Reporter: Damien Hardy > > We have a M/R job that use Mapper + Combiner but have nothing to do in > Reducer : > Bulk indexing of HBase data in ElasticSearch, > Map output is K / V : #bulk / json_data_to_be_indexed. > So job is launched maps work, combiners index and a reducer is created for > nothing (sometimes waiting for other M/R job to free a tasktracker slot for > reducer cf. MAPREDUCE-5019 ) > When we put ```job.setNumReduceTasks(0);``` in our job .run(), mapper are > started but combiner are not used. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira