hive.map.aggr

[Steven Wong <sw...@netflix.com>]

I have a query doing JOIN and GROUP BY:

SELECT ...
FROM x JOIN y ON (...)
GROUP BY ...;

In the first MR job, the reduce phase performs hash aggregation when hive.map.aggr=true, but the reduce phase doesn't perform hash aggregation when hive.map.aggr=false. Why does hive.map.aggr affect the reduce phase behavior when its name and description indicate it controls map-side aggregation?

<property>
  <name>hive.map.aggr</name>
  <value>true</value>
  <description>Whether to use map-side aggregation in Hive Group By queries</description>
</property>

Thanks.
Steven