[HIVE-1249] MapJoin to Check Number of Files to Match Metadata

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 0.10.0
• Component/s: Query Processor
• Assignee: Namit Jain
• Reporter: Siying Dong
• Created: 16/Mar/10 22:16
• Updated: 24/Mar/14 20:57
• Resolved: 14/Aug/12 05:57

Description

Sorted Map Join only can execute when two tables' bucket numbers are the same.
Currently, the checking is done by looking at metadata of the table. However, for some reason, the real physical storage is not guaranteed to follow what the metadata indicates. In that case, Sorted Map Join might fail to identify it and deliver wrong results. An extra check when storage scan is necessary to deal with the scenario.

This issue may also apply to normal map join.