Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Fixed
-
None
-
None
-
None
Description
the biggest problem is that select * is assuming that the output is a regular text file. it can't read out from compressed text files. (which is what happens when we start using the output compression options).
the core issue is that select * is special code. and it should not be. it should go through the same 'fileinputformat' and serde code that any map-side task goes through.
the second issue is that a select * without any data transformations/filters goes through a map-reduce task unnecessarily. we need to fix this. this seems related to the limit N jira that ashish opened. (see hadoop-4086)
Attachments
Issue Links
- is related to
-
HADOOP-4086 Add limit to Hive QL
- Closed