Details
-
Bug
-
Status: Closed
-
Minor
-
Resolution: Duplicate
-
None
-
None
Description
The buildSchema method of external sort calls allocateNew() on each vector in schema batch, which normally isn't much memory, but when there are thousands of columns, this ends up being significant. We should avoid allocating any memory for the schema batch.
Attachments
Issue Links
- duplicates
-
DRILL-5117 Compile error when query a json file with 1000+columns
- Closed