Details
-
Improvement
-
Status: Open
-
Minor
-
Resolution: Unresolved
-
None
-
None
-
None
-
Spark 2.1 + CarbonData 1.1.0
Description
Currently, we will clear access count for all segments after getSplits finished, this will keep the index in memory more time, but if we don't have enough memory, then will query failed, so clear access count after getSplits for each segment, then we can remove the index tree if we don't have enough memory to load index tree for other segments.