Fix an issue where CPU time and scheduled time are swapped in query plans.
Fix a correctness issue for queries containing nested coercible type casts.
Add support for recoverable grouped execution when writing unpartitioned table.
Add support for index-based access to fields of type
Fix a security issue in the client protocol.
Web UI Changes#
Hive Connector Changes#
Improve performance for bucketed table insertion when some buckets are empty.
Add config property
hive.max-buckets-for-grouped-executionand session property
max_buckets_for_grouped_executionto limit the number of buckets a query can access while still taking the advantage of grouped execution. If more than the configured number of buckets are queried, the query will run without grouped execution.
Add session property
hive.virtual_bucket_countto support grouped execution for queries that read unbucketed tables and have no joins or aggregations. This session property controls the number of virtual buckets that the connector will create for unbucketed tables.
Raptor Connector Changes#
Fix query failures on colocated joins when one table has empty data.
Add the reason that a test case is skipped to the output event.