[HIVE-10231] Compute partition column stats fails if partition col type is date

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.14.0, 1.0.0, 1.1.0
• Fix Version/s: 1.2.0
• Component/s: Statistics
• Assignee: Chaoyu Tang
• Reporter: Chaoyu Tang
• Created: 06/Apr/15 22:37
• Updated: 18/May/15 19:53
• Resolved: 08/Apr/15 16:42

Description

Currently the command "analyze table .. partition .. compute statistics for columns" may only work for partition column type of string, numeric types, but not others like date. See following case using date as partition coltype:

create table colstatspartdate (key int, value string) partitioned by (ds date, hr int);
insert into colstatspartdate partition (ds=date '2015-04-02', hr=2) select key, value from src limit 20;
analyze table colstatspartdate partition (ds=date '2015-04-02', hr=2) compute statistics for columns;

you will get RuntimeException:

FAILED: RuntimeException Cannot convert to Date from: int
15/04/06 17:30:01 ERROR ql.Driver: FAILED: RuntimeException Cannot convert to Date from: int
java.lang.RuntimeException: Cannot convert to Date from: int
	at org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorUtils.getDate(PrimitiveObjectInspectorUtils.java:1048)
	at org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorConverter$DateConverter.convert(PrimitiveObjectInspectorConverter.java:264)
	at org.apache.hadoop.hive.ql.optimizer.ConstantPropagateProcFactory.typeCast(ConstantPropagateProcFactory.java:163)
	at org.apache.hadoop.hive.ql.optimizer.ConstantPropagateProcFactory.propagate(ConstantPropagateProcFactory.java:333)
	at org.apache.hadoop.hive.ql.optimizer.ConstantPropagateProcFactory.foldExpr(ConstantPropagateProcFactory.java:242)
....