[HIVE-11288] Avro SerDe InstanceCache returns incorrect schema

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.3.0, 2.0.0
• Component/s: Serializers/Deserializers
• Labels: Avro AvroSerde
• Flags: Patch
• Assignee: Greg Phillips
• Reporter: Greg Phillips
• Created: 17/Jul/15 12:37
• Updated: 16/Feb/16 23:51
• Resolved: 28/Jul/15 18:42

Description

To reproduce this error, take two fields in an avro schema document matching the following:
"type" : { "type": "array", "items": [ "null",

{ "type": "map", "values": [ "null", "string" ] }

] }
"type" : { "type": "map", "values": [ "null" ,

{ "type": "array", "items": [ "null" , "string"] }

] }
After creating two tables in hive with these schemas, the describe statement on each of them will only return the schema for the first one loaded. This is due to a hashCode() collision in the InstanceCache.

A patch will be included in this ticket shortly which removes the hashCode call from the InstanceCache's internal HashMap, and instead provides the entire schema object.