[FLINK-31879] org.apache.avro.util.Utf8 cannot be serialized with avro when used in state

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: API / Type Serialization System
• Assignee: Unassigned
• Reporter: Feroze Daud
• Created: 21/Apr/23 18:10
• Updated: 21/Apr/23 18:10

Description

Scenario:

Write a flink app that reads avro messages from a kafka topic.

The avro pojos are generated with org.apache.avro.util.Utf8 type instead of java.lang.String

When this happens, Flink logs an error message as follows:

Class class org.apache.avro.util.Utf8 cannot be used as a POJO type because not all fields are valid POJO fields, and must be processed as GenericType. Please read the Flink documentation on "Data Types & Serialization" for details of the effect on performance. 

 

This is problematic because `Utf8` is designed to be a fast serialized/deserialized type for Avro. But since it is not inheriting from SpecificRecordBase, it seems as if it gets handled by Kryo serializer.