[NIFI-4746] ConfluentSchemaRegistry Controller service only uses latest schema, failing if schema changed

• Type: Bug
• Status: Patch Available
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.4.0
• Component/s: Core Framework
• Labels: confluent-kafka confluentic schema
• Flags: Important
• Assignee: Unassigned
• Reporter: AdFel
• Created: 08/Jan/18 10:58
• Updated: 11/Jan/18 13:35

Description

When ConsumeKafkaRecord_01X uses the ConfluentSchemaRegistry 1.4.0 the id in the magic byte is not used to merge the latest schema with the writer schema, as stated in the Apache Avro documentation. Nifi instead compares the id to the latest schema's ID, and if they do not match, the flowfile is sent to the parse.failure. To fix this bug the ConfluentSchemaRegistry controller should merge the writer's schema with the latest schema, using the ID written in the magic byte