-
-
Notifications
You must be signed in to change notification settings - Fork 756
Description
When upgrading to Confluent 8.x, the schema id can either be passed inside the message (existing behavior as a 4 byte prefix in the value) or new as a Kafka Header. Therefore it would be nice to change AKHQ to support it:
-
A change will be needed in the consumer's deserialization logic to use the https://github.com/confluentinc/schema-registry/blob/master/schema-serializer/src/main/java/io/confluent/kafka/serializers/schema/id/DualSchemaIdDeserializer.java, which handles both of the scenarios.
-
when displaying the headers, it would be nice to convert the byte value into a GUID, if the header key is
__value_schema_idor__key_schema_id(reusing the logic in the DualSchemaIdDeserializer:schemaId.fromBytes(ByteBuffer.wrap(headerValue));) -
The produce screen could be changed as well to offer the choice of the two strategies (with the impact that the user has to understand what he is doing). You would not want a user to produce a message with the schema in the header if a consumer still expects it to be in the value/key.
Metadata
Metadata
Assignees
Labels
Projects
Status