voonhous opened a new issue, #19421: URL: https://github.com/apache/hudi/issues/19421
AvroKafkaSource's `readFromCheckpoint` has a branch that re-configures a custom `KafkaAvroSchemaDeserializer` with the latest source schema before reading (`AvroKafkaSource.java`, the `configure` call guarded by the deserializer check). That branch has regressed twice: - #10118 -- Fix schema refresh for KafkaAvroSchemaDeserializer - #12111 (HUDI-8378) -- Fix Avro schema deserializer failing with schema evolution Despite that history, no test wires `hoodie.streamer.source.kafka.value.deserializer.class` to `KafkaAvroSchemaDeserializer` through `AvroKafkaSource`: the branch is uncovered on codecov master, and `deser/TestKafkaAvroSchemaDeserializer` only exercises the deserializer in isolation. Suggested coverage: construct `AvroKafkaSource` with the custom deserializer configured (embedded Kafka + mocked schema provider), fetch once, evolve the provider's schema, fetch again, and assert the second batch decodes with the refreshed schema. That executes the re-configure path that both past fixes touched. Found while adding tail coverage in #19416; left out of that PR because it needs a schema-evolution fixture rather than a small unit test. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
