timestampExtractorBeanName setting in the Spring Cloud Stream application doesn't override default value

问题

I have the following properties for my Spring Cloud Stream application that uses Kafka Streams Binder:

spring.cloud.stream.bindings:
  windowStream-in-0:
    destination: input
  windowStream-out-0:
    destination: window1
  hint1Stream-in-0:
    destination: window1
  hint1Stream-out-0:
    destination: hints
  realityStream-in-0:
    destination: input
  realityStream-in-1:
    destination: window1
    consumer:
      timestampExtractorBeanName: anotherTimestampExtractor
  realityStream-out-0:
    destination: hints
  countStream-in-0:
    destination: hints

spring.cloud.stream.kafka.streams:
  default:
    consumer:
      timestampExtractorBeanName: timestampExtractor
  binder:
    functions:
      windowStream:
        applicationId: mock-stream-window1
      hint1Stream:
        applicationId: mock-stream-hints
      realityStream:
        applicationId: mock-stream-reality
      countStream:
        applicationId: mock-stream-count
    stateStoreRetry:
      maxAttempts: 3
      backOffInterval: 1000
    configuration:
      schema.registry.url: mock://mock-stream-registry
      default.key.serde: org.apache.kafka.common.serialization.Serdes$StringSerde
      default.value.serde: io.confluent.kafka.streams.serdes.avro.SpecificAvroSerde
      commit.interval.ms: 100

What I am trying to do is to use the "timestampExtractor" for all streams except for one, called "realityStream".

For that I set the spring.cloud.stream.kafka.streams.default.consumer.timestampExtractorBeanName to timestampExtractor and then try to "override" it for the "realityStream" by setting spring.cloud.stream.bindings.realityStream-in-1.consumer.timestampExtractorBeanName

Unfortunately, it seems that my override doesn't work, since only "timestampExtractor" is getting called only, as I can see in the debugger (and in my test results).

Am I applying the configuration wrong, or have wrong expectation?

Here's a picture of the streams in my single Spring Cloud Streams application:

(orange circle is the place where I want to apply non-default timestamp extractor)

回答1:

The override is in the wrong place; it needs to be under the ....kafka.streams.bindings.realityStream-in-1.... property.

It's a kafka-specific property; you have it in the generic bindings properties (common to all binders).

来源：https://stackoverflow.com/questions/62039706/timestampextractorbeanname-setting-in-the-spring-cloud-stream-application-doesn

标签

apache-kafka-streams

spring-cloud-stream