micronaut-kafka-avro

A Demo application for Schema-Evolution by using Kafka, Micronaut and Avro serialization.

Scope:

• Creating auto generated SpecificRecord class using gradle-avro-plugin and Avro schema file
• Producer can send Avro serialized data with different schema version
• Consumer can use auto generated class consume predefined schema version without breaking
• Evaluating polymorphism and inheritance with Avro serialization

Technologies:

• Kotlin 1.4.0
• Confluent Platform 6.0.0
• Micronaut 2.1.2
• Avro 1.10.0
• gradle-avro-plugin 0.20.0

Schema Evolution via Specific Record

Step 1 Start Confluent Platform:

$ confluent start

Step 2 Prepare topics for kafka streams:

$ ./createTopics topics.txt

Step 3 Use gradle-avro-plug to generate Java Pojo class, which is based on avro schema file:

# check out branch, which has schema version 1
$ git checkout v1/spec-record  

# compile java pojo class for producer
$ gradlew clean build   

Step 4 Start backend service

$ gradlew run

Step 5 Send some record via GraphiQL interface

open graphiql interface in broswer http://localhsot:8080/graphiql and send record with graphql mutation request

mutation{
  createPartnerV1(partnerV1: {
    {
      partnerId: "123",
      lastEventId: "test-event-id",
      type: "NaturalPerson",
      firstName: "bai",
      secondName: "xia",
      birthDay: "1984-04-11",
      telephone: "7891234",
      email: "[email protected]"
    })
  {
    partnerId
    lastEventId
  }
}

Check record in console output...

After that checkout the other branch "v2/spec-record" and send some new record with schema version 2, repeat from Step 3

mutation {
  createPartnerV2(partnerV2:
    {
      partnerId: "123",
      lastEventId: "test-event-id",
      type: "NaturalPerson",
      firstName: "bai",
      secondName: "xia",
      birthDay: "1984-04-11",
      telephone: "7891234",
      email: "[email protected]",
      age: 18
    })
  {
    partnerId
    lastEventId
  }
}

At the end switch to main branch. There is a stream-app, which uses the current schema to consume data from partner-topic.

You will see, this stream deserializer can handle all schema versions.