Spring Kafka - Event sourcing - Example of how to query some entity state using Kafka + KafkaStreams API

Question

I am using Kafka to implement an architecture based on event sourcing.

Suppose I store the events in JSON format:

{"name": "ProductAdded", "productId":"1", quantity=3, dateAdded="2017-04-04" }

I would like to implement a query to get the quantity of a product with productId=X in a certain date.

Can you show an approximate implementation of this query with Spring Kafka KStreams?

UPDATE: I've advanced a little with this using Spring Kafka KStreams, but I'm getting a deserialization error.

This is my Spring Cloud Stream Kafka Producer:

public interface ProductProducer{

    final String OUTPUT = "productsOut";

    @Output(ProductProducer.OUTPUT)
    MessageChannel output();

}

Config:

spring:
  application:
    name: product-generator-service
  cloud:
    stream:
      kafka:
        binder:
          brokers:
          - kafka
          zk-nodes:
          - kafka
        bindings:
          productsOut:
            producer:
              sync: true
      bindings:
        productsOut: 
          destination: orders
          content-type: application/json

I send a message using the following code, that serializes a Map into a JSON object correctly:

Map<String, Object> event = new HashMap<>();
event.put("name", "ProductCreated");
event.put("productId", product.getId());
event.put("quantity", product.getQuantity());
event.put("dateAdded", new Date());
        productProducer.output().send(MessageBuilder.withPayload(event).build(), 500);

MessageBuilder.withPayload(event).build() -> GenericMessage [payload={quantity=1, productId=1, name=ProductCreated, dateAdded="xxxxx"}, headers={id=fc531176-e3e9-61b8-40e3-08074fabee4d, timestamp=1499845483095}]

In the ProductService application I am able to read this message using a Spring Cloud Stream listener:

@Component
public class ProductListener{

    @StreamListener(ProductConsumer.INPUT)
    public void handleProduct(Map<String, Object> event){

However with KStream I am getting a deserialization error:

@Configuration
public class KStreamsConfig {

    private static final String STREAMING_TOPIC1 = "orders";

    @Bean(name = KafkaStreamsDefaultConfiguration.DEFAULT_STREAMS_CONFIG_BEAN_NAME)
    public StreamsConfig kStreamsConfigs() {
        Map<String, Object> props = new HashMap<>();
        props.put(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
        props.put(StreamsConfig.APPLICATION_ID_CONFIG, "product-service-kstream");
        props.put(StreamsConfig.KEY_SERDE_CLASS_CONFIG, Serdes.String().getClass().getName());
        //props.put(StreamsConfig.VALUE_SERDE_CLASS_CONFIG, Serdes.serdeFrom(jsonSerializer, jsonDeserializer).getClass().getName());
        props.put(StreamsConfig.TIMESTAMP_EXTRACTOR_CLASS_CONFIG, WallclockTimestampExtractor.class.getName());
        return new StreamsConfig(props);
    }

    @Bean
    public FactoryBean<KStreamBuilder> myKStreamBuilder(StreamsConfig streamsConfig) {
        return new KStreamBuilderFactoryBean(streamsConfig);
    }

    @Bean
    public KStream<?, ?> kStream(KStreamBuilder kStreamBuilder) {

        Serde<Integer> integerSerde = Serdes.Integer();
        final Serializer<JsonNode> jsonSerializer = new JsonSerializer();
        final Deserializer<JsonNode> jsonDeserializer = new JsonDeserializer();
        final Serde<JsonNode> jsonSerde = Serdes.serdeFrom(jsonSerializer, jsonDeserializer);

        KStream<Integer, JsonNode> stream = kStreamBuilder.stream(null, integerSerde, jsonSerde, STREAMING_TOPIC1);
        stream.print();
        return stream;
    }

}

Exception:

org.apache.kafka.common.errors.SerializationException: com.fasterxml.jackson.core.JsonParseException: Unrecognized token 'ÿ': was expecting ('true', 'false' or 'null')
 at [Source: [B@288e4e9a; line: 1, column: 4]
Caused by: com.fasterxml.jackson.core.JsonParseException: Unrecognized token 'ÿ': was expecting ('true', 'false' or 'null')
 at [Source: [B@288e4e9a; line: 1, column: 4]
    at com.fasterxml.jackson.core.JsonParser._constructError(JsonParser.java:1702)
    at com.fasterxml.jackson.core.base.ParserMinimalBase._reportError(ParserMinimalBase.java:558)
    at com.fasterxml.jackson.core.json.UTF8StreamJsonParser._reportInvalidToken(UTF8StreamJsonParser.java:3528)
    at com.fasterxml.jackson.core.json.UTF8StreamJsonParser._handleUnexpectedValue(UTF8StreamJsonParser.java:2686)
    at com.fasterxml.jackson.core.json.UTF8StreamJsonParser._nextTokenNotInObject(UTF8StreamJsonParser.java:878)
    at com.fasterxml.jackson.core.json.UTF8StreamJsonParser.nextToken(UTF8StreamJsonParser.java:772)
    at com.fasterxml.jackson.databind.ObjectMapper._initForReading(ObjectMapper.java:3834)
    at com.fasterxml.jackson.databind.ObjectMapper._readMapAndClose(ObjectMapper.java:3783)
    at com.fasterxml.jackson.databind.ObjectMapper.readTree(ObjectMapper.java:2404)
    at org.apache.kafka.connect.json.JsonDeserializer.deserialize(JsonDeserializer.java:50)
    at org.apache.kafka.connect.json.JsonDeserializer.deserialize(JsonDeserializer.java:30)
    at org.apache.kafka.streams.processor.internals.SourceNode.deserializeValue(SourceNode.java:46)
    at org.apache.kafka.streams.processor.internals.SourceNodeRecordDeserializer.deserialize(SourceNodeRecordDeserializer.java:44)
    at org.apache.kafka.streams.processor.internals.RecordQueue.addRawRecords(RecordQueue.java:85)
    at org.apache.kafka.streams.processor.internals.PartitionGroup.addRawRecords(PartitionGroup.java:117)
    at org.apache.kafka.streams.processor.internals.StreamTask.addRecords(StreamTask.java:158)
    at org.apache.kafka.streams.processor.internals.StreamThread.runLoop(StreamThread.java:605)
    at org.apache.kafka.streams.processor.internals.StreamThread.run(StreamThread.java:361)

UPDATE 2:

To find out what was getting to the KStream I changed into String deserializers for both the key and the value and this is what is being printed:

KStream<Integer, String> stream = kStreamBuilder.stream(null, integerSerde, stringSerde, STREAMING_TOPIC1);

Printed value:

[KSTREAM-SOURCE-0000000000]: null , �contentType

Why am I not getting the JSON string?

UPDATE 3: I fixed the deserialization problem, the cause was that the message producer (Spring Cloud Stream) adds some headers by default as part of the payload. I only had to disable this header inclusion to start receiving the messages correctly in Kafka Streams:

spring:
  application:
    name: product-service
  cloud:
    stream:
      kafka:
        binder:
          brokers:
          - kafka
          zk-nodes:
          - kafka
        bindings:
          productsOut:
            producer:
              sync: true
      bindings:
        productsIn:
          group: product-service 
          destination: orders
          consumer:
            max-attempts: 5
            header-mode: raw
        productsOut: 
          destination: orders
          content-type: application/json
          producer:
            header-mode: raw

KStream definition:

KStream<Integer, JsonNode> stream = kStreamBuilder.stream(integerSerde, jsonSerde, STREAMING_TOPIC1);

Output:

[KSTREAM-SOURCE-0000000000]: null , {"quantity":0,"productId":0,"name":"ProductCreated","dateAdded":1499930385450}

Now with everything set up correctly: how can I implement an interactive query like the one I need? -> Get the quantity of a product with productId=X in a certain date

Answer 1

I managed to solve this using a mixture of Spring Cloud Streams (to generate the messages) and Spring Kafka to process the KafkaStreams and implement the interactive queries (IMPORTANT: notice the Question UPDATE 3: to be able to combine both):

Kafka Streams Config:

@Configuration
public class KStreamsConfig {

    private static final String STREAMING_TOPIC1 = "orders";

    @Bean(name = KafkaStreamsDefaultConfiguration.DEFAULT_STREAMS_CONFIG_BEAN_NAME)
    public StreamsConfig kStreamsConfigs() {
        Map<String, Object> props = new HashMap<>();
        props.put(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
        props.put(StreamsConfig.APPLICATION_ID_CONFIG, "product-service-streams");
        props.put(StreamsConfig.KEY_SERDE_CLASS_CONFIG, Serdes.Integer().getClass().getName());
        props.put(StreamsConfig.VALUE_SERDE_CLASS_CONFIG, Serdes.Integer().getClass().getName());
        //props.put(StreamsConfig.VALUE_SERDE_CLASS_CONFIG, Serdes.serdeFrom(jsonSerializer, jsonDeserializer).getClass().getName());
        props.put(StreamsConfig.TIMESTAMP_EXTRACTOR_CLASS_CONFIG, WallclockTimestampExtractor.class.getName());
        return new StreamsConfig(props);
    }

    @Bean
    public KStreamBuilderFactoryBean myKStreamBuilder(StreamsConfig streamsConfig) {
        return new KStreamBuilderFactoryBean(streamsConfig);
    }

    @Bean
    public KStream<?, ?> kStream(KStreamBuilder kStreamBuilder, KStreamBuilderFactoryBean kStreamBuilderFactoryBean) {

        Serde<Integer> integerSerde = Serdes.Integer();
        final Serializer<JsonNode> jsonSerializer = new JsonSerializer();
        final Deserializer<JsonNode> jsonDeserializer = new JsonDeserializer();
        final Serde<JsonNode> jsonSerde = Serdes.serdeFrom(jsonSerializer, jsonDeserializer);

        KStream<Integer, JsonNode> stream = kStreamBuilder.stream(integerSerde, jsonSerde, STREAMING_TOPIC1);

        stream.map( (key, value) -> {
            return new KeyValue<>(value.get("productId").asInt(), value.get("quantity").asInt());
        }).groupByKey().reduce( (v1, v2) -> v1 + v2, "ProductsStock");

        stream.print();
        return stream;
    }

}

Notice how I generate a KTable store ProductsStock that I'll query later in the service.

ProductService:

@Autowired
private KStreamBuilderFactoryBean kStreamBuilderFactoryBean;

@Override
    public Integer getProductStock(Integer id) {
        KafkaStreams streams = kStreamBuilderFactoryBean.getKafkaStreams();
        ReadOnlyKeyValueStore<Integer, Integer> keyValueStore =
        streams.store("ProductsStock", QueryableStoreTypes.keyValueStore());
        return keyValueStore.get(id);
}

Answer 2

The upcoming 1.3.0.M1 release of the spring cloud stream kafka binder will have support for kstream binding. There is a PR where you can track the progress of this initiative.

Here is a more general sample (WordCount) using the KStream binder: WordCount Sample using Spring Cloud Stream support for Kafka Streams

With this, you can acheive what you are looking for in the following way.

This StreamListener method will listen from a Kafka topic and write to another topic with the count for products with ID equal to 123 in a time window of last 30 seconds.

@SpringBootApplication
@EnableBinding(KStreamProcessor.class)
public class ProductCountApplication {

  public static final int = 123;

  @StreamListener("input")
  @SendTo("output")
  public KStream<?, String> process(KStream<?, Product> input) {

        return input
                .filter((key, product) -> product.getID() == PRODUCT_ID)
                .map((k,v) -> new KeyValue<>(v, v))
                .groupByKey(new JsonSerde<>(Product.class), new JsonSerde<>(Product.class))
                .count(TimeWindows.of(30000), "product-store")
                .toStream()
                .map((w,c) -> new KeyValue<>(null, "Product with id 123 count: " + c));
  }

}

Here is the application.yml used:

spring.cloud.stream.kstream.binder.streamConfiguration:
  key.serde: org.apache.kafka.common.serialization.Serdes$StringSerde # Use a native Kafka Serde for the key
  value.serde: org.apache.kafka.common.serialization.Serdes$StringSerde # Use a native Kafka Serde for the value
spring.cloud.stream.bindings.output.producer:
  headerMode: raw # Incoming data has no embedded headers
  useNativeEncoding: true # Write data using the native Serde
spring.cloud.stream.bindings.input.consumer:
  headerMode: raw # Outbound data has no embedded headers

When you run the program, you need to pass in the input/output destinations (topics):

--spring.cloud.stream.bindings.input.destination=products 
--spring.cloud.stream.bindings.output.destination=counts