Using KafkaBolt to write to a kafka topic

https://community.hortonworks.com/questions/27187/using-kafkabolt-to-write-to-a-kafka-topic.html

--------------------------------------------------------------------------------------------------------------

I'm building a kafka and storm based streaming application based on the usecase:

1. Application produces JSON message to a kafka topic

2. Kafka storm spout ingests the message and does processing

3. Produce output to another kafka topic using a storm bolt

Here are components and versions I'm using:

Storm: 0.10.0

Kafka_2.10: 0.9.0.1

HDP: 2.3.4

storm-kafka: 0.10.0

I accomplished steps 1 and 2 using "storm.kafka.KafkaSpout" that comes with storm-kafka

I'm trying to use "storm.kafka.KafkaBolt" to write processed data to a topic and I couldn't figure out exactly how to do it. Here is the code snippet:

　　

Properties props = new Properties();
props.put("metadata.broker.list", "192.168.56.102:9092");
props.put("request.required.acks", "1");
props.put("serializer.class", "kafka.serializer.StringEncoder");
 
Config conf = new Config();        
conf.put(KafkaBolt.KAFKA_BROKER_PROPERTIES, props);
 
KafkaBolt kafkaBolt = new KafkaBolt().withTopicSelector(new DefaultTopicSelector("OUTBOUND_TOPIC"))
　　　　　　　　　　　　　　　　　　　　　　.withTupleToKafkaMapper(new FieldNameBasedTupleToKafkaMapper());
 
builder.setBolt("kafka_outbound_bolt", kafkaBolt, 3).shuffleGrouping("process_bolt");

　　

"process_bolt" sends out a tuple that is a serializable domain model object (SubModel.java). I would like to send SubModel.toString() data to the outbound topic. I'm running storm in local cluster mode connected to kafka in a local VM.

1. Am I doing anything wrong ?
2. How do I use "FieldNameBasedTupleToKafkaMapper"?