org.apache.storm.kafka.spout

Class Subscription

java.lang.Object

org.apache.storm.kafka.spout.Subscription

All Implemented Interfaces:

Serializable

Direct Known Subclasses:

ManualPartitionSubscription, NamedSubscription, PatternSubscription

public abstract class Subscription
extends Object
implements Serializable

A subscription to kafka.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
Subscription()

Method Summary

Modifier and Type	Method and Description
abstract String	getTopicsString()
void	refreshAssignment() NOOP is the default behavior, which means that Kafka will internally handle partition assignment.
abstract <K,V> void	subscribe(org.apache.kafka.clients.consumer.KafkaConsumer<K,V> consumer, org.apache.kafka.clients.consumer.ConsumerRebalanceListener listener, TopologyContext context) Subscribe the KafkaConsumer to the proper topics.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Subscription

public Subscription()

Method Detail

subscribe

public abstract <K,V> void subscribe(org.apache.kafka.clients.consumer.KafkaConsumer<K,V> consumer,
                                     org.apache.kafka.clients.consumer.ConsumerRebalanceListener listener,
                                     TopologyContext context)

Subscribe the KafkaConsumer to the proper topics. Implementations must ensure that a given topic partition is always assigned to the same spout task. Adding and removing partitions as necessary is fine, but partitions must not move from one task to another. This constraint is only important for use with the Trident spout.

Parameters:

consumer - the Consumer to get.

listener - the rebalance listener to include in the subscription

getTopicsString

public abstract String getTopicsString()

Returns:

A human-readable string representing the subscribed topics.

refreshAssignment

public void refreshAssignment()

NOOP is the default behavior, which means that Kafka will internally handle partition assignment. If you wish to do manual partition management, you must provide an implementation of this method that will check with kafka for any changes and call the ConsumerRebalanceListener from subscribe to inform the rest of the system of those changes.