public class DConsumer {

    public static void main(String[] args) {
        Properties prop = new Properties();
        prop.put("bootstrap.servers","node:9092");
        prop.put("group.id","test8");
        prop.put("key.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
        prop.put("value.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
//如果是之前存在的group.id
        Consumer consumer = new KafkaConsumer(prop);
        TopicPartition p = new TopicPartition("test2",2);
//        指定消費topic的那個分割槽
        consumer.assign(Arrays.asList(p));
//        指定從topic的分割槽的某個offset開始消費
//        consumer.seekToBeginning(Arrays.asList(p));
        consumer.seek(p,5);
//        consumer.subscribe(Arrays.asList("test2"));

//如果是之前不存在的group.id
//        Map<TopicPartition, OffsetAndMetadata> hashMaps = new HashMap<TopicPartition, OffsetAndMetadata>();
//        hashMaps.put(new TopicPartition("test2", 0), new OffsetAndMetadata(0));
//        consumer.commitSync(hashMaps);
//        consumer.subscribe(Arrays.asList("test2"));
        while (true) {
            ConsumerRecords<String, String> c = consumer.poll(100);
            for(ConsumerRecord<String, String> c1: c) {
                System.out.println("Key: " + c1.key() + " Value: " + c1.value() + " Offset: " + c1.offset() + " Partitions: " + c1.partition());

            }
        }
    }
}
 

topic分割槽為 併發操作提供了優勢，訂閱消費關係如圖：