作為 kafka 消費者同時消費多個主題

Question

我正在關注這個 url https://cwiki.apache.org/confluence/display/KAFKA/Consumer+Group+Example 中關於並發消費 kafka 主題的示例。

在創建線程池部分，他們有以下代碼

public void run(int a_numThreads) {
    Map<String, Integer> topicCountMap = new HashMap<String, Integer>();
    topicCountMap.put(topic, new Integer(a_numThreads));
    Map<String, List<KafkaStream<byte[], byte[]>>> consumerMap = consumer.createMessageStreams(topicCountMap);
    List<KafkaStream<byte[], byte[]>> streams = consumerMap.get(topic);


    // now launch all the threads
    //
    executor = Executors.newFixedThreadPool(a_numThreads);

    // now create an object to consume the messages
    //
    int threadNumber = 0;
    for (final KafkaStream stream : streams) {
        executor.submit(new ConsumerTest(stream, threadNumber));
        threadNumber++;
    }
}

我可以向 topicCountMap 添加更多主題。 例如，

topicCountMap.put("channel1", new Integer(a_numThreads));
topicCountMap.put("channe2", new Integer(a_numThreads));
topicCountMap.put("channel3", new Integer(a_numThreads));

在上面的代碼中，在我看來，流對象只映射到主題之一

List<KafkaStream<byte[], byte[]>> streams = consumerMap.get(topic);

我不完全確定如何創建多個流對象，每個對象都映射到給定的主題，然后遍歷它們以從每個通道獲取數據並將它們提交給執行程序。

Answer 1

假設你有：

String topic1 = "channel1";
String topic2 = "channel2";
String topic3 = "channel3";

然后，你確實可以這樣做：

topicCountMap.put(topic1, new Integer(a_numThreads_topic1));
topicCountMap.put(topic2, new Integer(a_numThreads_topic2));
topicCountMap.put(topic3, new Integer(a_numThreads_topic3));

一旦您獲得了您的 consumerMap（執行此操作的代碼不會改變），您將能夠檢索每個主題的流：

List<KafkaStream<byte[], byte[]>> topic1_streams = consumerMap.get(topic1);
List<KafkaStream<byte[], byte[]>> topic2_streams = consumerMap.get(topic2);
List<KafkaStream<byte[], byte[]>> topic3_streams = consumerMap.get(topic3);

要從流中使用，您需要創建正確數量的執行程序：

executors_topic1 = Executors.newFixedThreadPool(a_numThreads_topic1);
executors_topic2 = Executors.newFixedThreadPool(a_numThreads_topic2);
executors_topic3 = Executors.newFixedThreadPool(a_numThreads_topic3);

最后 ：

int threadNumber = 0;
for (final KafkaStream stream : topic1_streams) {
    executors_topic1.submit(new ConsumerTest(streams, threadNumber));
    threadNumber++;
}
for (final KafkaStream stream : topic2_streams) {
    executors_topic2.submit(new ConsumerTest(stream, threadNumber));
    threadNumber++;
}
for (final KafkaStream stream : topic3_streams) {
    executor_topic3.submit(new ConsumerTest(stream, threadNumber));
    threadNumber++;
}

當然，這只是給你的想法。 顯然，代碼可以改進。