摘要：核心實現(xiàn)是這個方法通過不同的模式可以實現(xiàn)發(fā)送即忘忽略返回結(jié)果同步發(fā)送獲取返回的對象，回調(diào)函數(shù)置為異步發(fā)送設置回調(diào)函數(shù)三種消息模式。

Kafka是一款很棒的消息系統(tǒng)，可以看看我之前寫的 后端好書閱讀與推薦來了解一下它的整體設計。今天我們就來深入了解一下它的實現(xiàn)細節(jié)（我fork了一份代碼），首先關注Producer這一方。

要使用kafka首先要實例化一個KafkaProducer，需要有brokerIP、序列化器等必要Properties以及acks（0、1、n）、compression、retries、batch.size等非必要Properties，通過這個簡單的接口可以控制Producer大部分行為，實例化后就可以調(diào)用send方法發(fā)送消息了。

核心實現(xiàn)是這個方法：

public Future send(ProducerRecord record, Callback callback) {
    // intercept the record, which can be potentially modified; this method does not throw exceptions
    ProducerRecord interceptedRecord = this.interceptors.onSend(record);//①
    return doSend(interceptedRecord, callback);//②
}

通過不同的模式可以實現(xiàn)發(fā)送即忘（忽略返回結(jié)果）、同步發(fā)送（獲取返回的future對象，回調(diào)函數(shù)置為null）、異步發(fā)送（設置回調(diào)函數(shù)）三種消息模式。

我們來看看消息類ProducerRecord有哪些屬性：

private final String topic;//主題
private final Integer partition;//分區(qū)
private final Headers headers;//頭
private final K key;//鍵
private final V value;//值
private final Long timestamp;//時間戳

它有多個構(gòu)造函數(shù)，可以適應不同的消息類型：比如有無分區(qū)、有無key等。

①中ProducerInterceptors（有0 ~ 無窮多個，形成一個攔截鏈）對ProducerRecord進行攔截處理（比如打上時間戳，進行審計與統(tǒng)計等操作）

public ProducerRecord onSend(ProducerRecord record) {
    ProducerRecord interceptRecord = record;
    for (ProducerInterceptor interceptor : this.interceptors) {
        try {
            interceptRecord = interceptor.onSend(interceptRecord);
        } catch (Exception e) {
            // 不拋出異常，繼續(xù)執(zhí)行下一個攔截器
            if (record != null)
                log.warn("Error executing interceptor onSend callback for topic: {}, partition: {}", record.topic(), record.partition(), e);
            else
                log.warn("Error executing interceptor onSend callback", e);
        }
    }
    return interceptRecord;
}

如果用戶有定義就進行處理并返回處理后的ProducerRecord，否則直接返回本身。
然后②中doSend真正發(fā)送消息，并且是異步的（源碼太長只保留關鍵）：

private Future doSend(ProducerRecord record, Callback callback) {
    TopicPartition tp = null;
    try {
        // 序列化 key 和 value
        byte[] serializedKey;
        try {
            serializedKey = keySerializer.serialize(record.topic(), record.headers(), record.key());
        } catch (ClassCastException cce) {
        }
        byte[] serializedValue;
        try {
            serializedValue = valueSerializer.serialize(record.topic(), record.headers(), record.value());
        } catch (ClassCastException cce) {
        }
        // 計算分區(qū)獲得主題與分區(qū)
        int partition = partition(record, serializedKey, serializedValue, cluster);
        tp = new TopicPartition(record.topic(), partition);
        // 回調(diào)與事務處理省略。
        Header[] headers = record.headers().toArray();
        // 消息追加到RecordAccumulator中
        RecordAccumulator.RecordAppendResult result = accumulator.append(tp, timestamp, serializedKey,
                serializedValue, headers, interceptCallback, remainingWaitMs);
        // 該批次滿了或者創(chuàng)建了新的批次就要喚醒IO線程發(fā)送該批次了，也就是sender的wakeup方法
        if (result.batchIsFull || result.newBatchCreated) {
            log.trace("Waking up the sender since topic {} partition {} is either full or getting a new batch", record.topic(), partition);
            this.sender.wakeup();
        }
        return result.future;
    } catch (Exception e) {
        // 攔截異常并拋出
        this.interceptors.onSendError(record, tp, e);
        throw e;
    }
}

下面是計算分區(qū)的方法：

private int partition(ProducerRecord record, 
byte[] serializedKey, byte[] serializedValue, Cluster cluster) {
    Integer partition = record.partition();
    // 消息有分區(qū)就直接使用，否則就使用分區(qū)器計算
    return partition != null ?
            partition :
            partitioner.partition(
                    record.topic(), record.key(), serializedKey,
                     record.value(), serializedValue, cluster);
}

默認的分區(qū)器DefaultPartitioner實現(xiàn)方式是如果partition存在就直接使用，否則根據(jù)key計算partition，如果key也不存在就使用round robin算法分配partition。

/**
 * The default partitioning strategy:
 * 

 * 
If a partition is specified in the record, use it
 * 
If no partition is specified but a key is present choose a partition based on a hash of the key
 * 
If no partition or key is present choose a partition in a round-robin fashion
 */
public class DefaultPartitioner implements Partitioner {

    private final ConcurrentMap topicCounterMap = new ConcurrentHashMap<>();
    
    public int partition(String topic, Object key, byte[] keyBytes, Object value, byte[] valueBytes, Cluster cluster) {
        List partitions = cluster.partitionsForTopic(topic);
        int numPartitions = partitions.size();
        if (keyBytes == null) {//key為空 
            int nextValue = nextValue(topic);
            List availablePartitions = cluster.availablePartitionsForTopic(topic);//可用的分區(qū)
            if (availablePartitions.size() > 0) {//有分區(qū)，取模就行
                int part = Utils.toPositive(nextValue) % availablePartitions.size();
                return availablePartitions.get(part).partition();
            } else {// 無分區(qū)，
                return Utils.toPositive(nextValue) % numPartitions;
            }
        } else {// key 不為空，計算key的hash并取模獲得分區(qū)
            return Utils.toPositive(Utils.murmur2(keyBytes)) % numPartitions;
        }
    }

    private int nextValue(String topic) {
        AtomicInteger counter = topicCounterMap.get(topic);
        if (null == counter) {
            counter = new AtomicInteger(ThreadLocalRandom.current().nextInt());
            AtomicInteger currentCounter = topicCounterMap.putIfAbsent(topic, counter);
            if (currentCounter != null) {
                counter = currentCounter;
            }
        }
        return counter.getAndIncrement();//返回并加一，在取模的配合下就是round robin
    }
}

以上就是發(fā)送消息的邏輯處理，接下來我們再看看消息發(fā)送的物理處理。

Sender（是一個Runnable，被包含在一個IO線程ioThread中，該線程不斷從RecordAccumulator隊列中的讀取消息并通過Selector將數(shù)據(jù)發(fā)送給Broker）的wakeup方法，實際上是KafkaClient接口的wakeup方法，由NetworkClient類實現(xiàn)，采用了NIO，也就是java.nio.channels.Selector.wakeup()方法實現(xiàn)。

Sender的run中主要邏輯是不停執(zhí)行準備消息和等待消息：

long pollTimeout = sendProducerData(now);//③
client.poll(pollTimeout, now);//④

③完成消息設置并保存到信道中，然后監(jiān)聽感興趣的key，由KafkaChannel實現(xiàn)。

public void setSend(Send send) {
    if (this.send != null)
        throw new IllegalStateException("Attempt to begin a send operation with prior send operation still in progress, connection id is " + id);
    this.send = send;
    this.transportLayer.addInterestOps(SelectionKey.OP_WRITE);
}

// transportLayer的一種實現(xiàn)中的相關方法
public void addInterestOps(int ops) {
    key.interestOps(key.interestOps() | ops);
}

④主要是Selector的poll，其select被wakeup喚醒：

public void poll(long timeout) throws IOException {
    /* check ready keys */
    long startSelect = time.nanoseconds();
    int numReadyKeys = select(timeout);//wakeup使其停止阻塞
    long endSelect = time.nanoseconds();
    this.sensors.selectTime.record(endSelect - startSelect, time.milliseconds());

    if (numReadyKeys > 0 || !immediatelyConnectedKeys.isEmpty() || dataInBuffers) {
        Set readyKeys = this.nioSelector.selectedKeys();

        // Poll from channels that have buffered data (but nothing more from the underlying socket)
        if (dataInBuffers) {
            keysWithBufferedRead.removeAll(readyKeys); //so no channel gets polled twice
            Set toPoll = keysWithBufferedRead;
            keysWithBufferedRead = new HashSet<>(); //poll() calls will repopulate if needed
            pollSelectionKeys(toPoll, false, endSelect);
        }

        // Poll from channels where the underlying socket has more data
        pollSelectionKeys(readyKeys, false, endSelect);
        // Clear all selected keys so that they are included in the ready count for the next select
        readyKeys.clear();

        pollSelectionKeys(immediatelyConnectedKeys, true, endSelect);
        immediatelyConnectedKeys.clear();
    } else {
        madeReadProgressLastPoll = true; //no work is also "progress"
    }

    long endIo = time.nanoseconds();
    this.sensors.ioTime.record(endIo - endSelect, time.milliseconds());
}

其中pollSelectionKeys方法會調(diào)用如下方法完成消息發(fā)送：

public Send write() throws IOException {
    Send result = null;
    if (send != null && send(send)) {
        result = send;
        send = null;
    }
    return result;
}

private boolean send(Send send) throws IOException {
    send.writeTo(transportLayer);
    if (send.completed())
        transportLayer.removeInterestOps(SelectionKey.OP_WRITE);
    return send.completed();
}

Send是一次數(shù)據(jù)發(fā)包，一般由ByteBufferSend或者MultiRecordsSend實現(xiàn)，其writeTo調(diào)用transportLayer的write方法，一般由PlaintextTransportLayer或者SslTransportLayer實現(xiàn)，區(qū)分是否使用ssl：

public long writeTo(GatheringByteChannel channel) throws IOException {
    long written = channel.write(buffers);
    if (written < 0)
        throw new EOFException("Wrote negative bytes to channel. This shouldn"t happen.");
    remaining -= written;
    pending = TransportLayers.hasPendingWrites(channel);
    return written;
}

public int write(ByteBuffer src) throws IOException {
    return socketChannel.write(src);
}

到此就把Producer的業(yè)務相關邏輯處理和非業(yè)務相關的網(wǎng)絡 2方面的主要流程梳理清楚了。其他額外的功能是通過一些配置保證的。

比如順序保證就是max.in.flight.requests.per.connection，InFlightRequests的doSend會進行判斷（由NetworkClient的canSendRequest調(diào)用），只要該參數(shù)設為1即可保證當前包未確認就不能發(fā)送下一個包從而實現(xiàn)有序性

public boolean canSendMore(String node) {
    Deque queue = requests.get(node);
    return queue == null || queue.isEmpty() ||
           (queue.peekFirst().send.completed() && queue.size() < this.maxInFlightRequestsPerConnection);
}

再比如可靠性，通過設置acks，Sender中sendProduceRequest的clientRequest加入了回調(diào)函數(shù)：

    RequestCompletionHandler callback = new RequestCompletionHandler() {
        public void onComplete(ClientResponse response) {
            handleProduceResponse(response, recordsByPartition, time.milliseconds());//調(diào)用completeBatch
        }
    };
    
     /**
     * 完成或者重試投遞，這里如果acks不對就會重試
     *
     * @param batch The record batch
     * @param response The produce response
     * @param correlationId The correlation id for the request
     * @param now The current POSIX timestamp in milliseconds
     */
    private void completeBatch(ProducerBatch batch, ProduceResponse.PartitionResponse response, long correlationId,
                               long now, long throttleUntilTimeMs) {
    }
    
    public class ProduceResponse extends AbstractResponse {
      /**
         * Possible error code:
         * INVALID_REQUIRED_ACKS (21)
         */
    }
    

kafka源碼一層一層包裝很多，錯綜復雜，如有錯誤請大家不吝賜教。