Kafka核心参数详解

消费者分组消费机制

生产者拦截器机制

消息序列化机制

消息分区路由机制

消费者分组消费机制

在Consumer中，需要指定一个GROUP_ID_CONFIG属性，这表示当前Consumer所属的消费者组。他的描述是这样的：

 public static final String GROUP_ID_CONFIG = "group.id";
    public static final String GROUP_ID_DOC = "A unique string that identifies the consumer group this consumer belongs to. This property is required if the consumer uses either the group management functionality by using <code>subscribe(topic)</code> or the Kafka-based offset management strategy.";

        Properties props = new Properties();
        //kafka地址
        props.put(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, BOOTSTRAP_SERVERS);
        //每个消费者要指定一个group
        props.put(ConsumerConfig.GROUP_ID_CONFIG, "test");
        //key序列化类
        props.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, "org.apache.kafka.common.serialization.StringDeserializer");
        //value序列化类
        props.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, "org.apache.kafka.common.serialization.StringDeserializer");
        Consumer<String, String> consumer = new KafkaConsumer<>(props);

通过对properties集合设置，在创建消费者时将指定集合传入消费者。

对于Consumer，如果需要在subcribe时使用组管理功能以及Kafka提供的offset管理策略，那就必须要配置GROUP_ID_CONFIG属性。这个分组消费机制简单描述就是这样的：

生产者往Topic下发消息时，会尽量均匀的将消息发送到Topic下的各个Partition当中。而这个消息，会向所有订阅了该Topic的消费者推送。推送时，每个ConsumerGroup中只会推送一份。也就是同一个消费者组中的多个消费者实例，只会共同消费一个消息副本。而不同消费者组之间，会重复消费消息副本。这就是消费者组的作用。

与之相关的还有Offset偏移量。这个偏移量表示每个消费者组在每个Partiton中已经消费处理的进度。

bin\windows\kafka-consumer-groups.bat --bootstrap-server localhost:9092 --describe --group test

这个Offset偏移量，需要消费者处理完成后主动向Kafka的Broker提交。提交完成后，Broker就会更新消费进度，表示这个消息已经被这个消费者组处理完了。但是如果消费者没有提交Offset，Broker就会认为这个消息还没有被处理过，就会重新往对应的消费者组进行推送，不过这次，一般会尽量推送给同一个消费者组当中的其他消费者实例。

在之前文章示例当中，是通过业务端主动调用Consumer的commitAsync方法或者commitSync方法主动提交的，Kafka中自然也提供了自动提交Offset的方式。使用自动提交，只需要在Comsumer中配置ENABLE_AUTO_COMMIT_CONFIG属性即可。

public static final String ENABLE_AUTO_COMMIT_CONFIG = "enable.auto.commit";
private static final String ENABLE_AUTO_COMMIT_DOC = "If true the consumer's offset will be periodically committed in the background.";

Offset是Kafka进行消息推送控制的关键之处。

生产者拦截器机制

生产者拦截机制允许客户端在生产者在消息发送到Kafka集群之前，对消息进行拦截，甚至可以修改消息内容。

 public static final String INTERCEPTOR_CLASSES_CONFIG = "interceptor.classes";
 public static final String INTERCEPTOR_CLASSES_DOC = "A list of classes to use as interceptors. "
                                                        + "Implementing the <code>org.apache.kafka.clients.producer.ProducerInterceptor</code> interface allows you to intercept (and possibly mutate) the records "
                                                        + "received by the producer before they are published to the Kafka cluster. By default, there are no interceptors.";

按照说明，我们可以自定义一个拦截器实现类：

public class MyInterceptor implements ProducerInterceptor {
    //发送消息时触发
    @Override
    public ProducerRecord onSend(ProducerRecord producerRecord) {
        System.out.println("prudocerRecord : " + producerRecord.toString());
        return producerRecord;
    }

    //收到服务端响应时触发
    @Override
    public void onAcknowledgement(RecordMetadata recordMetadata, Exception e) {
        System.out.println("acknowledgement recordMetadata:"+recordMetadata.toString());
    }

    //连接关闭时触发
    @Override
    public void close() {
        System.out.println("producer closed");
    }

    //整理配置项
    @Override
    public void configure(Map<String, ?> map) {
        System.out.println("=====config start======");
        for (Map.Entry<String, ?> entry : map.entrySet()) {
            System.out.println("entry.key:"+entry.getKey()+" === entry.value: "+entry.getValue());
        }
        System.out.println("=====config end======");
    }
}

在生产者中指定拦截器类（多个拦截器类，用逗号隔开）

 props.put(ProducerConfig.INTERCEPTOR_CLASSES_CONFIG,"com.kfk.basic.MyInterceptor");

消息序列化机制

在上篇简单示例中，Producer指定了两个属性KEY_SERIALIZER_CLASS_CONFIG和VALUE_SERIALIZER_CLASS_CONFIG，对于这两个属性，在ProducerConfig中都有配套的说明属性。

    public static final String KEY_SERIALIZER_CLASS_CONFIG = "key.serializer";
    public static final String KEY_SERIALIZER_CLASS_DOC = "Serializer class for key that implements the <code>org.apache.kafka.common.serialization.Serializer</code> interface.";
    public static final String VALUE_SERIALIZER_CLASS_CONFIG = "value.serializer";
    public static final String VALUE_SERIALIZER_CLASS_DOC = "Serializer class for value that implements the <code>org.apache.kafka.common.serialization.Serializer</code> interface.";

通过这两个参数，可以指定消息生产者如何将消息的key和value序列化成二进制数据。在Kafka的消息定义中，key和value的作用是不同的。

1. key是用来进行分区的可选项。Kafka通过key来判断消息要分发到哪个Partition。

如果没有填写key，那么Kafka会使Round-robin轮询的方式，自动选择Partition。

如果填写了key，那么会通过声明的Serializer序列化接口，将key转换成一个byte[]数组，然后对key进行hash，选择Partition。这样可以保证key相同的消息会分配到相同的Partition中。

2. Value是业务上比较关心的消息。Kafka同样需要将Value对象通过Serializer序列化接口，将Key转换成byte[]数组，这样才能比较好的在网络上传输Value信息，以及将Value信息落盘到操作系统的文件当中。

生产者要对消息进行序列化，那么消费者拉取消息时，自然需要进行反序列化。所以，在Consumer中，也有反序列化的两个配置：

    public static final String KEY_DESERIALIZER_CLASS_CONFIG = "key.deserializer";
    public static final String KEY_DESERIALIZER_CLASS_DOC = "Deserializer class for key that implements the <code>org.apache.kafka.common.serialization.Deserializer</code> interface.";
    public static final String VALUE_DESERIALIZER_CLASS_CONFIG = "value.deserializer";
    public static final String VALUE_DESERIALIZER_CLASS_DOC = "Deserializer class for value that implements the <code>org.apache.kafka.common.serialization.Deserializer</code> interface.";

在Kafka中，对于常用的一些基础数据类型，都已经提供了对应的实现类。但是，如果需要使用一些自定义的消息格式，比如自己定制的POJO，就需要定制具体的实现类了。

消息分区路由机制

在Producer中，可以指定一个Partitioner来对消息进行分配。

public static final String PARTITIONER_CLASS_CONFIG = "partitioner.class";
    private static final String PARTITIONER_CLASS_DOC = "A class to use to determine which partition to be send to when produce the records. Available options are:" +
        "<ul>" +
            "<li>If not set, the default partitioning logic is used. " +
        "This strategy will try sticking to a partition until at least " + BATCH_SIZE_CONFIG + " bytes is produced to the partition. It works with the strategy:" +
                "<ul>" +
                    "<li>If no partition is specified but a key is present, choose a partition based on a hash of the key</li>" +
                    "<li>If no partition or key is present, choose the sticky partition that changes when at least " + BATCH_SIZE_CONFIG + " bytes are produced to the partition.</li>" +
                "</ul>" +
            "</li>" +
            "<li><code>org.apache.kafka.clients.producer.RoundRobinPartitioner</code>: This partitioning strategy is that " +
        "each record in a series of consecutive records will be sent to a different partition(no matter if the 'key' is provided or not), " +
        "until we run out of partitions and start over again. Note: There's a known issue that will cause uneven distribution when new batch is created. " +
        "Please check KAFKA-9965 for more detail." +
            "</li>" +
        "</ul>" +
        "<p>Implementing the <code>org.apache.kafka.clients.producer.Partitioner</code> interface allows you to plug in a custom partitioner.";

Kafka是通过一个Partitioner接口的具体实现来决定一个消息如何根据Key分配到对应的Partition上的。在之前的3.2.0版本，Kafka提供了三种默认的Partitioner实现类，RoundRobinPartitioner，DefaultPartitioner和UniformStickyPartitioner。目前后面两种实现已经标记为过期，被替换成了默认的实现机制。

可以自行指定一个Partitioner实现类，定制分区逻辑。在Partitioner接口中，核心要实现的就是partition方法。根据相关信息，选择一个Partition。比如用key对partition的个数取模之类的。而Topic下的所有Partition信息在源码中都在cluster参数中。

//获取所有的Partition信息。
List<PartitionInfo> partitions = cluster.partitionsForTopic(topic);

在Consumer中，可以指定一个PARTITION_ASSIGNMENT_STRATEGY分区分配策略，决定如何在多个Consumer实例和多个Partitioner之间建立关联关系。

  public static final String PARTITION_ASSIGNMENT_STRATEGY_CONFIG = "partition.assignment.strategy";
    private static final String PARTITION_ASSIGNMENT_STRATEGY_DOC = "A list of class names or class types, " +
        "ordered by preference, of supported partition assignment strategies that the client will use to distribute " +
        "partition ownership amongst consumer instances when group management is used. Available options are:" +
        "<ul>" +
        "<li><code>org.apache.kafka.clients.consumer.RangeAssignor</code>: Assigns partitions on a per-topic basis.</li>" +
        "<li><code>org.apache.kafka.clients.consumer.RoundRobinAssignor</code>: Assigns partitions to consumers in a round-robin fashion.</li>" +
        "<li><code>org.apache.kafka.clients.consumer.StickyAssignor</code>: Guarantees an assignment that is " +
        "maximally balanced while preserving as many existing partition assignments as possible.</li>" +
        "<li><code>org.apache.kafka.clients.consumer.CooperativeStickyAssignor</code>: Follows the same StickyAssignor " +
        "logic, but allows for cooperative rebalancing.</li>" +
        "</ul>" +
        "<p>The default assignor is [RangeAssignor, CooperativeStickyAssignor], which will use the RangeAssignor by default, " +
        "but allows upgrading to the CooperativeStickyAssignor with just a single rolling bounce that removes the RangeAssignor from the list.</p>" +
        "<p>Implementing the <code>org.apache.kafka.clients.consumer.ConsumerPartitionAssignor</code> " +
        "interface allows you to plug in a custom assignment strategy.</p>";

Kafka默认提供了三种消费者的分区分配策略：

range策略：
比如一个Topic有10个Partiton(partition 0~9) 一个消费者组下有三个Consumer(consumer1~3)。Range策略就会将分区0~3分给一个Consumer，4~6给一个Consumer，7~9给一个Consumer。
round-robin策略：
轮询分配策略，可以理解为在Consumer中一个一个轮流分配分区。比如0，3，6，9分区给一个Consumer，1，4，7分区给一个Consumer，然后2，5，8给一个Consumer
sticky策略：粘性策略。这个策略有两个原则：
1、在开始分区时，尽量保持分区的分配均匀。比如按照Range策略分(这一步实际上是随机的)。2、分区的分配尽可能的与上一次分配的保持一致。比如在range分区的情况下，第三个Consumer的服务宕机了，那么按照sticky策略，就会保持consumer1和consumer2原有的分区分配情况。然后将consumer3分配的7~9分区尽量平均的分配到另外两个consumer上。这种粘性策略可以很好的保持Consumer的数据稳定性。

生产者消息缓存机制

Kafka生产者为了避免高并发请求对服务端造成过大压力，每次发消息时并不是一条一条发往服务端，而是增加了一个高速缓存，将消息集中到缓存后，批量进行发送。这种缓存机制也是高并发处理时非常常用的一种机制。

Kafka的消息缓存机制涉及到KafkaProducer中的两个关键组件： accumulator 和 sender。

//1.记录累加器
int batchSize = Math.max(1, config.getInt(ProducerConfig.BATCH_SIZE_CONFIG));
this.accumulator = new RecordAccumulator(logContext,batchSize,this.compressionType,lingerMs(config),retryBackoffMs,deliveryTimeoutMs, partitionerConfig,metrics,PRODUCER_METRIC_GROUP_NAME,time,apiVersions,transactionManager,new BufferPool(this.totalMemorySize, batchSize, metrics, time, PRODUCER_METRIC_GROUP_NAME));
//2. 数据发送线程
this.sender = newSender(logContext, kafkaClient, this.metadata);

其中RecordAccumulator，就是Kafka生产者的消息累加器。KafkaProducer要发送的消息都会在ReocrdAccumulator中缓存起来，然后再分批发送给kafka broker。在RecordAccumulator中，会针对每一个Partition，维护一个Deque双端队列，这些Dequeue队列基本上是和Kafka服务端的Topic下的Partition对应的。每个Dequeue里会放入若干个ProducerBatch数据。KafkaProducer每次发送的消息，都会根据key分配到对应的Deque队列中。然后每个消息都会保存在这些队列中的某一个ProducerBatch中。而消息分发的规则，就是由上面的Partitioner组件完成的。

主要涉及到两个参数

//RecordAccumulator缓冲区大小
public static final String BUFFER_MEMORY_CONFIG = "buffer.memory";
private static final String BUFFER_MEMORY_DOC = "The total bytes of memory the producer can use to buffer records waiting to be sent to the server. If records are "
                                                    + "sent faster than they can be delivered to the server the producer will block for <code>" + MAX_BLOCK_MS_CONFIG + "</code> after which it will throw an exception."
                                                    + "<p>"
                                                    + "This setting should correspond roughly to the total memory the producer will use, but is not a hard bound since "
                                                    + "not all memory the producer uses is used for buffering. Some additional memory will be used for compression (if "
                                                    + "compression is enabled) as well as for maintaining in-flight requests.";

//缓冲区每一个batch的大小
public static final String BATCH_SIZE_CONFIG = "batch.size";
private static final String BATCH_SIZE_DOC = "The producer will attempt to batch records together into fewer requests whenever multiple records are being sent"
                                                 + " to the same partition. This helps performance on both the client and the server. This configuration controls the "
                                                 + "default batch size in bytes. "
                                                 + "<p>"
                                                 + "No attempt will be made to batch records larger than this size. "
                                                 + "<p>"
                                                 + "Requests sent to brokers will contain multiple batches, one for each partition with data available to be sent. "
                                                 + "<p>"
                                                 + "A small batch size will make batching less common and may reduce throughput (a batch size of zero will disable "
                                                 + "batching entirely). A very large batch size may use memory a bit more wastefully as we will always allocate a "
                                                 + "buffer of the specified batch size in anticipation of additional records."
                                                 + "<p>"
                                                 + "Note: This setting gives the upper bound of the batch size to be sent. If we have fewer than this many bytes accumulated "
                                                 + "for this partition, we will 'linger' for the <code>linger.ms</code> time waiting for more records to show up. "
                                                 + "This <code>linger.ms</code> setting defaults to 0, which means we'll immediately send out a record even the accumulated "
                                                 + "batch size is under this <code>batch.size</code> setting.";

接下来，sender就是KafkaProducer中用来发送消息的一个单独的线程。从这里可以看到，每个KafkaProducer对象都对应一个sender线程。他会负责将RecordAccumulator中的消息发送给Kafka。

Sender也并不是一次就把RecordAccumulator中缓存的所有消息都发送出去，而是每次只拿一部分消息。他只获取RecordAccumulator中缓存内容达到BATCH_SIZE_CONFIG大小的ProducerBatch消息。当然，如果消息比较少，ProducerBatch中的消息大小长期达不到BATCH_SIZE_CONFIG的话，Sender也不会一直等待。最多等待LINGER_MS_CONFIG时长。然后就会将ProducerBatch中的消息读取出来。LINGER_MS_CONFIG默认值是0。

然后，Sender对读取出来的消息，会以Broker为key，缓存到一个对应的队列当中。这些队列当中的消息就称为InflightRequest。接下来这些Inflight就会一一发往Kafka对应的Broker中，直到收到Broker的响应，才会从队列中移除。这些队列也并不会无限缓存，最多缓存MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION(默认值为5)个请求。

涉及到的几个主要参数如下：

 public static final String LINGER_MS_CONFIG = "linger.ms";
    private static final String LINGER_MS_DOC = "The producer groups together any records that arrive in between request transmissions into a single batched request. "
                                                + "Normally this occurs only under load when records arrive faster than they can be sent out. However in some circumstances the client may want to "
                                                + "reduce the number of requests even under moderate load. This setting accomplishes this by adding a small amount "
                                                + "of artificial delay&mdash;that is, rather than immediately sending out a record, the producer will wait for up to "
                                                + "the given delay to allow other records to be sent so that the sends can be batched together. This can be thought "
                                                + "of as analogous to Nagle's algorithm in TCP. This setting gives the upper bound on the delay for batching: once "
                                                + "we get <code>" + BATCH_SIZE_CONFIG + "</code> worth of records for a partition it will be sent immediately regardless of this "
                                                + "setting, however if we have fewer than this many bytes accumulated for this partition we will 'linger' for the "
                                                + "specified time waiting for more records to show up. This setting defaults to 0 (i.e. no delay). Setting <code>" + LINGER_MS_CONFIG + "=5</code>, "
                                                + "for example, would have the effect of reducing the number of requests sent but would add up to 5ms of latency to records sent in the absence of load.";



public static final String MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION = "max.in.flight.requests.per.connection";
private static final String MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION_DOC = "The maximum number of unacknowledged requests the client will send on a single connection before blocking."
                                                                            + " Note that if this configuration is set to be greater than 1 and <code>enable.idempotence</code> is set to false, there is a risk of"
                                                                            + " message reordering after a failed send due to retries (i.e., if retries are enabled); "
                                                                            + " if retries are disabled or if <code>enable.idempotence</code> is set to true, ordering will be preserved."
                                                                            + " Additionally, enabling idempotence requires the value of this configuration to be less than or equal to " + MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION_FOR_IDEMPOTENCE + "."
                                                                            + " If conflicting configurations are set and idempotence is not explicitly enabled, idempotence is disabled. ";

最后，Sender会通过其中的一个Selector组件完成与Kafka的IO请求，并接收Kafka的响应。

//org.apache.kafka.clients.producer.KafkaProducer#doSend
if (result.batchIsFull || result.newBatchCreated) {
                log.trace("Waking up the sender since topic {} partition {} is either full or getting a new batch", record.topic(), appendCallbacks.getPartition());
                this.sender.wakeup();
            }

Kafka的生产者缓存机制是Kafka面对海量消息时非常重要的优化机制。合理优化这些参数，对于Kafka集群性能提升是非常重要的。比如如果你的消息体比较大，那么应该考虑加大batch.size，尽量提升batch的缓存效率。而如果Producer要发送的消息确实非常多，那么就需要考虑加大total.memory参数，尽量避免缓存不够造成的阻塞。如果发现生产者发送消息比较慢，那么可以考虑提升max.in.flight.requests.per.connection参数，这样能加大消息发送的吞吐量。

发送应答机制

在Producer将消息发送到Broker后，要怎么确定消息是不是成功发到Broker上了呢？

及到的是在Producer端一个不太起眼的属性ACKS_CONFIG。

public static final String ACKS_CONFIG = "acks";
    private static final String ACKS_DOC = "The number of acknowledgments the producer requires the leader to have received before considering a request complete. This controls the "
                                           + " durability of records that are sent. The following settings are allowed: "
                                           + " <ul>"
                                           + " <li><code>acks=0</code> If set to zero then the producer will not wait for any acknowledgment from the"
                                           + " server at all. The record will be immediately added to the socket buffer and considered sent. No guarantee can be"
                                           + " made that the server has received the record in this case, and the <code>retries</code> configuration will not"
                                           + " take effect (as the client won't generally know of any failures). The offset given back for each record will"
                                           + " always be set to <code>-1</code>."
                                           + " <li><code>acks=1</code> This will mean the leader will write the record to its local log but will respond"
                                           + " without awaiting full acknowledgement from all followers. In this case should the leader fail immediately after"
                                           + " acknowledging the record but before the followers have replicated it then the record will be lost."
                                           + " <li><code>acks=all</code> This means the leader will wait for the full set of in-sync replicas to"
                                           + " acknowledge the record. This guarantees that the record will not be lost as long as at least one in-sync replica"
                                           + " remains alive. This is the strongest available guarantee. This is equivalent to the acks=-1 setting."
                                           + "</ul>"
                                           + "<p>"
                                           + "Note that enabling idempotence requires this config value to be 'all'."
                                           + " If conflicting configurations are set and idempotence is not explicitly enabled, idempotence is disabled.";

acks=0，生产者不关心Broker端有没有将消息写入到Partition，只发送消息就不管了。吞吐量是最高的，但是数据安全性是最低的。
acks=all or -1，生产者需要等Broker端的所有Partiton(Leader Partition以及其对应的Follower Partition都写完了才能得到返回结果，这样数据是最安全的，但是每次发消息需要等待更长的时间，吞吐量是最低的。
acks设置成1，则是一种相对中和的策略。Leader Partition在完成自己的消息写入后，就向生产者返回结果。

在生产环境中，acks=0可靠性太差，很少使用。acks=1，一般用于传输日志等，允许个别数据丢失的场景。使用范围最广。acks=-1，一般用于传输敏感数据，比如与钱相关的数据。如果ack设置为all或者-1，Kafka也并不是强制要求所有Partition都写入数据后才响应。在Kafka的Broker服务端会有一个配置参数min.insync.replicas，控制Leader Partition在完成多少个Partition的消息写入后，往Producer返回响应。这个参数可以在broker.conf文件中进行配置。

min.insync.replicas
When a producer sets acks to "all" (or "-1"), min.insync.replicas specifies the minimum number of replicas that must acknowledge a write for the write to be considered successful. If this minimum cannot be met, then the producer will raise an exception (either NotEnoughReplicas or NotEnoughReplicasAfterAppend).
When used together, min.insync.replicas and acks allow you to enforce greater durability guarantees. A typical scenario would be to create a topic with a replication factor of 3, set min.insync.replicas to 2, and produce with acks of "all". This will ensure that the producer raises an exception if a majority of replicas do not receive a write.

Type:	int
Default:	1
Valid Values:	[1,...]
Importance:	high
Update Mode:	cluster-wide

生产者消息幂等性

当Producer的acks设置成1或-1时，Producer每次发送消息都是需要获取Broker端返回的RecordMetadata的。这个过程中就需要两次跨网络请求。

如果要保证消息安全，那么对于每个消息，这两次网络请求就必须要求是幂等的。但是，网络是不靠谱的，在高并发场景下，往往没办法保证这两个请求是幂等的。Producer发送消息的过程中，如果第一步请求成功了，但是第二步却没有返回。这时，Producer就会认为消息发送失败了。那么Producer必然会发起重试。重试次数由参数ProducerConfig.RETRIES_CONFIG，默认值是Integer.MAX。Producer会重复发送多条消息到Broker中。Kafka如何保证无论Producer向Broker发送多少次重复的数据，Broker端都只保留一条消息，而不会重复保存多条消息呢？这就是Kafka消息生产者的幂等性问题。

    public static final String ENABLE_IDEMPOTENCE_CONFIG = "enable.idempotence";
    public static final String ENABLE_IDEMPOTENCE_DOC = "When set to 'true', the producer will ensure that exactly one copy of each message is written in the stream. If 'false', producer " 
        + "retries due to broker failures, etc., may write duplicates of the retried message in the stream. "
        + "Note that enabling idempotence requires <code>" + MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION + "</code> to be less than or equal to " + MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION_FOR_IDEMPOTENCE
        + " (with message ordering preserved for any allowable value), <code>" + RETRIES_CONFIG + "</code> to be greater than 0, and <code>"
        + ACKS_CONFIG + "</code> must be 'all'. "
        + "<p>"
        + "Idempotence is enabled by default if no conflicting configurations are set. "
        + "If conflicting configurations are set and idempotence is not explicitly enabled, idempotence is disabled. "
        + "If idempotence is explicitly enabled and conflicting configurations are set, a <code>ConfigException</code> is thrown.";

// max.in.flight.requests.per.connection should be less than or equal to 5 when idempotence producer enabled to ensure message ordering
    private static final int MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION_FOR_IDEMPOTENCE = 5;

    /** <code>max.in.flight.requests.per.connection</code> */
    public static final String MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION = "max.in.flight.requests.per.connection";
    private static final String MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION_DOC = "The maximum number of unacknowledged requests the client will send on a single connection before blocking."
                                                                            + " Note that if this config is set to be greater than 1 and <code>enable.idempotence</code> is set to false, there is a risk of"
                                                                            + " message re-ordering after a failed send due to retries (i.e., if retries are enabled)."
                                                                            + " Additionally, enabling idempotence requires this config value to be less than or equal to " + MAX_IN_FLIGHT_REQUESTS_PER_CONNECTION_FOR_IDEMPOTENCE + "."
                                                                            + " If conflicting configurations are set and idempotence is not explicitly enabled, idempotence is disabled.";

分布式数据传递过程中的三个数据语义：at-least-once:至少一次；at-most-once:最多一次；exactly-once:精确一次。

at-least-once可以保证数据不丢失，但是不能保证数据不重复。而at-most-once保证数据不重复，但是又不能保证数据不丢失。这两种语义虽然都有缺陷，但是实现起来相对来说比较简单。但是对一些敏感的业务数据，往往要求数据即不重复也不丢失，这就需要支持Exactly-once语义。而要支持Exactly-once语义，需要有非常精密的设计。

Kafka为了保证消息发送的Exactly-once语义，增加了几个概念：

1. PID：每个新的Producer在初始化的过程中就会被分配一个唯一的PID。这个PID对用户是不可见的。
2. Sequence Numer: 对于每个PID，这个Producer针对Partition会维护一个sequenceNumber。这是一个从0开始单调递增的数字。当Producer要往同一个Partition发送消息时，这个Sequence Number就会加1。然后会随着消息一起发往Broker。
3. Broker端则会针对每个<PID,Partition>维护一个序列号（SN），只有当对应的SequenceNumber = SN+1时，Broker才会接收消息，同时将SN更新为SN+1。否则，SequenceNumber过小就认为消息已经写入了，不需要再重复写入。而如果SequenceNumber过大，就会认为中间可能有数据丢失了。对生产者就会抛出一个OutOfOrderSequenceException。

这样，Kafka在打开idempotence幂等性控制后，在Broker端就会保证每条消息在一次发送过程中，Broker端最多只会刚刚好持久化一条。这样就能保证at-most-once语义。再加上之前分析的将生产者的acks参数设置成1或-1，保证at-least-once语义，这样就整体上保证了Exactaly-once语义。

给Producer打开幂等性后，不管Producer往同一个Partition发送多少条消息，都可以通过幂等机制保证消息的Exactly-only语义。

生产者消息事务

通过生产者消息幂等性问题，能够解决单生产者消息写入单分区的的幂等性问题。但是，如果是要写入多个分区呢？比如发送五条消息，他们的key都是不同的。这批消息就有可能写入多个Partition，而这些Partition是分布在不同Broker上的。这意味着，Producer需要对多个Broker同时保证消息的幂等性。

这时候，通过上面的生产者消息幂等性机制就无法保证所有消息的幂等了。这时候就需要有一个事务机制，保证这一批消息最好同时成功的保持幂等性。或者这一批消息同时失败，这样生产者就可以开始进行整体重试，消息不至于重复。

Kafka引入了消息事务机制。涉及到Producer中的几个API：

// 1 初始化事务
void initTransactions();
// 2 开启事务
void beginTransaction() throws ProducerFencedException;
// 3 提交事务
void commitTransaction() throws ProducerFencedException;
// 4 放弃事务（类似于回滚事务的操作）
void abortTransaction() throws ProducerFencedException;

public class TransactionErrorDemo {

    private static final String BOOTSTRAP_SERVERS = "127.0.0.1:9092";
    private static final String TOPIC = "disTopic";

    public static void main(String[] args) throws ExecutionException, InterruptedException {
        Properties props = new Properties();
        // 此处配置的是kafka的端口
        props.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, BOOTSTRAP_SERVERS);
        // 事务ID
        props.put(ProducerConfig.TRANSACTIONAL_ID_CONFIG,"111");
        // 配置key的序列化类
        props.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringSerializer");
        // 配置value的序列化类
        props.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringSerializer");

        Producer<String,String> producer = new KafkaProducer<>(props);
        producer.initTransactions();
        producer.beginTransaction();
        for(int i = 0; i < 5; i++) {
            ProducerRecord<String, String> record = new ProducerRecord<>(TOPIC, Integer.toString(i), "MyProducer" + i);
            //异步发送。
            producer.send(record);
            if(i == 3){
                //第三条消息放弃事务之后，整个这一批消息都回退了。
                System.out.println("error");
                producer.abortTransaction();
            }
        }
        System.out.println("message sended");
        try {
            Thread.sleep(10000);
        } catch (Exception e) {
            e.printStackTrace();
        }
//        producer.commitTransaction();
        producer.close();
    }
}

可以先启动一个订阅了disTopic这个Topic的消费者，然后启动这个生产者，进行试验。在这个试验中，发送到第3条消息时，主动放弃事务，此时之前的消息也会一起回滚。

实际上，Kafka的事务消息还会做两件事情：

1. 一个TransactionId只会对应一个PID

如果当前一个Producer的事务没有提交，而另一个新的Producer保持相同的TransactionId，这时旧的生产者会立即失效，无法继续发送消息。

2. 跨会话事务对齐

如果某个Producer实例异常宕机了，事务没有被正常提交。那么新的TransactionId相同的Producer实例会对旧的事务进行补齐。保证旧事务要么提交，要么终止。这样新的Producer实例就可以以一个正常的状态开始工作。

如果一个Producer需要发送多条消息，通常比较安全的发送方式是这样的：

public class TransactionProducer {
    private static final String BOOTSTRAP_SERVERS = "127.0.0.1:9092";
    private static final String TOPIC = "disTopic";
    public static void main(String[] args) throws ExecutionException, InterruptedException {
        Properties props = new Properties();
        // 此处配置的是kafka的端口
        props.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, BOOTSTRAP_SERVERS);
        // 事务ID。
        props.put(ProducerConfig.TRANSACTIONAL_ID_CONFIG,"111");
        // 配置key的序列化类
        props.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringSerializer");
        // 配置value的序列化类
        props.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringSerializer");

        Producer<String,String> producer = new KafkaProducer<>(props);
        producer.initTransactions();
        producer.beginTransaction();
        try{
            for(int i = 0; i < 5; i++) {
                ProducerRecord<String, String> record = new ProducerRecord<>(TOPIC, Integer.toString(i), "MyProducer" + i);
                //异步发送。
                producer.send(record);
            }
            producer.commitTransaction();
        }catch (ProducerFencedException e){
            producer.abortTransaction();
        }finally {
            producer.close();
        }
    }
}

其中对于事务ID这个参数，可以任意起名，但是建议包含一定的业务唯一性。

生产者的事务消息机制保证了Producer发送消息的安全性，但是，他并不保证已经提交的消息就一定能被所有消费者消费。