alpakka-kafka(1)-producer

雪川大蟲發表於2021-02-20
Kafka

  alpakka專案是一個基於akka-streams流處理程式設計工具的scala/java開源專案，通過提供connector連線各種資料來源並在akka-streams裡進行資料處理。alpakka-kafka就是alpakka專案裡的kafka-connector。對於我們來說：可以用alpakka-kafka來對接kafka，使用kafka提供的功能。或者從另外一個角度講：alpakka-kafka就是一個用akka-streams實現kafka功能的scala開發工具。

alpakka-kafka提供了kafka的核心功能：producer、consumer，分別負責把akka-streams裡的資料寫入kafka及從kafka中讀出資料並輸入到akka-streams裡。用akka-streams整合kafka的應用場景通常出現在業務整合方面：在一項業務A中產生一些業務操作指令寫入kafka，然後通過kafka把指令傳送給另一項業務B，業務B從kafka中獲取操作指令並進行相應的業務操作。如：有兩個業務模組：收貨管理和庫存管理，一方面收貨管理向kafka寫入收貨記錄。另一頭庫存管理從kafka中讀取收貨記錄並更新相關庫存數量記錄。注意，這兩項業務是分別操作的。在alpakka中，實際的業務操作基本就是在akka-streams裡的資料處理（transform），其實是典型的CQRS模式：讀寫兩方互不關聯，寫時不管受眾是誰，如何使用、讀者不關心誰是寫方。這裡的寫和讀兩方分別代表kafka裡的producer和consumer。

本篇我們先介紹alpakka-kafka的producer功能及其使用方法。如前所述：alpakka是用akka-streams實現了kafka-producer功能。alpakka提供的producer也就是akka-streams的一種元件，可以與其它的akka-streams元件組合形成更大的akka-streams個體。構建一個producer需要先完成幾個配件類構成：

1、producer-settings配置：alpakka-kafka在reference.conf裡的akka.kafka.producer配置段落提供了足夠支援基本運作的預設producer配置。使用者可以通過typesafe config配置檔案操作工具來靈活調整配置

2、de/serializer序列化工具：alpakka-kafka提供了String型別的序列化/反序列化函式，可以直接使用

4、bootstrap-server：一個以逗號分隔的kafka-cluster節點ip清單文字

下面是一個具體的例子：

  implicit val system = ActorSystem("kafka_sys")
  val bootstrapServers = "localhost:9092"
  val config = system.settings.config.getConfig("akka.kafka.producer")
  val producerSettings =
    ProducerSettings(config, new StringSerializer, new StringSerializer)
      .withBootstrapServers(bootstrapServers)

這裡使用ActorSystem只是為了讀取.conf檔案裡的配置，還沒有使用任何akka-streams元件。akka.kafka.producer配置段落在alpakka-kafka的reference.conf裡提供了預設配置，不需要在application.conf裡重新定義。

alpakka-kafka提供了一個最基本的producer，非akka-streams元件，sendProducer。下面我們示範一下sendProducer的使用和效果：

 

import akka.actor.ActorSystem
import akka.kafka.scaladsl.{Consumer, SendProducer}
import org.apache.kafka.clients.producer.{ProducerRecord, RecordMetadata}
import akka.kafka._
import org.apache.kafka.common.serialization._
import scala.concurrent.duration._
import scala.concurrent.{Await, Future}

object SendProducerDemo extends App {
  implicit val system = ActorSystem("kafka_sys")
  implicit val executionContext = system.dispatcher
  val bootstrapServers = "localhost:9092"
  val config = system.settings.config.getConfig("akka.kafka.producer")
  val producerSettings =
    ProducerSettings(config, new StringSerializer, new StringSerializer)
      .withBootstrapServers(bootstrapServers)
  val producer = SendProducer(producerSettings)
  val topic = "greatings"
  val lstfut: Seq[Future[RecordMetadata]] =
    (100 to 200).reverse
      .map(_.toString)
      .map(value => new ProducerRecord[String, String](topic, s"hello-$value"))
      .map(msg => producer.send(msg))

  val futlst = Future.sequence(lstfut)
  Await.result(futlst, 2.seconds)


  scala.io.StdIn.readLine()
  producer.close()
  system.terminate()
}

 

以上示範用sendProducer向kafka寫入100條hello訊息。使用的是集合遍歷，沒有使用akka-streams的Source。為了檢驗具體效果，我們可以使用kafka提供的一些手工指令，如下：

 

 

\w> ./kafka-topics --create --topic greatings --bootstrap-server localhost:9092
Created topic greatings.
\w> ./kafka-console-consumer --topic greatings  --bootstrap-server localhost:9092
hello-100
hello-101
hello-102
hello-103
hello-104
hello-105
hello-106
...

 

既然producer代表寫入功能，那麼在akka-streams裡就是Sink或Flow元件的功能了。下面這個例子是producer Sink元件plainSink的示範：

import akka.Done
import akka.actor.ActorSystem
import akka.kafka.scaladsl._
import akka.kafka._
import akka.stream.scaladsl._
import org.apache.kafka.clients.producer.ProducerRecord
import org.apache.kafka.common.serialization._

import scala.concurrent._
import scala.concurrent.duration._

object plain_sink extends App {
  implicit val system = ActorSystem("kafka_sys")
  val bootstrapServers = "localhost:9092"
  val config = system.settings.config.getConfig("akka.kafka.producer")
  val producerSettings =
    ProducerSettings(config, new StringSerializer, new StringSerializer)
      .withBootstrapServers(bootstrapServers)

  implicit val executionContext = system.dispatcher
  val topic = "greatings"
  val done: Future[Done] =
    Source(1 to 100)
      .map(_.toString)
      .map(value => new ProducerRecord[String, String](topic, s"hello-$value"))
      .runWith(Producer.plainSink(producerSettings))

  Await.ready(done,3.seconds)

  scala.io.StdIn.readLine()
  system.terminate()
}

這是一個典型的akka-streams應用例項，其中Producer.plainSink就是一個akka-streams Sink元件。

以上兩個示範都涉及到構建一個ProducerRecord型別並將之寫入kafka。ProducerRecord是一個基本的kafka訊息型別：

   public ProducerRecord(String topic, K key, V value) {
        this(topic, null, null, key, value, null);
    }

topic是String型別，key, value 是 Any 型別的。 alpakka-kafka在ProducerRecord之上又擴充了一個複雜點的訊息型別ProducerMessage.Envelope型別：

sealed trait Envelope[K, V, +PassThrough] {
    def passThrough: PassThrough
    def withPassThrough[PassThrough2](value: PassThrough2): Envelope[K, V, PassThrough2]
  }


  final case class Message[K, V, +PassThrough](
      record: ProducerRecord[K, V],
      passThrough: PassThrough
  ) extends Envelope[K, V, PassThrough] {
    override def withPassThrough[PassThrough2](value: PassThrough2): Message[K, V, PassThrough2] =
      copy(passThrough = value)
  }

ProducerMessage.Envelope增加了個PassThrough引數，用來與訊息一道傳遞額外的後設資料。alpakka-kafka streams元件使用這個訊息型別作為流元素，最終把它轉換成一或多條ProducerRecord寫入kafka。如下： 

object EventMessages {
//一對一條ProducerRecord
   def createMessage[KeyType,ValueType,PassThroughType](
      topic: String,
      key: KeyType,
      value: ValueType,
      passThrough: PassThroughType): ProducerMessage.Envelope[KeyType,ValueType,PassThroughType] = {
     val single = ProducerMessage.single(
       new ProducerRecord[KeyType,ValueType](topic,key,value),
       passThrough
     )
     single
   }
//一對多條ProducerRecord
  def createMultiMessage[KeyType,ValueType,PassThroughType] (
       topics: List[String],
       key: KeyType,
       value: ValueType,
       passThrough: PassThroughType): ProducerMessage.Envelope[KeyType,ValueType,PassThroughType] = {
    import scala.collection.immutable
    val msgs = topics.map { topic =>
      new ProducerRecord(topic,key,value)
    }.toSeq
    val multi = ProducerMessage.multi(
      msgs,
      passThrough
    )
    multi
  }
//只傳遞通過型後設資料
  def createPassThroughMessage[KeyType,ValueType,PassThroughType](
       topic: String,
       key: KeyType,
       value: ValueType,
       passThrough: PassThroughType): ProducerMessage.Envelope[KeyType,ValueType,PassThroughType] = {
    ProducerMessage.passThrough(passThrough)
  }

}

flexiFlow是一個alpakka-kafka Flow元件，流入ProducerMessage.Evelope，流出Results型別： 

  def flexiFlow[K, V, PassThrough](
      settings: ProducerSettings[K, V]
  ): Flow[Envelope[K, V, PassThrough], Results[K, V, PassThrough], NotUsed] = { ... }

Results型別定義如下：

  final case class Result[K, V, PassThrough] private (
      metadata: RecordMetadata,
      message: Message[K, V, PassThrough]
  ) extends Results[K, V, PassThrough] {
    def offset: Long = metadata.offset()
    def passThrough: PassThrough = message.passThrough
  }

也就是說flexiFlow可以返回寫入kafka後kafka返回的操作狀態資料。我們再看看flexiFlow的使用案例： 

import akka.kafka.ProducerMessage._
import akka.actor.ActorSystem
import akka.kafka.scaladsl._
import akka.kafka.{ProducerMessage, ProducerSettings}
import akka.stream.scaladsl.{Sink, Source}
import org.apache.kafka.clients.producer.ProducerRecord
import org.apache.kafka.common.serialization.StringSerializer

import scala.concurrent._
import scala.concurrent.duration._

object flexi_flow extends App {
  implicit val system = ActorSystem("kafka_sys")
  val bootstrapServers = "localhost:9092"
  val config = system.settings.config.getConfig("akka.kafka.producer")
  val producerSettings =
    ProducerSettings(config, new StringSerializer, new StringSerializer)
      .withBootstrapServers(bootstrapServers)

  // needed for the future flatMap/onComplete in the end
  implicit val executionContext = system.dispatcher
  val topic = "greatings"

  val done = Source(1 to 100)
    .map { number =>
      val value = number.toString
      EventMessages.createMessage(topic,"key",value,number)
    }
    .via(Producer.flexiFlow(producerSettings))
    .map {
      case ProducerMessage.Result(metadata, ProducerMessage.Message(record, passThrough)) =>
        s"${metadata.topic}/${metadata.partition} ${metadata.offset}: ${record.value}"

      case ProducerMessage.MultiResult(parts, passThrough) =>
        parts
          .map {
            case MultiResultPart(metadata, record) =>
              s"${metadata.topic}/${metadata.partition} ${metadata.offset}: ${record.value}"
          }
          .mkString(", ")

      case ProducerMessage.PassThroughResult(passThrough) =>
        s"passed through"
    }
    .runWith(Sink.foreach(println(_)))

  Await.ready(done,3.seconds)

  scala.io.StdIn.readLine()
  system.terminate()
}

object EventMessages {
   def createMessage[KeyType,ValueType,PassThroughType](
      topic: String,
      key: KeyType,
      value: ValueType,
      passThrough: PassThroughType): ProducerMessage.Envelope[KeyType,ValueType,PassThroughType] = {
     val single = ProducerMessage.single(
       new ProducerRecord[KeyType,ValueType](topic,key,value),
       passThrough
     )
     single
   }
  def createMultiMessage[KeyType,ValueType,PassThroughType] (
       topics: List[String],
       key: KeyType,
       value: ValueType,
       passThrough: PassThroughType): ProducerMessage.Envelope[KeyType,ValueType,PassThroughType] = {
    import scala.collection.immutable
    val msgs = topics.map { topic =>
      new ProducerRecord(topic,key,value)
    }.toSeq
    val multi = ProducerMessage.multi(
      msgs,
      passThrough
    )
    multi
  }
  def createPassThroughMessage[KeyType,ValueType,PassThroughType](
       topic: String,
       key: KeyType,
       value: ValueType,
       passThrough: PassThroughType): ProducerMessage.Envelope[KeyType,ValueType,PassThroughType] = {
    ProducerMessage.passThrough(passThrough)
  }

}

producer除向kafka寫入與業務相關的業務事件或業務指令外還會向kafka寫入當前訊息讀取的具體位置offset，所以alpakka-kafka的produce可分成兩種型別：上面示範的plainSink, flexiFlow只向kafka寫業務資料。還有一類如commitableSink還包括了把訊息讀取位置offset寫入commit的功能。如下：

val control =
  Consumer
    .committableSource(consumerSettings, Subscriptions.topics(topic1, topic2))
    .map { msg =>
      ProducerMessage.single(
        new ProducerRecord(targetTopic, msg.record.key, msg.record.value),
        msg.committableOffset
      )
    }
    .toMat(Producer.committableSink(producerSettings, committerSettings))(DrainingControl.apply)
    .run()

control.drainAndShutdown()

如上所示，committableSource從kafka讀取業務訊息及讀取位置committableOffsset，然後Producer.committableSink把業務訊息和offset再寫入kafka。

下篇討論我們再具體介紹consumer。

 

相關文章