Question

我有一个仅由一个经纪人组成的本地kafka集群。我有两个Akka流：

一个流会定期生成一条包含当前时间戳的消息，并将其保存到kafka主题
另一个流已订阅此主题并计算延迟。

似乎仅在流中的下一条消息生成时才生成消息。这是每两秒发布一条消息（以毫秒为单位）时观察到的延迟：

total: 2019. Before persist: 2001, after: 18
total: 2019. Before persist: 2002, after: 17
total: 2019. Before persist: 2002, after: 17
total: 2019. Before persist: 2002, after: 17
total: 2019. Before persist: 2002, after: 17
total: 2019. Before persist: 2002, after: 17
total: 2020. Before persist: 2002, after: 18
total: 2019. Before persist: 2001, after: 18
total: 2019. Before persist: 2002, after: 17
total: 2019. Before persist: 2002, after: 17
total: 2019. Before persist: 2002, after: 17

如果我将周期增加到5秒，则延迟会相应增加（+ -5000 ms）。

以下是产生的akka流代码：


  val producer =
    Source.fromIterator(() => Iterator.continually())
      .throttle(1, 5.second)
      .map { value =>
        val currentTime = System.currentTimeMillis()
        new ProducerRecord[ByteBuffer, ByteBuffer]("kafka-streams-input", ByteBuffer.wrap(currentTime.toString.getBytes))
      }
      .runWith(Producer.plainSink(KafkaClient.producerSettings))

Akka流kafka仅在下一条记录可用时才产生记录

0 个答案: