如何使用Akka持久性保存流数据

Question

我使用StreamRefs在集群中的参与者之间建立流式连接。目前，在writing节点中，我手动将传入消息保存到日志文件中，但我想知道是否可以将其替换为persistentSink，用于写入，以及persistentSource，用于从Akka Persistence journal启动actor时读取。我一直在考虑用Persistent actor的Persistent{evt替换日志文件接收器=

当前执行情况：

object Writer {
  case class WriteSinkRequest(userId: String) 
  case class WriteSinkReady(userId: String, sinkRef: SinkRef[ByteString])
  case class ReadSourceRequest(userId: String)
  case class ReadSourceReady(userId: String, sourceRef: SourceRef[ByteString])
}

class Writer extends Actor {

    // code omitted

    val logsDir = "logs"

    val path = Files.createDirectories(FileSystems.getDefault.getPath(logsDir))

    def logFile(id: String) = {
        path.resolve(id)
    }

    def logFileSink(logId: String): Sink[ByteString, Future[IOResult]] = FileIO.toPath(logFile(logId), Set(CREATE, WRITE, APPEND))
    def logFileSource(logId: String): Source[ByteString, Future[IOResult]] = FileIO.fromPath(logFile(logId))

    override def receive: Receive = {
        case WriteSinkRequest(userId) => 
            // obtain the source you want to offer:
            val sink = logFileSink(userId)
            // materialize the SinkRef (the remote is like a source of data for us):
            val ref: Future[SinkRef[ByteString]] = StreamRefs.sinkRef[ByteString]().to(sink).run()
            // wrap the SinkRef in some domain message, such that the sender knows what source it is
            val reply: Future[WriteSinkReady] = ref.map(WriteSinkReady(userId, _))
            // reply to sender
            reply.pipeTo(sender())

        case ReadSourceRequest(userId) =>
            val source = logFileSource(userId)
            val ref: Future[SourceRef[ByteString]] = source.runWith(StreamRefs.sourceRef())
            val reply: Future[ReadSourceReady] = ref.map(ReadSourceReady(userId, _))
            reply pipeTo sender()

    }
}

另外，是否可以不创建“保存到日志”接收器，而是创建流：要写入的传入数据~




Answer 1

以背压方式将数据流传输给持久参与者的一个想法是使用Sink。actorRefWithAck：让参与者在保留消息时发送确认消息。这将类似于以下内容：

// ...
case class WriteSinkReady(userId: String, sinkRef: SinkRef[MyMsg])    
// ...

def receive = {
  case WriteSinkRequest(userId) =>
    val persistentActor: ActorRef = ??? // a persistent actor that handles MyMsg messages
                                        // as well as the messages used in persistentSink

    val persistentSink: Sink[MyMsg, NotUsed] = Sink.actorRefWithAck[MyMsg](
      persistentActor,
      /* additional parameters: see the docs */
    )

    val ref: Future[SinkRef[MyMsg]] = StreamRefs.sinkRef[MyMsg]().to(persistentSink).run()
    val reply: Future[WriteSinkReady] = ref.map(WriteSinkReady(userId, _))
    reply.pipeTo(sender())

  case ReadSourceRequest(userId) =>
    // ...
}

上面的示例使用名为MyMsg的自定义case类，而不是ByteString。

在发送者中，假设是演员：

def receive = {
  case WriteSinkReady(userId, sinkRef) =>
    source.runWith(sinkRef) // source is a Source[MyMsg, _]

  // ...
}

发送方中的物化流将向持久参与者发送消息。