Search code examples
scalaakkaakka-typed

Will the stream be terminated?

I have actors, that looks like as follows:

enter image description here

As you can see on the image, the ActorStream is a child of Actor. The question is, when I terminate the Actor, will the ActorStream also be terminated?

Here is the way, how I create the ActorStream in an Actor: 

 def create(fsm: ActorRef[ServerHealth], cancel: Option[Cancellable]): Behavior[ServerHealthStreamer] =
    Behaviors.setup { context =>
      implicit val system = context.system
      implicit val materializer = ActorMaterializer()
      implicit val dispatcher = materializer.executionContext

      val kafkaServer = system
        .settings
        .config
        .getConfig("kafka")
        .getString("servers")

      val sink: Sink[ServerHealth, NotUsed] = ActorSink.actorRefWithAck[ServerHealth, ServerHealthStreamer, Ack](
        ref = context.self,
        onCompleteMessage = Complete,
        onFailureMessage = Fail.apply,
        messageAdapter = Message.apply,
        onInitMessage = Init.apply,
        ackMessage = Ack)

      val cancel = Source.tick(1.seconds, 15.seconds, NotUsed)
        .flatMapConcat(_ => Source.fromFuture(health(kafkaServer)))
        .map {
          case true =>
            KafkaActive
          case false =>
            KafkaInactive
        }
        .to(sink)
        .run()

      Behaviors.receiveMessage {
        case Init(ackTo) =>
          ackTo ! Ack
          Behaviors.same
        case Message(ackTo, msg) =>
          fsm ! msg
          ackTo ! Ack
          create(fsm, Some(cancel))
        case Complete =>
          Behaviors.same
        case Fail(_) =>
          fsm ! KafkaInactive
          Behaviors.same
      }
    }
Solution

  • In your case actor termination must terminate stream because under the hood stage actor watching passed actorRef and complete stage if Terminated arrived

    I think you can find more information here https://blog.colinbreck.com/integrating-akka-streams-and-akka-actors-part-ii/

    An extremely important aspect to understand is that the materialized stream is running as a set of actors on the threads of the execution context on which they were allocated. In other words, the stream is running independently from the actor that allocated it. This becomes very important if the stream is long-running, or even infinite, and we want the actor to manage the life-cycle of the stream, such that when the actor stops, the stream is terminated. Expanding on the example above, I will make the stream infinite and use a KillSwitch to manage the life-cycle of the stream.