Dispatchers

You are viewing the documentation for the new actor APIs, to view the Pekko Classic documentation, see Classic Dispatchers.

Dependency

Dispatchers are part of core Pekko, which means that they are part of the pekko-actor dependency. This page describes how to use dispatchers with pekko-actor-typed, which has dependency:

sbt

val PekkoVersion = "1.0.3"
libraryDependencies += "org.apache.pekko" %% "pekko-actor-typed" % PekkoVersion

Maven

<properties>
  <scala.binary.version>2.13</scala.binary.version>
</properties>
<dependencyManagement>
  <dependencies>
    <dependency>
      <groupId>org.apache.pekko</groupId>
      <artifactId>pekko-bom_${scala.binary.version}</artifactId>
      <version>1.0.3</version>
      <type>pom</type>
      <scope>import</scope>
    </dependency>
  </dependencies>
</dependencyManagement>
<dependencies>
  <dependency>
    <groupId>org.apache.pekko</groupId>
    <artifactId>pekko-actor-typed_${scala.binary.version}</artifactId>
  </dependency>
</dependencies>

Gradle

def versions = [
  ScalaBinary: "2.13"
]
dependencies {
  implementation platform("org.apache.pekko:pekko-bom_${versions.ScalaBinary}:1.0.3")

  implementation "org.apache.pekko:pekko-actor-typed_${versions.ScalaBinary}"
}

Introduction

A Pekko MessageDispatcher is what makes Pekko Actors “tick”, it is the engine of the machine so to speak. All MessageDispatcher implementations are also an ExecutionContextExecutor, which means that they can be used to execute arbitrary code, for instance FuturesCompletableFutures.

Default dispatcher

Every ActorSystem will have a default dispatcher that will be used in case nothing else is configured for an Actor. The default dispatcher can be configured, and is by default a Dispatcher with the configured pekko.actor.default-dispatcher.executor. If no executor is selected a “fork-join-executor” is selected, which gives excellent performance in most cases.

Internal dispatcher

To protect the internal Actors that are spawned by the various Pekko modules, a separate internal dispatcher is used by default. The internal dispatcher can be tuned in a fine-grained way with the setting pekko.actor.internal-dispatcher, it can also be replaced by another dispatcher by making pekko.actor.internal-dispatcher an alias.

Looking up a Dispatcher

Dispatchers implement the ExecutionContextExecutor interface and can thus be used to run FutureCompletableFuture invocations etc.

Scala

copysource// for use with Futures, Scheduler, etc.
import org.apache.pekko.actor.typed.DispatcherSelector
implicit val executionContext = context.system.dispatchers.lookup(DispatcherSelector.fromConfig("my-dispatcher"))

Java

copysource// this is scala.concurrent.ExecutionContextExecutor, which implements
// both scala.concurrent.ExecutionContext (for use with Futures, Scheduler, etc.)
// and java.util.concurrent.Executor (for use with CompletableFuture etc.)
final ExecutionContextExecutor ex =
    system.dispatchers().lookup(DispatcherSelector.fromConfig("my-dispatcher"));

Selecting a dispatcher

A default dispatcher is used for all actors that are spawned without specifying a custom dispatcher. This is suitable for all actors that don’t block. Blocking in actors needs to be carefully managed, more details here.

To select a dispatcher use DispatcherSelector to create a Props instance for spawning your actor:

Scala

copysourceimport org.apache.pekko.actor.typed.DispatcherSelector

context.spawn(yourBehavior, "DefaultDispatcher")
context.spawn(yourBehavior, "ExplicitDefaultDispatcher", DispatcherSelector.default())
context.spawn(yourBehavior, "BlockingDispatcher", DispatcherSelector.blocking())
context.spawn(yourBehavior, "ParentDispatcher", DispatcherSelector.sameAsParent())
context.spawn(yourBehavior, "DispatcherFromConfig", DispatcherSelector.fromConfig("your-dispatcher"))

Java

copysourcecontext.spawn(behavior, "DefaultDispatcher");
context.spawn(behavior, "ExplicitDefaultDispatcher", DispatcherSelector.defaultDispatcher());
context.spawn(behavior, "BlockingDispatcher", DispatcherSelector.blocking());
context.spawn(behavior, "ParentDispatcher", DispatcherSelector.sameAsParent());
context.spawn(
    behavior, "DispatcherFromConfig", DispatcherSelector.fromConfig("your-dispatcher"));

DispatcherSelector has a few convenience methods:

• DispatcherSelector.defaultDispatcherSelector.defaultDispatcher to look up the default dispatcher
• DispatcherSelector.blocking can be used to execute actors that block e.g. a legacy database API that does not support FutureCompletionStages
• DispatcherSelector.sameAsParent to use the same dispatcher as the parent actor

The final example shows how to load a custom dispatcher from configuration and relies on this being in your application.conf:

copysourceyour-dispatcher {
  type = Dispatcher
  executor = "thread-pool-executor"
  thread-pool-executor {
    fixed-pool-size = 32
  }
  throughput = 1
}

Types of dispatchers

There are 2 different types of message dispatchers:

• Dispatcher

  This is an event-based dispatcher that binds a set of Actors to a thread pool. The default dispatcher is used if no other is specified.

  • Shareability: Unlimited
  • Mailboxes: Any, creates one per Actor
  • Use cases: Default dispatcher, Bulkheading
  • Driven by: java.util.concurrent.ExecutorService. Specify using “executor” using “fork-join-executor”, “thread-pool-executor” or the fully-qualified class name of an org.apache.pekko.dispatcher.ExecutorServiceConfigurator implementation.

• PinnedDispatcher

  This dispatcher dedicates a unique thread for each actor using it; i.e. each actor will have its own thread pool with only one thread in the pool.

  • Shareability: None
  • Mailboxes: Any, creates one per Actor
  • Use cases: Bulkheading
  • Driven by: Any org.apache.pekko.dispatch.ThreadPoolExecutorConfigurator. By default a “thread-pool-executor”.

Here is an example configuration of a Fork Join Pool dispatcher:

copysourcemy-dispatcher {
  # Dispatcher is the name of the event-based dispatcher
  type = Dispatcher
  # What kind of ExecutionService to use
  executor = "fork-join-executor"
  # Configuration for the fork join pool
  fork-join-executor {
    # Min number of threads to cap factor-based parallelism number to
    parallelism-min = 2
    # Parallelism (threads) ... ceil(available processors * factor)
    parallelism-factor = 2.0
    # Max number of threads to cap factor-based parallelism number to
    parallelism-max = 10
  }
  # Throughput defines the maximum number of messages to be
  # processed per actor before the thread jumps to the next actor.
  # Set to 1 for as fair as possible.
  throughput = 100
}

For more configuration options, see the More dispatcher configuration examples section and the default-dispatcher section of the configuration.

Dispatcher aliases

When a dispatcher is looked up, and the given setting contains a string rather than a dispatcher config block, the lookup will treat it as an alias, and follow that string to an alternate location for a dispatcher config. If the dispatcher config is referenced both through an alias and through the absolute path only one dispatcher will be used and shared among the two ids.

Example: configuring internal-dispatcher to be an alias for default-dispatcher:

pekko.actor.internal-dispatcher = pekko.actor.default-dispatcher

Blocking Needs Careful Management

In some cases it is unavoidable to do blocking operations, i.e. to put a thread to sleep for an indeterminate time, waiting for an external event to occur. Examples are legacy RDBMS drivers or messaging APIs, and the underlying reason is typically that (network) I/O occurs under the covers.

The Managing Blocking in Akka video explains why it is bad to block inside an actor, and how you can use custom dispatchers to manage blocking when you cannot avoid it. The same principle applies with Pekko actors.

Problem: Blocking on default dispatcher

Simply adding blocking calls to your actor message processing like this is problematic:

Scala

copysourceimport org.apache.pekko
import pekko.actor.typed.Behavior
import pekko.actor.typed.scaladsl.Behaviors

object BlockingActor {
  def apply(): Behavior[Int] =
    Behaviors.receiveMessage { i =>
      // DO NOT DO THIS HERE: this is an example of incorrect code,
      // better alternatives are described further on.

      // block for 5 seconds, representing blocking I/O, etc
      Thread.sleep(5000)
      println(s"Blocking operation finished: $i")
      Behaviors.same
    }
}

Java

copysourceimport org.apache.pekko.actor.typed.*;
import org.apache.pekko.actor.typed.javadsl.*;

public class BlockingActor extends AbstractBehavior<Integer> {
  public static Behavior<Integer> create() {
    return Behaviors.setup(BlockingActor::new);
  }

  private BlockingActor(ActorContext<Integer> context) {
    super(context);
  }

  @Override
  public Receive<Integer> createReceive() {
    return newReceiveBuilder()
        .onMessage(
            Integer.class,
            i -> {
              // DO NOT DO THIS HERE: this is an example of incorrect code,
              // better alternatives are described further on.

              // block for 5 seconds, representing blocking I/O, etc
              Thread.sleep(5000);
              System.out.println("Blocking operation finished: " + i);
              return Behaviors.same();
            })
        .build();
  }
}

Without any further configuration the default dispatcher runs this actor along with all other actors. This is very efficient when all actor message processing is non-blocking. When all of the available threads are blocked, however, then all the actors on the same dispatcher will starve for threads and will not be able to process incoming messages.

To demonstrate this problem, let’s set up an application with the above BlockingActor and the following PrintActor:

Scala

copysourceobject PrintActor {
  def apply(): Behavior[Integer] =
    Behaviors.receiveMessage { i =>
      println(s"PrintActor: $i")
      Behaviors.same
    }
}

Java

copysourceclass PrintActor extends AbstractBehavior<Integer> {

  public static Behavior<Integer> create() {
    return Behaviors.setup(PrintActor::new);
  }

  private PrintActor(ActorContext<Integer> context) {
    super(context);
  }

  @Override
  public Receive<Integer> createReceive() {
    return newReceiveBuilder()
        .onMessage(
            Integer.class,
            i -> {
              System.out.println("PrintActor: " + i);
              return Behaviors.same();
            })
        .build();
  }
}

Scala

copysourceval root = Behaviors.setup[Nothing] { context =>
  for (i <- 1 to 100) {
    context.spawn(BlockingFutureActor(), s"futureActor-$i") ! i
    context.spawn(PrintActor(), s"printActor-$i") ! i
  }
  Behaviors.empty
}
val system = ActorSystem[Nothing](root, "BlockingDispatcherSample")

Java

copysourceBehavior<Void> root =
    Behaviors.setup(
        context -> {
          for (int i = 0; i < 100; i++) {
            context.spawn(BlockingActor.create(), "BlockingActor-" + i).tell(i);
            context.spawn(PrintActor.create(), "PrintActor-" + i).tell(i);
          }
          return Behaviors.ignore();
        });

Here the app is sending 100 messages to BlockingActors and PrintActors and large numbers of pekko.actor.default-dispatcher threads are handling requests. When you run the above code, you will likely to see the entire application gets stuck somewhere like this:

>　PrintActor: 44
>　PrintActor: 45

PrintActor is considered non-blocking, however it is not able to proceed with handling the remaining messages, since all the threads are occupied and blocked by the other blocking actors - thus leading to thread starvation.

In the thread state diagrams below the colors have the following meaning:

• Turquoise - Sleeping state
• Orange - Waiting state
• Green - Runnable state

The thread information was recorded using the YourKit profiler, however any good JVM profiler has this feature (including the free and bundled with the Oracle JDK VisualVM, as well as Java Mission Control).

The orange portion of the thread shows that it is idle. Idle threads are fine - they’re ready to accept new work. However, a large number of turquoise (blocked, or sleeping as in our example) threads leads to thread starvation.

In the above example we put the code under load by sending hundreds of messages to blocking actors which causes threads of the default dispatcher to be blocked. The fork join pool based dispatcher in Pekko then attempts to compensate for this blocking by adding more threads to the pool (default-pekko.actor.default-dispatcher 18,19,20,...). This however is not able to help if those too will immediately get blocked, and eventually the blocking operations will dominate the entire dispatcher.

In essence, the Thread.sleep operation has dominated all threads and caused anything executing on the default dispatcher to starve for resources (including any actor that you have not configured an explicit dispatcher for).

implicit val executionContext: ExecutionContext = context.executionContext

Solution: Dedicated dispatcher for blocking operations

An efficient method of isolating the blocking behavior, such that it does not impact the rest of the system, is to prepare and use a dedicated dispatcher for all those blocking operations. This technique is often referred to as “bulk-heading” or simply “isolating blocking”.

In application.conf, the dispatcher dedicated to blocking behavior should be configured as follows:

copysourcemy-blocking-dispatcher {
  type = Dispatcher
  executor = "thread-pool-executor"
  thread-pool-executor {
    fixed-pool-size = 16
  }
  throughput = 1
}

A thread-pool-executor based dispatcher allows us to limit the number of threads it will host, and this way we gain tight control over the maximum number of blocked threads the system may use.

The exact size should be fine tuned depending on the workload you’re expecting to run on this dispatcher.

Whenever blocking has to be done, use the above configured dispatcher instead of the default one:

Scala

copysourceobject SeparateDispatcherFutureActor {
  def apply(): Behavior[Int] =
    Behaviors.setup { context =>
      implicit val executionContext: ExecutionContext =
        context.system.dispatchers.lookup(DispatcherSelector.fromConfig("my-blocking-dispatcher"))

      Behaviors.receiveMessage { i =>
        triggerFutureBlockingOperation(i)
        Behaviors.same
      }
    }

  def triggerFutureBlockingOperation(i: Int)(implicit ec: ExecutionContext): Future[Unit] = {
    println(s"Calling blocking Future: $i")
    Future {
      Thread.sleep(5000) // block for 5 seconds
      println(s"Blocking future finished $i")
    }
  }
}

Java

copysourceclass SeparateDispatcherFutureActor extends AbstractBehavior<Integer> {
  private final Executor ec;

  public static Behavior<Integer> create() {
    return Behaviors.setup(SeparateDispatcherFutureActor::new);
  }

  private SeparateDispatcherFutureActor(ActorContext<Integer> context) {
    super(context);
    ec =
        context
            .getSystem()
            .dispatchers()
            .lookup(DispatcherSelector.fromConfig("my-blocking-dispatcher"));
  }

  @Override
  public Receive<Integer> createReceive() {
    return newReceiveBuilder()
        .onMessage(
            Integer.class,
            i -> {
              triggerFutureBlockingOperation(i, ec);
              return Behaviors.same();
            })
        .build();
  }

  private static void triggerFutureBlockingOperation(Integer i, Executor ec) {
    System.out.println("Calling blocking Future on separate dispatcher: " + i);
    CompletableFuture<Integer> f =
        CompletableFuture.supplyAsync(
            () -> {
              try {
                Thread.sleep(5000);
                System.out.println("Blocking future finished: " + i);
                return i;
              } catch (InterruptedException e) {
                return -1;
              }
            },
            ec);
  }
}

The thread pool behavior is shown in the below diagram.

Messages sent to SeparateDispatcherFutureActorSeparateDispatcherCompletionStageActor and PrintActor are handled by the default dispatcher - the green lines, which represent the actual execution.

When blocking operations are run on the my-blocking-dispatcher, it uses the threads (up to the configured limit) to handle these operations. The sleeping in this case is nicely isolated to just this dispatcher, and the default one remains unaffected, allowing the rest of the application to proceed as if nothing bad was happening. After a certain period of idleness, threads started by this dispatcher will be shut down.

In this case, the throughput of other actors was not impacted - they were still served on the default dispatcher.

This is the recommended way of dealing with any kind of blocking in reactive applications.

For a similar discussion specifically about Pekko HTTP, refer to Handling blocking operations in Pekko HTTP.

Available solutions to blocking operations

The non-exhaustive list of adequate solutions to the “blocking problem” includes the following suggestions:

• Do the blocking call within a FutureCompletionStage, ensuring an upper bound on the number of such calls at any point in time (submitting an unbounded number of tasks of this nature will exhaust your memory or thread limits).
• Do the blocking call within a Future, providing a thread pool with an upper limit on the number of threads which is appropriate for the hardware on which the application runs, as explained in detail in this section.
• Dedicate a single thread to manage a set of blocking resources (e.g. a NIO selector driving multiple channels) and dispatch events as they occur as actor messages.
• Do the blocking call within an actor (or a set of actors) managed by a router, making sure to configure a thread pool which is either dedicated for this purpose or sufficiently sized.

The last possibility is especially well-suited for resources which are single-threaded in nature, like database handles which traditionally can only execute one outstanding query at a time and use internal synchronization to ensure this. A common pattern is to create a router for N actors, each of which wraps a single DB connection and handles queries as sent to the router. The number N must then be tuned for maximum throughput, which will vary depending on which DBMS is deployed on what hardware.

More dispatcher configuration examples

Fixed pool size

Configuring a dispatcher with fixed thread pool size, e.g. for actors that perform blocking IO:

copysourceblocking-io-dispatcher {
  type = Dispatcher
  executor = "thread-pool-executor"
  thread-pool-executor {
    fixed-pool-size = 32
  }
  throughput = 1
}

Cores

Another example that uses the thread pool based on the number of cores (e.g. for CPU bound tasks)

copysourcemy-thread-pool-dispatcher {
  # Dispatcher is the name of the event-based dispatcher
  type = Dispatcher
  # What kind of ExecutionService to use
  executor = "thread-pool-executor"
  # Configuration for the thread pool
  thread-pool-executor {
    # minimum number of threads to cap factor-based core number to
    core-pool-size-min = 2
    # No of core threads ... ceil(available processors * factor)
    core-pool-size-factor = 2.0
    # maximum number of threads to cap factor-based number to
    core-pool-size-max = 10
  }
  # Throughput defines the maximum number of messages to be
  # processed per actor before the thread jumps to the next actor.
  # Set to 1 for as fair as possible.
  throughput = 100
}

Pinned

A separate thread is dedicated for each actor that is configured to use the pinned dispatcher.

Configuring a PinnedDispatcher:

copysourcemy-pinned-dispatcher {
  executor = "thread-pool-executor"
  type = PinnedDispatcher
}

Note that thread-pool-executor configuration as per the above my-thread-pool-dispatcher example is NOT applicable. This is because every actor will have its own thread pool when using PinnedDispatcher, and that pool will have only one thread.

Note that it’s not guaranteed that the same thread is used over time, since the core pool timeout is used for PinnedDispatcher to keep resource usage down in case of idle actors. To use the same thread all the time you need to add thread-pool-executor.allow-core-timeout=off to the configuration of the PinnedDispatcher.

Thread shutdown timeout

Both the fork-join-executor and thread-pool-executor may shutdown threads when they are not used. If it’s desired to keep the threads alive longer there are some timeout settings that can be adjusted.

copysourcemy-dispatcher-with-timeouts {
  type = Dispatcher
  executor = "thread-pool-executor"
  thread-pool-executor {
    fixed-pool-size = 16
    # Keep alive time for threads
    keep-alive-time = 60s
    # Allow core threads to time out
    allow-core-timeout = off
  }
  # How long time the dispatcher will wait for new actors until it shuts down
  shutdown-timeout = 60s
}

When using the dispatcher as an ExecutionContext without assigning actors to it the shutdown-timeout should typically be increased, since the default of 1 second may cause too frequent shutdown of the entire thread pool.