Classic Scheduler

For the new API see typed scheduling.

Dependency

To use Scheduler, you must add the following dependency in your project:

sbt

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

Maven

<properties>
  <scala.binary.version>2.13</scala.binary.version>
</properties>
<dependencyManagement>
  <dependencies>
    <dependency>
      <groupId>org.apache.pekko</groupId>
      <artifactId>pekko-bom_${scala.binary.version}</artifactId>
      <version>1.0.3</version>
      <type>pom</type>
      <scope>import</scope>
    </dependency>
  </dependencies>
</dependencyManagement>
<dependencies>
  <dependency>
    <groupId>org.apache.pekko</groupId>
    <artifactId>pekko-actor_${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_${versions.ScalaBinary}"
}

Introduction

Sometimes, the need for making things happen in the future arises, and where do you go look then? Look no further than ActorSystemActorSystem! There you find the schedulergetScheduler() method that returns an instance of SchedulerScheduler, this instance is unique per ActorSystem and is used internally for scheduling things to happen at specific points in time.

You can schedule sending of messages to actors and execution of tasks (functions or Runnable). You will get a CancellableCancellable back that you can call cancelcancel() on to cancel the execution of the scheduled operation.

When scheduling periodic or single messages in an actor to itself, it is recommended to use the Actor Timers instead of using the SchedulerScheduler directly.

The scheduler in Pekko is designed for high-throughput of thousands up to millions of triggers. The prime use-case being triggering Actor receive timeouts, Future timeouts, circuit breakers and other time dependent events which happen all-the-time and in many instances at the same time. The implementation is based on a Hashed Wheel Timer, which is a known datastructure and algorithm for handling such use cases, refer to the Hashed and Hierarchical Timing Wheels whitepaper by Varghese and Lauck if you’d like to understand its inner workings.

The Pekko scheduler is not designed for long-term scheduling (see pekko-quartz-scheduler instead for this use case) nor is it to be used for highly precise firing of the events. The maximum amount of time into the future you can schedule an event to trigger is around 8 months, which in practice is too much to be useful since this would assume the system never went down during that period. If you need long-term scheduling we highly recommend looking into alternative schedulers, as this is not the use-case the Pekko scheduler is implemented for.

Some examples

Scala

copysourceimport org.apache.pekko
import pekko.actor.Actor
import pekko.actor.Props
import scala.concurrent.duration._

Java

copysourceimport java.time.Duration;

Schedule to send the “foo”-message to the testActor after 50ms:

Scala

copysource// Use the system's dispatcher as ExecutionContext
import system.dispatcher

// Schedules to send the "foo"-message to the testActor after 50ms
system.scheduler.scheduleOnce(50 milliseconds, testActor, "foo")

Java

copysourcesystem
    .scheduler()
    .scheduleOnce(
        Duration.ofMillis(50), testActor, "foo", system.dispatcher(), ActorRef.noSender());

Schedule a functionRunnable, that sends the current time to the testActor, to be executed after 50ms:

Scala

copysource// Schedules a function to be executed (send a message to the testActor) after 50ms
system.scheduler.scheduleOnce(50 milliseconds) {
  testActor ! System.currentTimeMillis
}

Java

copysourcesystem
    .scheduler()
    .scheduleOnce(
        Duration.ofMillis(50),
        new Runnable() {
          @Override
          public void run() {
            testActor.tell(System.currentTimeMillis(), ActorRef.noSender());
          }
        },
        system.dispatcher());

Schedule to send the “Tick”-message to the tickActor after 0ms repeating every 50ms:

Scala

copysourceclass TickActor extends Actor {
  def receive: Receive = {
    case "tick" => // Do something
  }
}

  val tickActor = system.actorOf(Props(classOf[TickActor]))
  // Use system's dispatcher as ExecutionContext
  import system.dispatcher

  // This will schedule to send the Tick-message
  // to the tickActor after 0ms repeating every 50ms
  val cancellable =
    system.scheduler.scheduleWithFixedDelay(Duration.Zero, 50.milliseconds, tickActor, "tick")

  // This cancels further Ticks to be sent
  cancellable.cancel()

Java

copysourceclass Ticker extends AbstractActor {
  @Override
  public Receive createReceive() {
    return receiveBuilder()
        .matchEquals(
            "Tick",
            m -> {
              // Do something
            })
        .build();
  }
}

ActorRef tickActor = system.actorOf(Props.create(Ticker.class, this));

// This will schedule to send the Tick-message
// to the tickActor after 0ms repeating every 50ms
Cancellable cancellable =
    system
        .scheduler()
        .scheduleWithFixedDelay(
            Duration.ZERO,
            Duration.ofMillis(50),
            tickActor,
            "Tick",
            system.dispatcher(),
            ActorRef.noSender());

// This cancels further Ticks to be sent
cancellable.cancel();

Schedule periodically

Scheduling of recurring tasks or messages can have two different characteristics:

• fixed-delay - The delay between subsequent execution will always be (at least) the given delay. Use scheduleWithFixedDelay.
• fixed-rate - The frequency of execution over time will meet the given interval. Use scheduleAtFixedRate.

If you are uncertain of which one to use you should pick scheduleWithFixedDelay.

When using fixed-delay it will not compensate the delay between tasks or messages if the execution takes long time or if scheduling is delayed longer than specified for some reason. The delay between subsequent execution will always be (at least) the given delay. In the long run, the frequency of execution will generally be slightly lower than the reciprocal of the specified delay.

Fixed-delay execution is appropriate for recurring activities that require “smoothness.” In other words, it is appropriate for activities where it is more important to keep the frequency accurate in the short run than in the long run.

When using fixed-rate it will compensate the delay for a subsequent task if the previous tasks took too long to execute. For example, if the given interval is 1000 milliseconds and a task takes 200 milliseconds to execute the next task will be scheduled to run after 800 milliseconds. In such cases, the actual execution interval will differ from the interval passed to the scheduleAtFixedRate method.

If the execution of the tasks takes longer than the interval, the subsequent execution will start immediately after the prior one completes (there will be no overlap of executions). This also has the consequence that after long garbage collection pauses or other reasons when the JVM was suspended all “missed” tasks will execute when the process wakes up again. For example, scheduleAtFixedRate with an interval of 1 second and the process is suspended for 30 seconds will result in 30 tasks (or messages) being executed in rapid succession to catch up. In the long run, the frequency of execution will be exactly the reciprocal of the specified interval.

Fixed-rate execution is appropriate for recurring activities that are sensitive to absolute time or where the total time to perform a fixed number of executions is important, such as a countdown timer that ticks once every second for ten seconds.

The Scheduler interface

The actual scheduler implementation is loaded reflectively upon ActorSystemActorSystem start-up, which means that it is possible to provide a different one using the pekko.scheduler.implementation configuration property. The referenced class must implement the SchedulerSchedulerAbstractSchedulerAbstractScheduler interface.

The Cancellable interface

Scheduling a task will result in a CancellableCancellable (or throw an IllegalStateException if attempted after the scheduler’s shutdown). This allows you to cancel something that has been scheduled for execution.