LATEST RELEASE: 24.05 — Zero Trust, DB Sharding, Java 21, Rust Support, and More. Read Release Notes

News & Articles

Full archive

October 04


Akka Typed: New Cluster Tools API

Note: Code examples in this blog post are out of date, see the Akka documentation for latest information on this topic.

In previous post we looked at the the Cluster and Receptionist for Akka Typed. In this post you will be introduced to the new typed APIs for Distributed Data, Cluster Singleton and Cluster Sharding. These features are still using the existing implementations from the untyped modules and they require that you run with the untyped ActorSystem as described in Akka Typed: Coexistence.

Distributed Data

The API for Distributed Data is very similar to the untyped API, with the difference that you have to pass the replyTo actor reference in the messages as there is no sender. Here is an example:

import akka.cluster.ddata.GCounter
import akka.cluster.ddata.GCounterKey
import akka.cluster.ddata.ReplicatedData
import akka.typed.ActorRef
import akka.typed.Behavior
import akka.typed.cluster.ddata.scaladsl.DistributedData
import akka.typed.cluster.ddata.scaladsl.Replicator
import akka.typed.scaladsl.Actor
import akka.typed.scaladsl.adapter._

object Counter {
  sealed trait ClientCommand
  final case object Increment extends ClientCommand
  final case class GetValue(replyTo: ActorRef[Int]) extends ClientCommand

  private sealed trait InternalMsg extends ClientCommand
  private case class InternalUpdateResponse[A <: ReplicatedData](rsp: Replicator.UpdateResponse[A]) extends InternalMsg
  private case class InternalGetResponse[A <: ReplicatedData](rsp: Replicator.GetResponse[A]) extends InternalMsg

  val Key = GCounterKey("counter")

  def behavior: Behavior[ClientCommand] =
    Actor.deferred[ClientCommand] { ctx 
      // The ddata types still need the implicit untyped Cluster.
      // We will look into another solution for that.
      implicit val cluster = akka.cluster.Cluster(ctx.system.toUntyped)
      val replicator: ActorRef[Replicator.Command] = DistributedData(ctx.system).replicator

      // use message adapters to map the external messages (replies) to the message types
      // that this actor can handle (see InternalMsg)
      val updateResponseAdapter: ActorRef[Replicator.UpdateResponse[GCounter]] =

      val getResponseAdapter: ActorRef[Replicator.GetResponse[GCounter]] =

      Actor.immutable[ClientCommand] { (ctx, msg) 
        msg match {
          case Increment 
            replicator ! Replicator.Update(Key, GCounter.empty, Replicator.WriteLocal, updateResponseAdapter)(_ + 1)

          case GetValue(replyTo) 
            replicator ! Replicator.Get(Key, Replicator.ReadLocal, getResponseAdapter, Some(replyTo))

          case internal: InternalMsg  internal match {
            case InternalUpdateResponse(_)  Actor.same // ok

            case InternalGetResponse(rsp @ Replicator.GetSuccess(Key, Some(replyTo: ActorRef[Int] @unchecked))) 
              val value = rsp.get(Key).value.toInt
              replyTo ! value

            case InternalGetResponse(rsp) 
              Actor.unhandled // not dealing with failures


Note that the messages such as Replicator.Update are defined in akka.typed.cluster.ddata.scaladsl.Replicator, but the actual CRDTs are the same as in untyped, for example akka.cluster.ddata.GCounter.

Cluster Singleton

The API for Cluster Singleton is simplified compared to the untyped API. It is based on an extension that provides a single spawn method that starts both the SingletonManager and the SingletonProxy. It returns an ActorRef to the SingletonProxy, which is used for sending messages to the singleton instance that is running somewhere (oldest node) in the cluster. You should call this ClusterSingleton.spawn method on all nodes at system startup.

If the node doesn’t have the given role then only the proxy is started, and not the manager.

Here is an example of a global sequence number generator:

object SequenceNumberGenerator {

  sealed trait Message
  final case class Next(replyTo: ActorRef[Long]) extends Message
  case object Stop extends Message

  def generator(n: Long = System.currentTimeMillis()): Behavior[Message] = Actor.immutable {
    // a real generator would perhaps store the counter with Distributed Data to
    // be able to continue with next number after fail over
    case (_, Next(replyTo)) =>
      val next = n + 1
      replyTo ! next
    case (_, Stop) =>


and this is how to start the singleton manager and proxy:

val singletonProxy: ActorRef[SequenceNumberGenerator.Message] =
    behavior = SequenceNumberGenerator.generator(),
    singletonName = "seqNr",
    props = Props.empty,
    settings = ClusterSingletonSettings(typedSystem),
    terminationMessage = SequenceNumberGenerator.Stop)


Cluster Sharding

The API for Cluster Sharding has also been simplified. You start the sharded type with the ClusterSharding extension:

import akka.typed.cluster.sharding.ClusterSharding
import akka.typed.cluster.sharding.ClusterShardingSettings

  behavior = BlogPost.shardingBehavior,
  props = Props.empty,
  typeKey = BlogPost.ShardingTypeName,
  settings = ClusterShardingSettings(typedSystem),
  maxNumberOfShards = BlogPost.MaxNumberOfShards,
  handOffStopMessage = PassivatePost)

The BlogPost actor is a persistent actor but we will not look at the details of that now. Persistent actors for Akka Typed will be described in a separate blog post in a few days.

The typeKey is defined as:

val ShardingTypeName = EntityTypeKey[BlogCommand]("BlogPost")

This key is then used to retrieve an EntityRef for a given entity identifier. Messages to a specific entity are then sent via this EntityRef.

  val postId = UUID.randomUUID().toString
  val content = PostContent(postId, "Title...", "Body...")

  val entityRef: EntityRef[BlogCommand] =
    sharding.entityRefFor(BlogPost.ShardingTypeName, postId)

  entityRef ! AddPost(content, addPostReplyAdapter)

Note that in above ClusterSharding.spawn we are not defining any extractEntityId or extractShardId functions. Instead, the messages are sent in an envelope message that carry the entity identifier, and the shard identifier is derived from the entity id and the maxNumberOfShards parameter with a simple hashing function. That means that the actual domain messages don’t have to include the entity id. This is what the EntityRef is using under the hood. It has the entity id and can wrap the sent message in such an envelope.

It is still possible to define extractor functions and send messages directly to the shard region instead of using the EntityRef, if you prefer that.

The full source code of these examples, are available in patriknw/akka-typed-blog.

Let us know what you think about these APIs. In the next blog post we will take a look at the typed APIs for Persistence.

This post is part of the "Akka Typed Cluster and Persistence" series. Explore other posts in this series:

  1. Akka Typed: New Cluster API
  2. → Akka Typed: New Cluster Tools API
  3. Akka Typed: New Persistence API