Serialization - Version 2.4.20

Serialization

Akka has a built-in Extension for serialization, and it is both possible to use the built-in serializers and to write your own.

The serialization mechanism is both used by Akka internally to serialize messages, and available for ad-hoc serialization of whatever you might need it for.

Usage

Configuration

For Akka to know which Serializer to use for what, you need edit your Configuration, in the "akka.actor.serializers"-section you bind names to implementations of the akka.serialization.Serializer you wish to use, like this:

akka {
  actor {
    serializers {
      java = "akka.serialization.JavaSerializer"
      proto = "akka.remote.serialization.ProtobufSerializer"
      myown = "docs.serialization.MyOwnSerializer"
    }
  }
}

After you've bound names to different implementations of Serializer you need to wire which classes should be serialized using which Serializer, this is done in the "akka.actor.serialization-bindings"-section:

akka {
  actor {
    serializers {
      java = "akka.serialization.JavaSerializer"
      proto = "akka.remote.serialization.ProtobufSerializer"
      myown = "docs.serialization.MyOwnSerializer"
    }

    serialization-bindings {
      "java.lang.String" = java
      "docs.serialization.Customer" = java
      "com.google.protobuf.Message" = proto
      "docs.serialization.MyOwnSerializable" = myown
      "java.lang.Boolean" = myown
    }
  }
}

You only need to specify the name of an interface or abstract base class of the messages. In case of ambiguity, i.e. the message implements several of the configured classes, the most specific configured class will be used, i.e. the one of which all other candidates are superclasses. If this condition cannot be met, because e.g. java.io.Serializable and MyOwnSerializable both apply and neither is a subtype of the other, a warning will be issued.

Note

If you are using Scala for your message protocol and your messages are contained inside of a Scala object, then in order to reference those messages, you will need use the fully qualified Java class name. For a message named Message contained inside the Scala object named Wrapper you would need to reference it as Wrapper$Message instead of Wrapper.Message.

Akka provides serializers for java.io.Serializable and protobuf com.google.protobuf.GeneratedMessage by default (the latter only if depending on the akka-remote module), so normally you don't need to add configuration for that; since com.google.protobuf.GeneratedMessage implements java.io.Serializable, protobuf messages will always be serialized using the protobuf protocol unless specifically overridden. In order to disable a default serializer, map its marker type to “none”:

akka.actor.serialization-bindings {
  "java.io.Serializable" = none
}

Verification

If you want to verify that your messages are serializable you can enable the following config option:

akka {
  actor {
    serialize-messages = on
  }
}

Warning

We only recommend using the config option turned on when you're running tests. It is completely pointless to have it turned on in other scenarios.

If you want to verify that your Props are serializable you can enable the following config option:

akka {
  actor {
    serialize-creators = on
  }
}

Warning

We only recommend using the config option turned on when you're running tests. It is completely pointless to have it turned on in other scenarios.

Programmatic

If you want to programmatically serialize/deserialize using Akka Serialization, here's some examples:

import akka.actor.*;
import akka.serialization.*;
ActorSystem system = ActorSystem.create("example");

// Get the Serialization Extension
Serialization serialization = SerializationExtension.get(system);

// Have something to serialize
String original = "woohoo";

// Find the Serializer for it
Serializer serializer = serialization.findSerializerFor(original);

// Turn it into bytes
byte[] bytes = serializer.toBinary(original);

// Turn it back into an object,
// the nulls are for the class manifest and for the classloader
String back = (String) serializer.fromBinary(bytes);

// Voilá!
assertEquals(original, back);

For more information, have a look at the ScalaDoc for akka.serialization._

Customization

So, lets say that you want to create your own Serializer, you saw the docs.serialization.MyOwnSerializer in the config example above?

Creating new Serializers

First you need to create a class definition of your Serializer, which is done by extending akka.serialization.JSerializer, like this:

import akka.actor.*;
import akka.serialization.*;
public class MyOwnSerializer extends JSerializer {

  // This is whether "fromBinary" requires a "clazz" or not
  @Override public boolean includeManifest() {
    return false;
  }

  // Pick a unique identifier for your Serializer,
  // you've got a couple of billions to choose from,
  // 0 - 40 is reserved by Akka itself
  @Override public int identifier() {
    return 1234567;
  }

  // "toBinary" serializes the given object to an Array of Bytes
  @Override public byte[] toBinary(Object obj) {
    // Put the code that serializes the object here
    // ... ...
  }

  // "fromBinary" deserializes the given array,
  // using the type hint (if any, see "includeManifest" above)
  @Override public Object fromBinaryJava(byte[] bytes,
                                         Class<?> clazz) {
    // Put your code that deserializes here
    // ... ...
  }
}

The manifest is a type hint so that the same serializer can be used for different classes. The manifest parameter in fromBinaryJava is the class of the object that was serialized. In fromBinary you can match on the class and deserialize the bytes to different objects.

Then you only need to fill in the blanks, bind it to a name in your Configuration and then list which classes that should be serialized using it.

Serializer with String Manifest

The Serializer illustrated above supports a class based manifest (type hint). For serialization of data that need to evolve over time the SerializerWithStringManifest is recommended instead of Serializer because the manifest (type hint) is a String instead of a Class. That means that the class can be moved/removed and the serializer can still deserialize old data by matching on the String. This is especially useful for Persistence.

The manifest string can also encode a version number that can be used in fromBinary to deserialize in different ways to migrate old data to new domain objects.

If the data was originally serialized with Serializer and in a later version of the system you change to SerializerWithStringManifest the manifest string will be the full class name if you used includeManifest=true, otherwise it will be the empty string.

This is how a SerializerWithStringManifest looks like:

public class MyOwnSerializer2 extends SerializerWithStringManifest {

  private static final String CUSTOMER_MANIFEST = "customer";
  private static final String USER_MANIFEST = "user";
  private static final String UTF_8 = StandardCharsets.UTF_8.name();
  
  // Pick a unique identifier for your Serializer,
  // you've got a couple of billions to choose from,
  // 0 - 40 is reserved by Akka itself
  @Override public int identifier() {
    return 1234567;
  }
  
  @Override public String manifest(Object obj) {
    if (obj instanceof Customer)
      return CUSTOMER_MANIFEST;
    else if (obj instanceof User)
      return USER_MANIFEST;
    else 
      throw new IllegalArgumentException("Unknown type: " + obj);
  }

  // "toBinary" serializes the given object to an Array of Bytes
  @Override public byte[] toBinary(Object obj) {
    // Put the real code that serializes the object here
    try {
      if (obj instanceof Customer)
        return ((Customer) obj).name.getBytes(UTF_8);
      else if (obj instanceof User)
        return ((User) obj).name.getBytes(UTF_8);
      else 
        throw new IllegalArgumentException("Unknown type: " + obj);
    } catch (UnsupportedEncodingException e) {
      throw new RuntimeException(e.getMessage(), e);
    }
  }

  // "fromBinary" deserializes the given array,
  // using the type hint
  @Override public Object fromBinary(byte[] bytes, String manifest) {
    // Put the real code that deserializes here
    try {
      if (manifest.equals(CUSTOMER_MANIFEST))
        return new Customer(new String(bytes, UTF_8));
      else if (manifest.equals(USER_MANIFEST))
        return new User(new String(bytes, UTF_8));
      else 
        throw new IllegalArgumentException("Unknown manifest: " + manifest);
    } catch (UnsupportedEncodingException e) {
      throw new RuntimeException(e.getMessage(), e);
    }
  }
}

You must also bind it to a name in your Configuration and then list which classes that should be serialized using it.

It's recommended to throw java.io.NotSerializableException in fromBinary if the manifest is unknown. This makes it possible to introduce new message types and send them to nodes that don't know about them. This is typically needed when performing rolling upgrades, i.e. running a cluster with mixed versions for while. NotSerializableException is treated as a transient problem in the TCP based remoting layer. The problem will be logged and message is dropped. Other exceptions will tear down the TCP connection because it can be an indication of corrupt bytes from the underlying transport.

Serializing ActorRefs

All ActorRefs are serializable using JavaSerializer, but in case you are writing your own serializer, you might want to know how to serialize and deserialize them properly. In the general case, the local address to be used depends on the type of remote address which shall be the recipient of the serialized information. Use Serialization.serializedActorPath(actorRef) like this:

import akka.actor.*;
import akka.serialization.*;
// Serialize
// (beneath toBinary)
String identifier = Serialization.serializedActorPath(theActorRef);

// Then just serialize the identifier however you like

// Deserialize
// (beneath fromBinary)
final ActorRef deserializedActorRef = extendedSystem.provider().resolveActorRef(
  identifier);
// Then just use the ActorRef

This assumes that serialization happens in the context of sending a message through the remote transport. There are other uses of serialization, though, e.g. storing actor references outside of an actor application (database, etc.). In this case, it is important to keep in mind that the address part of an actor’s path determines how that actor is communicated with. Storing a local actor path might be the right choice if the retrieval happens in the same logical context, but it is not enough when deserializing it on a different network host: for that it would need to include the system’s remote transport address. An actor system is not limited to having just one remote transport per se, which makes this question a bit more interesting. To find out the appropriate address to use when sending to remoteAddr you can use ActorRefProvider.getExternalAddressFor(remoteAddr) like this:

public class ExternalAddressExt implements Extension {
  private final ExtendedActorSystem system;

  public ExternalAddressExt(ExtendedActorSystem system) {
    this.system = system;
  }

  public Address getAddressFor(Address remoteAddress) {
    final scala.Option<Address> optAddr = system.provider()
      .getExternalAddressFor(remoteAddress);
    if (optAddr.isDefined()) {
      return optAddr.get();
    } else {
      throw new UnsupportedOperationException(
        "cannot send to remote address " + remoteAddress);
    }
  }
}

public class ExternalAddress extends
  AbstractExtensionId<ExternalAddressExt> implements ExtensionIdProvider {
  public static final ExternalAddress ID = new ExternalAddress();

  public ExternalAddress lookup() {
    return ID;
  }

  public ExternalAddressExt createExtension(ExtendedActorSystem system) {
    return new ExternalAddressExt(system);
  }
}

public class ExternalAddressExample {
  public String serializeTo(ActorRef ref, Address remote) {
    return ref.path().toSerializationFormatWithAddress(
        ExternalAddress.ID.get(system).getAddressFor(remote));
  }
}

Note

ActorPath.toSerializationFormatWithAddress differs from toString if the address does not already have host and port components, i.e. it only inserts address information for local addresses.

toSerializationFormatWithAddress also adds the unique id of the actor, which will change when the actor is stopped and then created again with the same name. Sending messages to a reference pointing the old actor will not be delivered to the new actor. If you do not want this behavior, e.g. in case of long term storage of the reference, you can use toStringWithAddress, which does not include the unique id.

This requires that you know at least which type of address will be supported by the system which will deserialize the resulting actor reference; if you have no concrete address handy you can create a dummy one for the right protocol using new Address(protocol, "", "", 0) (assuming that the actual transport used is as lenient as Akka’s RemoteActorRefProvider).

There is also a default remote address which is the one used by cluster support (and typical systems have just this one); you can get it like this:

public class DefaultAddressExt implements Extension {
  private final ExtendedActorSystem system;

  public DefaultAddressExt(ExtendedActorSystem system) {
    this.system = system;
  }

  public Address getAddress() {
    return system.provider().getDefaultAddress();
  }
}

public class DefaultAddress extends
    AbstractExtensionId<DefaultAddressExt> implements ExtensionIdProvider {
  public static final DefaultAddress ID = new DefaultAddress();

  public DefaultAddress lookup() {
    return ID;
  }

  public DefaultAddressExt createExtension(ExtendedActorSystem system) {
    return new DefaultAddressExt(system);
  }
}

Deep serialization of Actors

The recommended approach to do deep serialization of internal actor state is to use Akka Persistence.

A Word About Java Serialization

When using Java serialization without employing the JavaSerializer for the task, you must make sure to supply a valid ExtendedActorSystem in the dynamic variable JavaSerializer.currentSystem. This is used when reading in the representation of an ActorRef for turning the string representation into a real reference. DynamicVariable is a thread-local variable, so be sure to have it set while deserializing anything which might contain actor references.

Serialization compatibility

It is not safe to mix major Scala versions when using the Java serialization as Scala does not guarantee compatibility and this could lead to very surprising errors.

If using the Akka Protobuf serializers (implicitly with akka.actor.allow-java-serialization = off or explicitly with enable-additional-serialization-bindings = true) for the internal Akka messages those will not require the same major Scala version however you must also ensure the serializers used for your own types does not introduce the same incompatibility as Java serialization does.

Contents