Add ClusterStartupTaskActor.Execute to protobuff definition

The following message is being displayed in logs when a ClusterStartupTaskActor.Execute message is being sent through the wire.

Using the default Java serializer for class [com.lightbend.lagom.internal.persistence.cluster.ClusterStartupTaskActor$Execute$│
] which is not recommended because of performance implications. Use another serializer or disable this warning using the setting 'akka.actor.warn-about-java-serializer-usage'

There may be other message being used internally by Lagom that are using Java Serializer by default.

This is only visible when deploying in a cluster. Single node instances will never serializer/deserialize messages.

[jroper]

Akka serialized objects carry a serializer id (an integer), and that id is used to identify which serializer should deserialize it, not the type. So, if com.example.Foo is configured to be serialized with the Java serializer on one node, and then it’s sent to another node, it doesn’t matter what that node has configured for com.example.Foo, the Java serializer will be used to deserialize it.

So the important thing is that before any nodes make a switch to serializing with the new serializer, all nodes have the new serializer. Then when the switch is made, some nodes will still be configured to use the Java serializer, but that doesn’t matter because the serializer will be looked up by the id that comes with the message, which will be the id of the new serializer.

[patriknw]

That is correct if all serializers are backported. The migration guide is misleading, but there are a few serializers that haven’t been backported so I’m not sure we can say that it will just work for everyone. Those missing are related to remote deployment so it will not be a problem for Lagom.

A serialized remote message (or persistent event) consists of serializerId, the manifest and the payload. When deserializing that it is only looking at the serializerId to pick which Serializer to use for fromBinary. The message class (the bindings) is not used for deserialization. The manifest is only used within the Serializer to decide how to deserialize the payload, so one Serializer can handle many classes.

A somewhat more safe way is to first use enable-additional-serialization-bindings = off on both 2.4 and 2.5 nodes. Rolling upgrade to 2.5 for all nodes. Then switch to on in another rolling upgrade.