Clarification about "A cluster should only span nodes that are running the same service."

The [Lagom Cluster] doc states:(https://www.lagomframework.com/documentation/latest/scala/Cluster.html#Cluster-composition):

A cluster should only span nodes that are running the same service.

What exactly does this mean relative to Lagom service descriptors?

For example, suppose I define 2 service descriptors, A and B.
Suppose then that I deploy 1x replica of A and 2x replicas of B.

According to the doc, this suggests that there would be 2 distinct Akka clusters:

  • one for the single replica of A
  • another for the two replicas of B

What happens if one does not follow this guideline?

For example, deploying all replicas of A and B to the same Akka cluster.
I have the impression that the problems I’m experiencing in this thread are about deploying everything to the same cluster.

  • Nicolas.

That’s correct, each service should form an independent cluster. This keeps the services loosely coupled and independently deployable. You can read more about this recommendation here:

https://doc.akka.io/docs/akka/current/cluster-usage.html#when-and-where-to-use-akka-cluster