ibm/akka-cluster-deploy-kubernetes

Deploy an AKKA cluster to kubernetes

akka-cluster-deploy-kubernetes's Issues

README file changes - improvements to Kubernetes instructions

Comments from Lightbend

StatefulSet should specify "imagePullPolicy: IfNotPresent" instead of "Always"
if they intend for the example to work with Minikube.

It would be helpful to have some more comments about prereqs, kubectl, minikube and so on.

Again, if providing instructions to run locally on minikube, you should mention that you need to run "eval$(minikube docker-env)" before building the application

Suggestions on the build steps in the README

Comments from Lightbend Reviewers

• sbt-native-packager used in this example is old, so it suffers from the duplicate layer bug
• DS_STORE is in version control, it shouldn't be.
• Various akka.io.dns settings in build.sbt are for a library that isn't included in this application and therefore not relevant; these settings should be removed

Diagram on web and Readme instructions are mismatched.

README file

Suggestions on README

README file changes

README file typos, grammar and clarity recommendations

There are numerous typos, grammatical errors, and sentence fragments in the README file which should be corrected.

In the Overview section, the description of the distinction between Akka Cluster and Kubernetes is a somewhat confusing. It leaves the impression that Akka Cluster runs your application instead of it being used by your application as a library. Also, the statements about seed nodes being masters implies Akka Cluster
isn't peer-to-peer. It is.

I offering the following as a replacement for the current README text:
.....................

In this Code Pattern, we will deploy a simple Akka application on Kubernetes to demonstrate how you can build an inherently scalable, distributed solution.

Akka is “a set of open-source libraries for designing scalable, resilient systems that span processor cores and networks”. Akka Cluster, in particular, enables the distribution of an application over a cluster, thereby lending elasticity to the application by automatically scaling out or in with the number of members in the cluster. Kubernetes is a container orchestration platform and has the ability to automatically scale the number of pods up or down in response to workload. Together, the two work in concert: your Akka cluster expands automatically as Kubernetes scales the number of pods running for a particular Kubernetes deployment.

One key enabler of this elasticity is the ability for new nodes to join the Akka cluster automatically. Akka clusters are a peer-to-peer framework, however some nodes must be designated “seed” nodes when the cluster is initialized. These serve as an initial point of contact for new nodes. When new nodes are brought online, they can join the cluster automatically by first sending a message to all seed nodes in the cluster, followed by a “join” command to the seed node that responds first. Without seed nodes you would have to manually or programmatically add nodes to the cluster.

This solution uses Docker to build a single container image. In Kubernetes, we use StatefulSet rather than Deployment to guarantee the orderly startup of seed nodes.

Overview