A basic stream aggregator that is built with help of Reactor Project (a fourth-generation reactive library, based on the Reactive Streams specification, for building non-blocking applications on the JVM).

This application is developed to be deployed in Kubernetes cluster, listen to the stream from other similar service conway-life-stream-server (deployed in the same cluster) and aggregates results with 2 internal streams and eventually publish the resulting aggregated stream through the single HTTP endpoint.

An oversimplified version of stream can be illustrated there:

• Manage data streams with Reactor Project

• Reactor library
• Spring WebFlux
• Java 15
• Java records
• GitHub Container Registry - a new Docker image repo from GitHub

• Reactor library
• Spring WebFlux

• JDK 15

TBD

As a build tool I use gradle and gradle-wrapper, so you can easily compile and build application with ./gradlew build and run with ./gradlew run.

For integration and deployment was used GitHub's framework - GitHub Actions. Workflow source is available there Active workflow could be found there.

For deployment Kubernetes cluster was used. Deployment descriptors are available in deployment sub-folder.

./gradlew build
docker build -t ghcr.io/fedor-malyshkin/conway-life-stream-reactor-client:latest -f docker/Dockerfile build/libs

export CR_PAT=YOUR_TOKEN
echo $CR_PAT | docker login ghcr.io -u fedor-malyshkin --password-stdin
docker push ghcr.io/fedor-malyshkin/conway-life-stream-reactor-client:latest

The initial deployment is done by the commands (in deployment sub-folder):

kubectl apply -f stream-reactor-client-deployment.yml
kubectl apply -f stream-reactor-client-service.yml
kubectl apply -f stream-reactor-client-loadbalancer.yml

The assigned IP can be found with the command:

kubectl describe service/stream-server-loadbalancer | egrep Ingress

Whereas the correctness of work with the command:

curl http://<IP>/health

As netty intensively uses some unsafe JVM features - they should be open by the set of special commands. Add such keys to your java app and don't forget to use them in your IDE during test (Dockerfile is updated for it)

--add-opens java.base/jdk.internal.misc=ALL-UNNAMED
-Dio.netty.tryReflectionSetAccessible=true
--illegal-access=warn

Don't forget to attach native JNI library that add features specific to a particular platform, generate less garbage, and generally improve performance when compared to the NIO based transport.

Gradle:

implementation "io.netty:netty-transport-native-epoll:${EPOLL_NATIVE_VER}:linux-x86_64"

It requires pay a close attention to version of serializer you use as support for java records has been added recently enough.

{
  "conwayCountOfAlive": 261,
  "conwayCountOfDead": 4739,
  "engVowelsStat": {
    "mean": 1.6,
    "variance": 0.46315789473684216,
    "standardDeviation": 0.6805570473787206,
    "max": 3.0,
    "min": 1.0,
    "n": 20,
    "sum": 32.0
  },
  "engConsonantsStat": {
    "mean": 2.0,
    "variance": 0.9473684210526315,
    "standardDeviation": 0.9733285267845752,
    "max": 5.0,
    "min": 1.0,
    "n": 20,
    "sum": 40.0
  },
  "engLengthStat": {
    "mean": 3.6,
    "variance": 2.0421052631578944,
    "standardDeviation": 1.429022485182754,
    "max": 8.0,
    "min": 2.0,
    "n": 20,
    "sum": 72.0
  },
  "engCountOfWords": 20,
  "rusVowelsStat": {
    "mean": 2.05,
    "variance": 1.628947368421053,
    "standardDeviation": 1.2763022245616642,
    "max": 5.0,
    "min": 0.0,
    "n": 20,
    "sum": 41.0
  },
  "rusConsonantsStat": {
    "mean": 3.75,
    "variance": 4.197368421052632,
    "standardDeviation": 2.0487480130686233,
    "max": 7.0,
    "min": 0.0,
    "n": 20,
    "sum": 75.0
  },
  "rusLengthStat": {
    "mean": 5.8,
    "variance": 9.43157894736842,
    "standardDeviation": 3.071087583799658,
    "max": 10.0,
    "min": 1.0,
    "n": 20,
    "sum": 116.0
  },
  "rusCountOfWords": 20,
  "mills": 1610570220753
}