Kubernetes Discovery with Embedded Infinispan

What You Will Learn

How to configure embedded Infinispan to form a cluster on Kubernetes using JGroups DNS_PING for node discovery, and how to build and deploy the application using JKube.

Prerequisites

Java 17+
A Kubernetes cluster (e.g., Minikube, OpenShift)
kubectl configured to access the cluster

Step 1: Add the Embedded Dependency

Add Infinispan core to your pom.xml:

<dependency>
   <groupId>org.infinispan</groupId>
   <artifactId>infinispan-core</artifactId>
</dependency>

Step 2: Configure JGroups Kubernetes Transport

Use the built-in Kubernetes JGroups configuration and pass the DNS query via a system property:

      //Configure Infinispan to use default transport and Kubernetes configuration
      GlobalConfiguration globalConfig = new GlobalConfigurationBuilder()
       .transport()
            .defaultTransport()
            .addProperty("configurationFile", "default-configs/default-jgroups-kubernetes.xml")
            .build();

The DNS query is set through the jgroups.dns.query system property, which points to the headless Kubernetes service:

<options.dnsPing>
   -Djgroups.dns.query=${options.service}.${options.namespace}.svc.cluster.local
</options.dnsPing>

Step 3: Create a Replicated Cache

Create a replicated cache so every pod in the deployment holds all entries:

         // We need a distributed cache for the purpose of this demo
         Configuration cacheConfiguration = new ConfigurationBuilder()
               .clustering()
               .cacheMode(CacheMode.REPL_SYNC)
               .build();

         Cache<String, String> cache = cacheManager.createCache("kubernetes", cacheConfiguration);

Step 4: Populate the Cache Periodically

Each pod updates its own entry on a schedule, demonstrating cluster-wide data replication:

         // Each cluster member will update its own entry in the cache
         String hostname = Inet4Address.getLocalHost().getHostName();
         ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(1);
         scheduler.scheduleAtFixedRate(() -> {
                  String time = Instant.now().toString();
                  cache.put(hostname, time);
                  System.out.println("[" + time + "][" + hostname + "] Values from the cache: ");
                  System.out.println(cache.entrySet());
               },
               0, 2, TimeUnit.SECONDS);

Step 5: Build and Deploy to Kubernetes

Build the Docker image and deploy to Kubernetes using JKube:

mvn package k8s:build k8s:resource k8s:apply -Pdocker

This creates a deployment with 2 replicas and a headless service for DNS-based cluster discovery.

What’s Next

Try replicated mode locally before deploying to Kubernetes
Explore distributed mode for large datasets where full replication is not feasible
Add cache listeners to react to cluster events