Containerizing and Deploying Boot Apps

Dockerfiles, health probes, graceful shutdown and Kubernetes deployment.

Expert12 min read

Read these first

Actuator Endpoints

Why this matters

"Works on my machine" disappears when the same Docker image runs in dev, staging, and production — only environment variables change.
Kubernetes rolling updates deploy new BookStore versions with zero downtime by gradually replacing pods while readiness probes gate traffic.
Graceful shutdown ensures in-flight checkout requests complete before the pod terminates, preventing lost orders during deploys.

StreamHub production architecture (AWS)

End-to-end path from user to data — reference this when placing any new service.

Deployment essentials

Multi-stage Dockerfile — build with Maven in one stage, run a slim JRE image in the final stage.
Liveness probe — restarts the pod if the JVM is wedged; hits /actuator/health/liveness.
Readiness probe — removes the pod from load balancing until it can serve traffic.
Resource limits — requests and limits for CPU and memory prevent one pod from starving the node.
Graceful shutdown — server.shutdown: graceful + terminationGracePeriodSeconds let requests drain.

Production Dockerfile

Java

# Build stage
FROM eclipse-temurin:21-jdk-alpine AS build
WORKDIR /app
COPY mvnw pom.xml ./
COPY .mvn .mvn
RUN ./mvnw dependency:go-offline -B
COPY src ./src
RUN ./mvnw package -DskipTests -B

# Runtime stage
FROM eclipse-temurin:21-jre-alpine
RUN addgroup -S bookstore && adduser -S bookstore -G bookstore
WORKDIR /app
COPY --from=build /app/target/bookstore-api-*.jar app.jar
USER bookstore
EXPOSE 8080

ENTRYPOINT ["java", \
    "-XX:+UseContainerSupport", \
    "-XX:MaxRAMPercentage=75.0", \
    "-Djava.security.egd=file:/dev/./urandom", \
    "-jar", "app.jar"]

Graceful shutdown configuration

Java

server:
  shutdown: graceful

spring:
  lifecycle:
    timeout-per-shutdown-phase: 30s

Java

@Bean
public TomcatConnectorCustomizer gracefulShutdownCustomizer() {
    return connector -> connector.setProperty("connectionTimeout", "20000");
}

Kubernetes deployment manifest

Java

apiVersion: apps/v1
kind: Deployment
metadata:
  name: bookstore-api
  labels:
    app: bookstore-api
spec:
  replicas: 3
  selector:
    matchLabels:
      app: bookstore-api
  template:
    metadata:
      labels:
        app: bookstore-api
    spec:
      terminationGracePeriodSeconds: 45
      containers:
        - name: bookstore-api
          image: registry.example.com/bookstore-api:2.4.1
          ports:
            - containerPort: 8080
          env:
            - name: SPRING_PROFILES_ACTIVE
              value: prod
            - name: SPRING_DATASOURCE_URL
              valueFrom:
                secretKeyRef:
                  name: bookstore-db
                  key: url
          resources:
            requests:
              cpu: 500m
              memory: 512Mi
            limits:
              cpu: 1000m
              memory: 1Gi
          livenessProbe:
            httpGet:
              path: /actuator/health/liveness
              port: 8080
            initialDelaySeconds: 60
            periodSeconds: 10
          readinessProbe:
            httpGet:
              path: /actuator/health/readiness
              port: 8080
            initialDelaySeconds: 30
            periodSeconds: 5
---
apiVersion: v1
kind: Service
metadata:
  name: bookstore-api
spec:
  selector:
    app: bookstore-api
  ports:
    - port: 80
      targetPort: 8080
  type: ClusterIP

Build and deploy pipeline

Java

docker build -t registry.example.com/bookstore-api:2.4.1 .
docker push registry.example.com/bookstore-api:2.4.1
kubectl apply -f k8s/deployment.yaml
kubectl rollout status deployment/bookstore-api

Quick recall

Everything you need if you only revisit this box.

Multi-stage Dockerfiles keep production images small — JRE only, no Maven or source code.
Enable server.shutdown: graceful and set terminationGracePeriodSeconds above your longest request.
Liveness restarts wedged pods; readiness controls whether the pod receives traffic.
Set CPU/memory requests and limits; use -XX:MaxRAMPercentage=75.0 so the JVM respects container memory.
Point probes at /actuator/health/liveness and /actuator/health/readiness with adequate initialDelaySeconds.

Test yourself

Answer these before moving on — recall is what makes it stick.