Best Practices to Spring (or Java) to Kubernetes Faster and Easier

For Spring developers, building containers is a common obstacle on the road to Kubernetes adoption. Traditionally, Dockerfiles define container builds imperatively, but can also be cumbersome, error-prone, and slow. The development cycle can be slow. IDE support is generally poor. Most likely, your Java application will crash with an OOMKill message.

In this talk, we'll introduce tools and best practices to help you adopt Kubernetes faster and easier. This includes tools such as Jib and Skaffold for fast image build and development cycle turnaround time. We'll also discuss best practices for configuring your Java applications to run inside of Kubernetes, such as health checks, zero downtime deployment, externalizing configurations, logging, and understanding memory usage to avoid OOMKilled situations.

Talk at CodeOne and SpringOne. Recording from SpringOne: https://www.youtube.com/watch?v=YTPUNesUIbI&list=PL4uYfigiauVYH4OwOyq8FGbPQOn-JueEf&index=3&t=0s

Transcript

1. Kubernetes Faster and Easier
2. 2 @saturnism @gcpcloud Ray Tsang Developer Advocate Google Cloud Platform Java Champion Spring Cloud GCP spring.io/projects/spring-cloud-gcp gcplab.me/spring @saturnism | saturnism.me
3. 3 @saturnism @gcpcloud Ray Tsang Traveler Photographer flickr.com/saturnism
4. 4 @saturnism @gcpcloud https://www.youtube.com/watch?v=Bcs-inRnLDc
5. 5 @saturnism @gcpcloud 5 Your App
6. 6 @saturnism @gcpcloud Don't start with Kubernetes
7. 7 @saturnism @gcpcloud It all starts with your application Twelve-Factor App 12factor.net
8. 8 @saturnism @gcpcloud Test, Test, Test Local Mock, Wiremock, Contract TestContainers
9. 9 @saturnism @gcpcloud 9 Containers
10. 10 @saturnism @gcpcloud Power → Responsibility Runtime Environments may be Your Responsibility Now!
11. 11 @saturnism @gcpcloud Choose a JDK Container Aware OpenJDK 8u192 or above
12. 12 @saturnism @gcpcloud OOMKilled Cloud Foundry Buildpack Memory Calculator https://github.com/cloudfoundry/java-buildpack-memory-calculator
13. 13 @saturnism @gcpcloud Native Memory Tracking -XX:NativeMemoryTracking=summary -XX:+PrintNMTStatistics (Doesn't work when set via JAVA_TOOL_OPTIONS - must be part of the argument)
14. 14 @saturnism @gcpcloud Container Best Practices saturnism.me/talk/docker-tips-and-tricks/ What's in that image? Don't run as root Multi-stage build Create small image Fat JAR to Thin JAR Layering Build cache Pin versions Reduce layer size ...
15. 15 @saturnism @gcpcloud Just Jib It
16. 16 @saturnism @gcpcloud Don't write to container filesystem! Those logs!
17. 17 @saturnism @gcpcloud Configuration via external sources Environmental variable Command line arguments
18. 18 @saturnism @gcpcloud 18 Finally, Kubernetes!
19. 19 @saturnism @gcpcloud Local Kubernetes Linux - consider k3s, k3d, kind, … Mac - Docker for Desktop, Minikube
20. 20 @saturnism @gcpcloud Keep Base Deployment Simple kubectl create deployment myservice --image=... --dry-run -oyaml > k8s/deployment.yaml kubectl create svc clusterip myservice --tcp=8080:8080 --dry-run -oyaml/service.yaml
21. 21 @saturnism @gcpcloud Continuous Development skaffold
22. 22 @saturnism @gcpcloud Resource Limits If you don't' set it, your app may use all the memory... Set it at namespace level, or for individual deployments
23. 23 @saturnism @gcpcloud Environments Kustomize
24. 24 @saturnism @gcpcloud Liveness Probe Readiness Probe Signals Lifecycle Hooks
25. 25 @saturnism @gcpcloud Liveness Probe → Restarts Readiness Probe → Remove from Service Signals → Shutdown or Killed Lifecycle Hooks → PreStart, PreStop
26. 26 @saturnism @gcpcloud When to use? Failure Means... Practices Example Liveness Probe If application is alive. Application will be restarted, and that a restart will help recover. Runs on serving port of the application, e.g., 8080. Don't check dependency. E.g., don't check dependent database connection, etc. A simple /alive URL that returns 200. Readiness Probe Ready to serve requests. Take the pod instance out of load balancer. Flip to ready when application has done all the initializations (cache preloaded). Upon SIGTERM, flip readiness to false. See Graceful Shutdown. /actuator/health on the management port.
27. 27 @saturnism @gcpcloud Anatomy of a Graceful Shutdown 1. Receive SIGTERM or PreStop Lifecycle Hook 2. Fail Readiness Probe 3. Receive requests until Kubernetes detects readiness probe failure 4. Kubernetes removes pod endpoint from Service 5. Finish serving in-flight requests 6. Shutdown
28. 28 @saturnism @gcpcloud Production is HARD Pod Security Policy / Pod Security Context Expect your app to not work in production environment with hardened security Try this early and fix issues
29. 29 @saturnism @gcpcloud 29 Thanks! spring.io/projects/spring-cloud-gcp cloud.google.com/java Come to the Google Cloud Platform Booth! @saturnism | saturnism.me