The cluster was already running, but the tests could fail at any moment. You needed certainty, not hope. Integration testing with kubectl gives you that certainty. It runs against the same resources, the same APIs, and the same network that production will use. When your tests pass here, they pass for real.
Integration testing in Kubernetes is not just a build step—it’s the gatekeeper for production. kubectl connects directly to the Kubernetes API server. From there, you can deploy test pods, apply manifests, seed data, and execute workloads in a true cluster environment. Every test interacts with actual services, ConfigMaps, Secrets, and persistent volumes, making false positives rare.
Start by defining a dedicated test namespace. Use kubectl create namespace integration-test to isolate workloads and avoid clashing with staging or production resources. Apply Kubernetes manifests for your services and dependencies using kubectl apply -f service.yaml. Then, run your test suite inside the cluster. kubectl exec lets you trigger tests from inside the pod, ensuring network calls, environment variables, and service discovery work as expected.
You can integrate kubectl commands into CI pipelines. A job can spin up the namespace, deploy containers, run tests, and tear everything down with kubectl delete namespace integration-test. This makes integration testing repeatable, automated, and clean.