jenvtest makes it easy to implement integration tests with Kubernetes API Server in Java. Inspired by envtest in Kubebuilder (Golang).
It runs the API Server binaries directly (without nodes and other components, but with etcd). Linux, Windows, Mac is supported.
See also this blog post regarding the motivation and more.
See more documentation in docs directory.
Include dependency:
<dependency>
<groupId>io.javaoperatorsdk</groupId>
<artifactId>jenvtest</artifactId>
<version>[version]</version>
<scope>test</scope>
</dependency>
See sample unit test here
@EnableKubeAPIServer
class JUnitExtensionSimpleCaseTest {
// Use @KubeConfig annotation to inject kube config yaml to init any client
@KubeConfig
static String kubeConfigYaml;
@Test
void simpleTestWithTargetVersion() {
var client = new KubernetesClientBuilder()
.withConfig(Config.fromKubeconfig(kubeConfigYaml))
.build();
client.resource(TestUtils.testConfigMap()).create();
var cm = client.resource(TestUtils.testConfigMap()).get();
Assertions.assertThat(cm).isNotNull();
}
}
The underlying API can be used directly. See KubeApiServer
See it's usage in a test.
class KubeApiServerTest {
@Test
void apiServerTest() {
var kubeApi = new KubeAPIServer();
kubeApi.start();
var client = new KubernetesClientBuilder()
.withConfig(Config.fromKubeconfig(kubeApi.getKubeConfigYaml()))
.build();
client.resource(TestUtils.testConfigMap()).create();
var cm = client.resource(TestUtils.testConfigMap()).get();
Assertions.assertThat(cm).isNotNull();
kubeApi.stop();
}
}
There is dedicated support for Fabric8 Kubernetes Client.
Using dependency:
<dependency>
<groupId>io.javaoperatorsdk</groupId>
<artifactId>jenvtest-fabric8-client-support</artifactId>
<version>[version]</version>
<scope>test</scope>
</dependency>
The client can be directly injected to the test. See sample test here.
@EnableKubeAPIServer
class JUnitFabric8ClientInjectionTest {
static KubernetesClient client;
// emitted code
}
Parallel test execution is explicitly supported for JUnit5, in fact the project tests are running parallel. Running a new instance for each test case. This speeds up the tests (in our case >75%) in a way that test cases are also fully isolated from each other. See the surefire plugin config.
An additional benefits os running K8S API Server this way, is that it makes easy to test Conversion Hooks and/or Dynamic Admission Controllers
In general, it is a best practice to use additional standard frameworks to implement Kubernetes webhooks, like kubernetes-webooks-framework with Quarkus or Spring. However, we demonstrate how it works in this test