Search code examples
javakubernetesfabric8tektontekton-pipelines

There is unknown field "container" in Tekton

I'm interested in Tekton these days.

However there are some issue when I implement Task with java fabric8.tekton apis.

There exist api which is adding steps in spec in units of container(withContainer) in TaskBuilder class.

However I got error message in rune time like below, enter image description here

Can I get some advices?

Tekton version - v0.10.1

I used packages like below:

io.fabric8:kubernetes-client:4.7.1

io.fabric8:tekton-client:4.7.1

Here is my complete test code.

package com.example.tekton;

import java.util.ArrayList;
import java.util.List;
import io.fabric8.kubernetes.api.model.Container;
import io.fabric8.kubernetes.api.model.ContainerBuilder;
import io.fabric8.kubernetes.client.BaseClient;
import io.fabric8.kubernetes.client.Config;
import io.fabric8.kubernetes.client.ConfigBuilder;
import io.fabric8.tekton.client.TektonClient;
import io.fabric8.tekton.client.DefaultTektonClient;
import io.fabric8.tekton.client.handlers.TaskHandler;
import io.fabric8.tekton.client.handlers.TaskRunHandler;
import io.fabric8.tekton.pipeline.v1alpha1.ArrayOrString;
import io.fabric8.tekton.pipeline.v1alpha1.Task;
import io.fabric8.tekton.pipeline.v1alpha1.TaskBuilder;
import io.fabric8.tekton.pipeline.v1alpha1.TaskRun;
import io.fabric8.tekton.pipeline.v1alpha1.TaskRunBuilder;
import io.fabric8.tekton.pipeline.v1alpha1.TaskRefBuilder;

public class DefaultKubernetesTest {

public Task getTask() {
  Container con = new ContainerBuilder()
                        .withNewImage("ubuntu")
                        .withNewName("echo-hello-world")
                        .addNewCommand("echo")
                        .addNewArg("hello jinwon world")
                        .build();

    Task task = new TaskBuilder()
        .withApiVersion("tekton.dev/v1alpha1")
        .withKind("Task")
        .withNewMetadata()
            .withName("echo-hello-world-test")
        .endMetadata()
        .withNewSpec()
            .addNewStep()
                .withContainer(con)
            .endStep()
        .endSpec()
        .build();

        return task;
}

public TaskRun getTaskRun() {
    TaskRun taskRun = new TaskRunBuilder()
    .withNewMetadata()
        .withName("taskrun")
        .endMetadata()
    .withNewSpec()
        .withTaskRef(new TaskRefBuilder().withName("echo-hello-world-test").withApiVersion("tekton.dev/v1alpha1").withKind("Task").build())
    .endSpec().build();
    return taskRun;
}

public static void main(String[] args) {
    ConfigBuilder config = new ConfigBuilder();
    DefaultKubernetesTest kubeTest = new DefaultKubernetesTest();
    String username = "testUser";
    String password = "testPwd";

    config = config.withMasterUrl("https://192.168.6.236:6443");
    config = config.withUsername(username);
    config = config.withPassword(password);
    Config kubeConfig = config.build();

    try (DefaultTektonClient test = new DefaultTektonClient(kubeConfig)) {
        Task task = kubeTest.getTask();
        TaskRun taskRun = kubeTest.getTaskRun();
        test.tasks().inNamespace("test").create(task);
        test.taskRuns().inNamespace("test").create(taskRun);
        test.close();
    }
}
}
Solution

  • Tekton ships with an admission controller, which validates the CRD specs before allowing them into the cluster. Because the project is still in alpha, its moving quite fast. Fabric8 may be templating out K8s objects against a different spec from what has been installed on your cluster. You should be able to validate the spec version used in Fabric8 and remove all the Tekton objects in your cluster and re-apply them at a specific version.