Camel K in Tekton Pipelines

Introduction

Available from Camel K 1.0.0-M2

Camel K can be directly used in Tekton Pipelines tasks since it’s container image ships the kamel CLI tool, that can be used to create all needed resources by interacting with the Kubernetes cluster.

For example, the following Tekton pipeline task can be used to run an integration:

apiVersion: tekton.dev/v1alpha1
kind: Task
metadata:
  name: camel-k-run-integration
spec:
  inputs:
    resources:
      - name: repository
        type: git
    params:
      - name: file
        description: The integration file to run
  steps:
    - name: run-integration
      image: apache/camel-k:1.0.0-M2 (1)
      workingDir: /workspace/repository
      command:
        - kamel
      args:
        - "run"
        - "--wait"
        - "$(inputs.params.file)" (2)

1	The base image for the step is `apache/camel-k:1.0.0-M2`
2	It executes command `kamel run --wait ${inputs.params.file}`, with file to run received as parameter

When executed, such task creates an integration resource from the given file (which is supposed to be contained in the input git repository) and waits for the integration to be fully running before completing.

This task is a building block for more complex scenarios that can be composed in Tekton pipelines. If you want to learn more, just follow the remainder of the tutorial.

Prerequisites

This tutorial assumes the following requirements are met:

OpenShift (or OKD) 4+ cluster (works also on "vanilla" Kubernetes with some adjustment) and oc binary tool
Tekton Pipelines 0.5.2 (different versions may need adjustments)
Camel K Client Tools 1.0.0-M2 (kamel binary tool)

Cluster Setup

This tutorial assumes that Tekton Pipelines are already installed in the cluster. Refer to the Tekton documentation to learn how to install them.

Camel K cluster resources need to be installed on the cluster:

# Use the oc tool to login as cluster admin to the target cluster, then
kamel install --cluster-setup

This will install the Camel K CRD (custom resource definitions) and roles to access them. You can switch to a standard user after doing this operation.

Creating the Pipeline

We’re going to create a pipeline on a new namespace. The first step is to create the namespace:

oc new-project camel-pipelines

Setting up a ServiceAccount

Any running container that needs to interact with the Kubernetes API (in our case, to create deployment and integration resources) must have special permissions granted to its service account.

Download the camel-k-pipeline-permissions.yaml file and save it into a directory in your hard drive. Then install it using the oc client tool:

oc apply -f camel-k-pipeline-permissions.yaml

This creates a Kubernetes ServiceAccount named camel-k-pipeline that is authorized to create the same kind of resources that the Camel K operator needs to create. We’ll use that service account to run our pipeline.

Creating the Pipeline Definition

Let’s now create the pipeline definition. Download the camel-k-pipeline-task-definition.yaml file and apply it to the cluster:

oc apply -f camel-k-pipeline-task-definition.yaml

This creates a series of resources, including a pipeline definition that you can immediately see on the OpenShift developer console.

View of the Camel K Tekton pipeline in the OpenShift developer console

The pipeline just created is composed of two tasks.

The last task (camel-k-run-integration) is the one we’ve already described in the introduction, while the first task just installs the operator in the current namespace. The definition of the first task is shown in the following excerpt:

apiVersion: tekton.dev/v1alpha1
kind: Task
metadata:
  name: camel-k-install-operator
spec:
  steps:
    - name: install
      image: apache/camel-k:1.0.0-M2
      command:
        - kamel
      args:
        - "install"
        - "--skip-cluster-setup"

As you see, it’s just doing a kamel install --skip-cluster-setup (we’ve already done the cluster setup in the preparation phase).

Both tasks are executed in sequence in the pipeline that is shown below.

apiVersion: tekton.dev/v1alpha1
kind: Pipeline
metadata:
  name: camel-k-pipeline
spec:
  resources:
    - name: source-repo
      type: git
  tasks:
    - name: install-operator
      taskRef:
        name: camel-k-install-operator
    - name: run-integration
      runAfter: [install-operator]
      taskRef:
        name: camel-k-run-integration
      resources:
        inputs:
          - name: repository
            resource: source-repo
      params:
        - name: file
          value: "examples/tekton/hello.groovy"

The integration file to be executed is set to examples/tekton/hello.groovy that is a simple "Hello World" integration contained in the Camel K github repository example/tekton directory.

In order to specify the actual source repository, the definition file contains also the following resource:

apiVersion: tekton.dev/v1alpha1
kind: PipelineResource
metadata:
  name: camel-k-examples-git
spec:
  type: git
  params:
    - name: revision
      value: main
    - name: url
      value: https://github.com/apache/camel-k

The repo will be now used as input for the run-integration task in the pipeline.

Triggering a Pipeline Execution

Everything is now ready to be executed and the last thing missing is a trigger. Download the camel-k-pipeline-task-run.yaml file and apply it to the cluster:

oc apply -f camel-k-pipeline-task-run.yaml

The file contains only a PipelineRun resource:

apiVersion: tekton.dev/v1alpha1
kind: PipelineRun
metadata:
  name: camel-k-pipeline-run-1
spec:
  pipelineRef:
    name: camel-k-pipeline
  serviceAccount: 'camel-k-pipeline' (1)
  resources:
    - name: source-repo
      resourceRef:
        name: camel-k-examples-git (2)

1	The PipelineRun binds the pipeline to the service account previously created
2	It also binds the pipeline to the Camel K repository containing the examples

The creation of the file starts the execution of the pipeline and its progress can be monitored on the OpenShift developer console.

View of the Camel K Tekton pipeline execution in the OpenShift developer console

To execute the PipelineRun again, delete previous runs before re-applying

The result of the pipeline execution is the Camel K operator and an integration named hello running on the cluster:

Result of the Camel K Tekton pipeline execution

There are certainly other ways to trigger an execution of a pipeline, like as reaction to a change in the git repository, but this is left to you as exercise ;)

Refer to the Tekton repository for more information.