tekton-kaniko-build

Demo: Jib-based Tekton Pipeline

This demo shows how to use Tekton Pipelines to build java code from source code in a su-directory of git repository. In this demo I will build a sample Spring app using Google Cloud Vision API and deploy it to a Knative cluster.

Assumes you already installed Tekton Pipelines and its dependencies kubectl apply -f https://storage.googleapis.com/tekton-releases/latest/release.yaml

Configure Dependance Cache

Maven will be downloading a lot of dependencies, to ensure the subsequent build run faster, first define a PersistentVolumeClaim on your cluster

kubectl apply -f cache.yaml

Install the Tekton Task on Knative cluster (one time task):

The Tekton Task for Maven builds Java/Kotlin/Groovy/Scala source into a container image using Google's Jib tool.

kubectl apply -f task-jdk-8.yaml

Install Secret (one time task):

First you need to create a GCP service account with sufficient rights to push images to GCR. Then edit the provided secret.yaml file and paste the secret into the password field.

Tekton will use this account to authenticate to the registry

kubectl apply -f secret.yaml

Deploy Build:

With Tokton configured on your Knative cluster, you can now build Java apps.

Make sure to edit the PipelineResource component to change cloudylabs to your own project name

apiVersion: tekton.dev/v1alpha1
kind: PipelineResource
metadata:
  name: vision-image-build
spec:
  type: image
  params:
    - name: url
      value: gcr.io/cloudylabs/vision:3.2.10

To submit your build now, run:

kubectl apply -f maven-build.yaml

Wait for the build to complete:

First time you build, Maven will have to download all the dependencies (~3 min). After that, things get a lot faster. Subsequent builds as fast as ~15 sec.

kubectl get pods

If you are interested in watching the on-cluster build, you can watch the build-step-build-and-push logs

kubectl logs jib-maven-vision-image-build-pod-**** -c step-build-and-push -f

Replace pod name in the above command with the unique name on your cluster with the name listed by kubectl get pods

The resulting output should look like something like this

[INFO] Built and pushed image as gcr.io/cloudylabs/vision:3.2.10
[INFO]
[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 33.250 s
[INFO] Finished at: 2019-09-14T13:34:29Z
[INFO] Final Memory: 58M/537M
[INFO] ------------------------------------------------------------------------

Deploy Vision app

With the image built, you can now deploy

Make sure to replace cloudylabs with the name of your project

container:
    image: gcr.io/cloudylabs/vision:3.2.10

Now apply the updated manifest

kubectl apply -f service.yaml

View

Navigate to the vision URL (https://vision.demo.knative.tech/) to see the results.

Cleanup

To remove the sample app from your cluster, delete the service record:

kubectl delete -f service.yaml