Use an existing image with Supply Chain Choreographer

This topic describes how you can use an existing image with Supply Chain Choreographer.

For apps that build container images in a predefined way, the supply chains in the Out of the Box packages enable you to specify a prebuilt image. This uses the same stages as any other workload.

Requirements for prebuilt images

Supply chains aim at Knative as the runtime for the container image you provide. Your app must adhere to the following Knative standards:

  • Container port listens on port 8080

    The Knative service is created with the container port set to 8080 in the pod template spec Therefore, your container image must have a socket listening on 8080.

    ports:
      - containerPort: 8080
        name: user-port
        protocol: TCP
    
  • Non-privileged user ID

    By default, the container initiated as part of the pod is run as user 1000.

    securityContext:
      runAsUser: 1000
    
  • Arguments other than the image’s default ENTRYPOINT

    In most cases the container image runs using the ENTRYPOINT it was configured with. In the case of Dockerfiles, the combination of ENTRYPOINT and CMD.

    If you need extra configuration for your image, use --env flags with the tanzu apps workload create command or modify spec.env in your workload.yaml file.

  • Credentials for pulling the container image at runtime

    The image you provide is not relocated to an internal container image registry. Any components associated with the image must have the necessary credentials to pull it. For the service accounts used for the workload and deliverable, you must attach a secret that contains the credentials to pull the container image.

    If the image is hosted in a registry that has certificates signed by a private certificate authority, the components of the supply chains, delivery, and the Kubernetes nodes in the run cluster must trust the certificate.

Configure your workload to use a prebuilt image

To select a prebuilt image, set the spec.image field in your workload.yaml file with the name of the container image that contains the app to deploy by running:

tanzu apps workload create WORKLOAD-NAME \
  --app APP-NAME \
  --type TYPE \
  --image IMAGE

Where:

  • WORKLOAD-NAME is the name you choose for your workload.
  • APP-NAME is the name of your app.
  • TYPE is the type of your app.
  • IMAGE is the container image that contains the app you want to deploy.

For example, if you have an image named IMAGE, you can create a workload with the flag mentioned earlier:

tanzu apps workload create tanzu-java-web-app \
  --app tanzu-java-web-app \
  --type web \
  --image IMAGE

Expected output:

Create workload:
      1 + |---
      2 + |apiVersion: carto.run/v1alpha1
      3 + |kind: Workload
      4 + |metadata:
      5 + |  labels:
      6 + |    app.kubernetes.io/part-of: hello-world
      7 + |    apps.tanzu.vmware.com/workload-type: web
      8 + |  name: tanzu-java-web-app
      9 + |  namespace: default
     10 + |spec:
     11 + |  image: IMAGE

When you run tanzu apps workload create command with the --image field, the source resolution and build phases of the supply chain are skipped.

Examples

The following examples show ways that you can build container images for a Java-based app and complete the supply chains to a running service.

Using a Dockerfile

Using a Dockerfile is the most common way of building container images. You can select a base image, on top of which certain operations must occur, such as compiling code, and mutate the contents of the file system to a final container image that has a build of your app and any required runtime dependencies.

Here you use the maven base image for compiling your app code, and then the minimal distroless java17-debian11 image for providing a JRE that can run your app when it is built.

After building the image, you push it to a container image registry, and then reference it in the workload.

  1. Create a Dockerfile that describes how to build your app and make it available as a container image:

    ARG BUILDER_IMAGE=maven
    ARG RUNTIME_IMAGE=gcr.io/distroless/java17-debian11
    
    
    FROM $BUILDER_IMAGE AS build
    
            ADD . .
            RUN unset MAVEN_CONFIG && ./mvnw clean package -B -DskipTests
    
    
    FROM $RUNTIME_IMAGE AS runtime
    
            COPY --from=build /target/demo-0.0.1-SNAPSHOT.jar /demo.jar
            CMD [ "/demo.jar" ]
    
  2. Push the container image to a container image registry by running:

    docker build -t IMAGE .
    docker push IMAGE
    
  3. Create a workload by running:

    tanzu apps workload create tanzu-java-web-app \
      --type web \
      --app tanzu-java-web-app \
      --image IMAGE
    

    Expected output:

    Create workload:
          1 + |---
          2 + |apiVersion: carto.run/v1alpha1
          3 + |kind: Workload
          4 + |metadata:
          5 + |  labels:
          6 + |    app.kubernetes.io/part-of: hello-world
          7 + |    apps.tanzu.vmware.com/workload-type: web
          8 + |  name: tanzu-java-web-app
          9 + |  namespace: default
         10 + |spec:
         11 + |  image: IMAGE
    
  4. Run the following workload:

    tanzu apps workload get tanzu-java-web-app
    

    Expected output:

    # tanzu-java-web-app: Ready
    ---
    lastTransitionTime: "2022-04-06T19:32:46Z"
    message: ""
    reason: Ready
    status: "True"
    type: Ready
    
    Workload pods
    NAME                                                   STATUS      RESTARTS   AGE
    tanzu-java-web-app-00001-deployment-7d7df5ccf5-k58rt   Running     0          32s
    tanzu-java-web-app-config-writer-xjmvw-pod             Succeeded   0          89s
    
    Workload Knative Services
    NAME                 READY   URL
    tanzu-java-web-app   Ready   http://tanzu-java-web-app.default.example.com
    

Using Spring Boot’s build-image Maven target

You can use Spring Boot’s build-image target to build a container image that runs your app. The build-image target must use a Dockerfile.

For example, using the same sample repository as mentioned before (https://github.com/vmware-tanzu/application-accelerator-samples/tree/main/tanzu-java-web-app):

  1. Build the image by running the following command from the root of the repository:

    IMAGE=ghcr.io/kontinue/hello-world:tanzu-java-web-app
    ./mvnw spring-boot:build-image -Dspring-boot.build-image.imageName=$IMAGE
    

    Expected output:

    [INFO] Scanning for projects...
    [INFO]
    [INFO] --------------------------< com.example:demo >--------------------------
    [INFO] Building demo 0.0.1-SNAPSHOT
    [INFO] --------------------------------[ jar ]---------------------------------
    [INFO]
    ...
    [INFO]
    [INFO] Successfully built image 'ghcr.io/kontinue/hello-world:tanzu-java-web-app'
    [INFO]
    [INFO] ------------------------------------------------------------------------
    [INFO] BUILD SUCCESS
    [INFO] ------------------------------------------------------------------------
    [INFO] Total time:  39.257 s
    [INFO] Finished at: 2022-04-06T19:40:16Z
    [INFO] ------------------------------------------------------------------------
    
  2. Push the image you built to the container image registry by running:

    IMAGE=ghcr.io/kontinue/hello-world:tanzu-java-web-app
    docker push $IMAGE
    

    Expected output:

    The push refers to repository [ghcr.io/kontinue/hello-world]
    1dc94a70dbaa: Preparing
    ...
    9d6787a516e7: Pushed
    tanzu-java-web-app: digest: sha256:7140722ea396af69fb3d0ad12e9b4419bc3e67d9c5d8a2f6a1421decc4828ace size: 4497
    

After you push the container image, you see the same results as building the image using a Dockerfile.

For more information about building container images for a Spring Boot app, see Spring Boot with Docker

About Out of the Box Supply Chains

In Tanzu Application Platform, the ootb-supply-chain-basic, ootb-supply-chain-testing, and ootb-supply-chain-testing-scanning packages each receive a new supply chain that provides a prebuilt container image for your app.

ootb-supply-chain-basic

    (cluster)  basic-image-to-url   ClusterSupplyChain            (!) new
    ^          source-to-url        ClusterSupplyChain


ootb-supply-chain-testing

    (cluster)  testing-image-to-url  ClusterSupplyChain           (!) new
    ^          source-test-to-url    ClusterSupplyChain


ootb-supply-chain-testing-scanning

    (cluster)  scanning-image-scan-to-url    ClusterSupplyChain   (!) new
    ^          source-test-scan-to-url       ClusterSupplyChain

To leverage the supply chains that expect a prebuilt image, you must set the spec.image field in the workload to the name of the container image that contains the app to deploy.

The new supply chains use a Cartographer feature that lets VMware increase the specificity of supply chain selection by using the matchFields selector rule.

The selection takes place as follows:

  • ootb-supply-chain-basic

    • From source: label apps.tanzu.vmware.com/workload-type: web
    • Prebuilt image: label apps.tanzu.vmware.com/workload-type: web and set spec.image in the workload.yaml
  • ootb-supply-chain-testing

    • From source: labels apps.tanzu.vmware.com/workload-type: web and apps.tanzu.vmware.com/has-tests: true
    • Prebuilt image: label apps.tanzu.vmware.com/workload-type: web and set spec.image in the workload.yaml
  • ootb-supply-chain-testing-scanning

    • From source: labels apps.tanzu.vmware.com/workload-type: web and apps.tanzu.vmware.com/has-tests: true
    • Prebuilt image: label apps.tanzu.vmware.com/workload-type: web and set spec.image in the workload.yaml

Workloads that already work with the supply chains before Tanzu Application Platform v1.1 continue to work with the same supply chain. Workloads that bring a prebuilt container image must set spec.image in the workload.yaml.

Understanding the supply chain for a prebuilt image

An ImageRepository object is created to keep track of new images pushed under that name. ImageRepository makes the image available to further resources in the supply chain, providing the final digest of the latest image.

Whenever a new image is pushed to the workload’s image location, the ImageRepository detects the change. The image is then available to further resources by updating its imagerepository.status.artifact.revision with an absolute reference to that image.

For example, if you create a workload using an image named hello-world, tagged tanzu-java-web-app hosted under ghcr.io in the kontinue repository:

tanzu apps workload create tanzu-java-web-app \
  --app tanzu-java-web-app \
  --type web \
  --image ghcr.io/kontinue/hello-world:tanzu-java-web-app

After a couple seconds, you see the ImageRepository object created to keep track of images named ghcr.io/kontinue/hello-world:tanzu-java-web-app:

Workload/tanzu-java-web-app
├─ImageRepository/tanzu-java-web-app
├─PodIntent/tanzu-java-web-app
├─ConfigMap/tanzu-java-web-app
└─Runnable/tanzu-java-web-app-config-writer
  └─TaskRun/tanzu-java-web-app-config-writer-p2lzv
    └─Pod/tanzu-java-web-app-config-writer-p2lzv-pod

If you inspect the status in status.resources in the workload.yaml, you see the image-provider resource promoting the image it found to further resources:

apiVersion: carto.run/v1alpha1
kind: Workload
spec:
  image: ghcr.io/kontinue/hello-world:tanzu-java-web-app
status:
  resources:
    - name: image-provider
      outputs:
        # output being made available to further resources in the supply chain
        # (in this case, the latest image it found under that name).
        #
        - name: image
          lastTransitionTime: "2022-04-01T15:05:01Z"
          preview: ghcr.io/kontinue/hello-world:tanzu-java-web-app@sha256:9fb930a...

      # reference to the object managed by the supply chain for this
      # resource
      #
      stampedRef:
        apiVersion: source.apps.tanzu.vmware.com/v1alpha1
        kind: ImageRepository
        name: tanzu-java-web-app
        namespace: workload

      # reference to the template that defined how this object should look
      # like
      #
      templateRef:
        apiVersion: carto.run/v1alpha1
        kind: ClusterImageTemplate
        name: image-provider-template

The image found by the ImageRepository object is carried through the supply chain to the final configuration. This is pushed to either a Git repository or image registry so that it is deployed in a run cluster.

Note

The image name matches the image name supplied in the spec.image field in the workload.yaml, but also includes the digest of the latest image found under the tag. If a new image is pushed to the same tag, you see the ImageRepository resolving the name to a different digest corresponding to the new image pushed.

check-circle-line exclamation-circle-line close-line
Scroll to top icon