Deployment Tasks with Keptn

A KeptnTaskDefinition resource defines one or more "executables" (functions, programs, scripts, etc.) that Keptn runs as part of the pre- and post-deployment phases of a KeptnApp or KeptnWorkload.

• pre-deployment (before the pod is scheduled)
• post-deployment (after the pod is scheduled)

These KeptnTask resources and the KeptnEvaluation resources (discussed in Evaluations) are part of the Keptn Release Lifecycle Management.

A KeptnTask executes as a runner in an application container, which runs as part of a Kubernetes job. A KeptnTaskDefinition includes calls to executables to be run.

To implement a KeptnTask:

• Define a KeptnTaskDefinition resource that defines the runner to use for the container and the executables to be run pre- and post-deployment
• Apply basic-annotations to your workloads.
• Generate the required KeptnApp resources following the instructions in Auto app discovery.
• Annotate your workload YAML files to associate your KeptnTaskDefinition with the pre-/post-deployment tasks that should be run.
• Create the appropriate KeptnAppContext resource to associate KeptnTaskDefinition resources to the generated KeptnApp.

This page provides information to help you create your tasks:

• Code your task in an appropriate runner
• How to control the execution order of functions, programs, and scripts since all KeptnTask resources at the same level run in parallel
• Understand how to use Context that contains a Kubernetes cluster, a user, a namespace, the application name, workload name, and version.
• Use parameterized functions if your task requires input parameters
• Create secret text and pass secrets to a function if necessary.

Runners and containers

Each KeptnTaskDefinition can use exactly one container with one runner. The runner you use determines the language you can use to define the task. The spec section of the KeptnTaskDefinition defines the runner to use for the container:

Keptn provides a general Kubernetes container runtime that you can configure to do almost anything you want:

• The container-runtime runner provides a pure custom Kubernetes application container that you define to includes a runtime, an application and its runtime dependencies. This gives you the greatest flexibility to define tasks using the language and facilities of your choice

Keptn also includes two "pre-defined" runners:

• Use the deno-runtime runner to define tasks using Deno scripts, which use JavaScript/Typescript syntax with a few limitations. You can use this to specify simple actions without having to define a container.
• Use the python-runtime runner to define your task using Python 3.

For the pre-defined runners (deno-runtime and python-runtime), the actual code to be executed can be configured in one of four different ways:

• inline
• referring to an HTTP script
• referring to another KeptnTaskDefinition
• referring to a ConfigMap resource that is populated with the function to execute

See the KeptnTaskDefinition reference page for the synopsis and examples for each runner.

Run a task associated with your workload deployment

To define pre-/post-deployment tasks, you must manually edit the YAML files to add annotations for your tasks to the appropriate workload YAML file.

Specify one of the following annotations/labels for each task you want to execute:

keptn.sh/pre-deployment-tasks: <task-name>
keptn.sh/post-deployment-tasks: <task-name>

The value of each annotation corresponds to the value of the name field of the KeptnTaskDefinition resource.

Run a task associated with your entire KeptnApp

To execute pre-/post-deployment tasks for a KeptnApp, create a KeptnAppContext with the same name and in the same namespace as the KeptnApp. The KeptnAppContext resource contains a list of pre-/post-deployment tasks that should be executed before and after the workloads within the KeptnApp are deployed.

See the Getting started guide for more information on how to configure a KeptnAppContext resource to execute pre-/post-deployment checks. KeptnAppContext is also used to collect user defined metadata information to use during your Task execution. As explained later in this guide. (See the context section)

Example of pre/post-deployment actions

A comprehensive example of pre-/post-deployment evaluations and tasks can be found in our examples folder, where we use Podtato-Head to run some simple pre-deployment checks. Check out the readme to learn how to test this example on your machine.

Executing sequential tasks

All KeptnTask resources that are defined by KeptnTaskDefinition resources at the same level (either pre-deployment or post-deployment) execute in parallel. This is by design, because Keptn is not a pipeline engine. Task sequences that are not part of the lifecycle workflow should not be handled by Keptn but should instead be handled by the pipeline engine tools being used such as Jenkins, Argo Workflows, Flux, and Tekton.

If your lifecycle workflow includes a sequence of executables that need to be run in order, you can put them all in one KeptnTaskDefinition resource, which can execute a virtually unlimited number of programs, scripts, and functions, as long as they are all using the same runner. You have the following options:

• Encode all your steps in the language of your choice and build an image that Keptn executes in a container-runtime runner. This is often the best solution if you need to execute complex sequences because it gives you the most flexibility.

• Use the inline syntax for one of the Keptn pre-defined runners (either deno-runtime or python-runtime) to code the actual calls inline in the KeptnTaskDefinition resource. See Fields for pre-defined containers for more information.

• Create a script that calls the functions, programs, and scripts that need to execute sequentially and install this on a remote webserver that Keptn can access. Then use the httpRef syntax for one of the pre-defined runners to call this script from your KeptnTaskDefinition, which can set parameters for the script if appropriate.

For more details about implementing these options, see the KeptnTaskDefinition page.

Context

The Keptn task context includes details about the current deployment, application name, version, object type and other user-defined metadata. Keptn populates this metadata while running tasks before and after deployments, to provide the necessary context associated with each task.

This contrasts with the Kubernetes context, which is a set of access parameters that defines the specific cluster, user and namespace with which you interact. For more information, see Configure Access to Multiple Clusters.

For Tasks generated for applications running in Kubernetes, Keptn populates a KEPTN_CONTEXT environment variable containing a set of parameters that correlate a task to a specific application/workload, information about the phase in which the task is being executed, as well as any metadata that has been attached to the related application/workload

You can use this context information in the function code in your KeptnTaskDefinition resource.

KEPTN_CONTEXT is encoded as JSON and by default, contains:

• "appName"
• "appVersion"
• "workloadName"
• "workloadVersion"
• "taskType"
• "objectType"
• "traceparent"
• "metadata"

A Job created by a KeptnTask with KEPTN_CONTEXT, may look like the following

apiVersion: batch/v1
kind: Job
spec:
  template:
    spec:
      containers:
        - name: "my-task-container"
          env:
            - name: KEPTN_CONTEXT
              value: '{
                         "workloadName":"waiter-waiter",
                         "appName":"waiter",
                         "appVersion":"",
                         "workloadVersion":"0.4",
                         "taskType":"pre",
                         "objectType":"Workload",
                         "metadata":{
                            "commit-id":"1234",
                            "stage":"dev",
                            "test-metadata":"test-metadata",
                            "traceparent": "00-traceid-spanid-01"
                         }
                      }'
            - name: SCRIPT
              value: /var/data/function.ts

You can customize the metadata field to hold any key-value pair of interest to share among your workloads and tasks in a KeptnApp (for instance a commit ID value). To do so, the metadata needs to be specified for the workload or for the application. Follow our guide on Context and Metadata here.

Note

For an example of how to access the KEPTN_CONTEXT, follow our reference page examples for deno and for python.

Parameterized functions

KeptnTaskDefinitions can use input parameters. Simple parameters are passed as a single map of key values, while the secret parameters refer to a single Kubernetes secret.

Consider the following example:

apiVersion: lifecycle.keptn.sh/v1
kind: KeptnTaskDefinition
metadata:
  name: slack-notification-dev
spec:
  deno:
    functionRef:
      name: slack-notification
    parameters:
      map:
        textMessage: "This is my configuration"
    secureParameters:
      secret: slack-token

Note the following about using parameters with functions:

• Keptn passes the values defined inside the map field as a JSON object.
• Multi-level maps are not currently supported.
• The JSON object can be read through the environment variable DATA using Deno.env.get("DATA");.
• Currently only one secret can be passed. The secret must have a key called SECURE_DATA. It can be accessed via the environment variable Deno.env.get("SECURE_DATA").

Working with secrets

A special case of parameterized functions is to pass secrets that may be required to access data that your task requires.

Create secret text

To create a secret to use in a KeptnTaskDefinition, execute this command:

kubectl create secret generic my-secret --from-literal=SECURE_DATA=foo

apiVersion: lifecycle.keptn.sh/v1
kind: KeptnTaskDefinition
metadata:
  name: dummy-task
  namespace: "default"
spec:
  deno:
    secureParameters:
      secret: my-secret
    inline:
      code: |
        let secret_text = Deno.env.get("SECURE_DATA");
        // secret_text = "foo"

To pass multiple variables you can create a Kubernetes secret using a JSON string:

kubectl create secret generic my-secret \
--from-literal=SECURE_DATA="{\"foo\": \"bar\", \"foo2\": \"bar2\"}"

apiVersion: lifecycle.keptn.sh/v1
kind: KeptnTaskDefinition
metadata:
  name: dummy-task
  namespace: "default"
spec:
  deno:
    secureParameters:
      secret: my-secret
    inline:
      code: |
        let secret_text = Deno.env.get("SECURE_DATA");
        let secret_text_obj = JSON.parse(secret_text);
        // secret_text_obj["foo"] = "bar"
        // secret_text_obj["foo2"] = "bar2"

Pass secrets to a function

Kubernetes secrets can be passed to the function using the secureParameters field.

Here, the secret value is the name of the Kubernetes secret, which contains a field with the key SECURE_DATA.
The value of that field is then available to the function's runtime via an environment variable called SECURE_DATA.

For example, if you have a task function that should make use of secret data, you must first ensure that the secret containing the SECURE_DATA key exists For example:

apiVersion: v1
kind: Secret
metadata:
  name: deno-demo-secret
  namespace: default
type: Opaque
data:
  SECURE_DATA: YmFyCg== # base64 encoded string, e.g. 'bar'

Then, you can make use of that secret as follows:

apiVersion: lifecycle.keptn.sh/v1
kind: KeptnTaskDefinition
metadata:
  name: deployment-hello
  namespace: "default"
spec:
  deno:
    secureParameters:
      secret: deno-demo-secret
    inline:
      code: |
        console.log("Deployment Hello Task has been executed");

        let foo = Deno.env.get('SECURE_DATA');
        console.log(foo);
        Deno.exit(0);

Comments