Learn how to manage your services in keptn.

Onboard your own service

After creating a project, which represents a repository in your GitHub organization, the keptn CLI allows to onboard services into this project. Please note that for describing the Kubernetes resources, Helm charts are used. Therefore, the keptn CLI allows setting a Helm values description in the before created project. Optionally, the user can also provide a Helm deployment and service description.

When onboarding your own service instead of the provided carts service, a values file has to be provided. In addition, a service and a deployment file can be provided. The following snippet will help to define your own values.yaml file:

replicaCount: 1
    repository: null
    tag: null
    pullPolicy: IfNotPresent
    name: myservice 
    internalPort: 8080
    name: myservice

First, define how many instances of your deployment should be running by providing this number as the replicaCount. Next, the image repository and tag can be set to null since they will be set with the keptn CLI command keptn send event new-artifact. For the service, provide the name of your service as well as the internal port you want your service to be reachable. For the container name provide a name you want to call your container. Additionally, make sure that your actual service provides a /health endpoint at port 8080 since this is needed for the liveness and readiness probe of Kubernetes.

Furthermore, keptn needs to have access to the perfspec.json file as well as the JMeter files. Therefore, fork the GitHub repo of your service into the GitHub organization that you have created earlier. Make sure there are the needed files in the corresponding folders in your repository:

│  ...    
└── jmeter
│   │   basiccheck.jmx
│   │   SERVICENAME_load.jmx
│   │   SERVICENAME_perfcheck.jmx
└── perfspec
│   │   perfspec.json
└── src
└── ...

An example of a perfspec file is listed below:

  "spec_version": "1.0",
  "indicators": [
  "objectives": {
    "pass": 90,
    "warning": 75

In this file, Prometheus is gathering metrics on the request latency in seconds. A threshold grader is then used to evaluate if the threshold is met. A score to this single metric is assigned, as well as objectives that have to be met to consider the quality of the service satisfying.

To onboard a service, use the command onboard service and provide the project name (flag --project), the Helm chart values (flag --values) and optionally also deployment (flag --deployment) and service (flag --service) configuration.

  • Use default deployment and service configuration:
keptn onboard service --project=your_project --values=values.yaml
  • Use custom deployment and service configuration:
keptn onboard service --project=your_project --values=values.yaml --deployment=deployment.yaml --service=service.yaml

Note: If you are using custom configurations and you would like to have the environment variables KEPTN_PROJECT, KEPTN_STAGE, and KEPTN_SERVICE within your service, add the following environment variables to your deployment configuration.

    value: "{{ .Chart.Name }}"
    value: "{{ }}"
  - name: KEPTN_STAGE
        fieldPath: "metadata.namespace"