To install and configure keptn in a Kubernetes cluster, follow these instructions:
The next steps expect that you have a working Kubernetes cluster in Google Container Engine (GKE). See the Getting Started Guides for details about creating a cluster with the folllowing configuration:
Node pool with 2 nodes:
image type: Ubuntu (preferred) or Container-Optimized OS (cos)
Note 1: To select Ubuntu, click on the “Advanced Edit” button in GKE and select “Ubuntu” as the Image Type in the Nodes section.
Note 2: If Container-Optimized OS (cos) is selected, make sure to follow the instructions for setting up the Dynatrace OneAgent Operator. This means that after the initial setup with
setupInfrastructure.sh, which is a step below, the
cr.yml has to be edited and applied again. In addition, all pods have to be restarted.
The scripts provided by keptn v.0.1 run in a BASH and require following tools locally installed:
kubectl that is logged in to your cluster.
Tip: View all the kubectl commands, including their options and descriptions in the kubectl CLI reference.
Additionally, the scripts need:
GitHub organizationto store the repositories of the sockshop application
GitHub personal access tokento push changes to the sockshop repositories
API Token must have the following permissions additional as shown in the screenshot below:
Go to the keptn release page to download the installation file using, e.g.,
$ cd ~ $ wget https://github.com/keptn/keptn/archive/0.1.0.tar.gz
Extract the package and move to the keptn directory:
$ tar -xvzf 0.1.0.tar.gz $ cd keptn-0.1.0
Keptn contains all scripts and instructions needed to deploy the demo application Sockshop on a Kubernetes cluster.
forkGitHubRepositories.sh script in the
scripts directory. This script takes the name of the GitHub organization you have created earlier. This script clones all needed repositories and uses
hub to fork those repositories to the passed GitHub organization. Afterwards, the script deletes all repositories and clones them again from the GitHub organization.
# from your keptn repo, navigate into your scripts folder $ cd scripts/ $ ./forkGitHubRepositories.sh <GitHubOrg>
Insert information in ./scripts/creds.json by executing
defineCredentials.sh in the
scripts directory. This script will prompt you for all information needed to complete the setup and populate the file scripts/creds.json with them.
Attention: This file will hold your personal access-token and credentials needed for the automatic setup of keptn. Take care of not leaking this file! (As a first precaution we have added this file to the
.gitignore file to prevent committing it to GitHub.)
setupInfrastructure.sh in the
scripts directory. This script deploys a container registry and Jenkins service within your cluster, as well as an initial deployment of the sockshop application in the dev, staging, and production namespaces. Please note that this is an initial step to provide you with running services in all three environments. In the course of the different use cases provided by keptn, the microservices will be built and deployed by the Jenkins pipelines set up by keptn.
Attention: The script will create several new resources for you and will also update the files shipped with keptn. Take care of not leaking any files that will hold personal information. Including:
Note: The script will run for some time (~10-15 min), since it will wait for Jenkins to boot before setting credentials via the Jenkins REST API.
To verify the deployment of the sockshop service, retrieve the URLs of your front-end in the dev, staging, and production environments with the
kubectl get svc
$ kubectl get svc front-end -n dev NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE front-end LoadBalancer 10.23.252.*** **.225.203.*** 8080:30438/TCP 5m
$ kubectl get svc front-end -n staging NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE front-end LoadBalancer 10.23.246.*** **.184.97.*** 8080:32501/TCP 6m
$ kubectl get svc front-end -n production NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE front-end LoadBalancer 10.23.248.*** **.226.62.*** 8080:32232/TCP 7m
kubectl get svc command to get the EXTERNAL-IP and PORT of Jenkins. Then user a browser to open Jenkins and login using the default Jenkins credentials:
Note: It is highly recommended to change these credentials right after the first login.
$ kubectl get svc jenkins -n cicd NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE jenkins LoadBalancer 10.23.245.*** ***.198.26.*** 24***:32478/TCP,50***:31867/TCP 10m
To verify the correct installation of Jenkins, go to the Jenkins dashboard where you see the following pipelines:
Finally, navigate to Jenkins > Manage Jenkins > Configure System and scroll to the environment variables to verify whether the variables are set correctly. Note: The value for the parameter DT_TENANT_URL must start with https://
Important: Due to a known issue in Jenkins, it is necessary to click Save although nothing is changed in this verification step. If the configuration is not saved at this point, the
$BUILD_URL variable won’t be initialized correctly for the builds and this meta-information will be missing later on in the use cases.
In case any value is missing in Jenkins, this can be fixed by adding the corresponding value in the
manifests/jenkins/k8s-jenkins-deployment.yml file and re-applying this file with
E.g., if the the value for
DOCKER_REGISTRY_IP is unset, retrieve the value with
kubectl get svc -n cicd and insert as value.
``` … env:
Save the file and apply with:
kubectl apply -f k8s-jenkins-deployment.yml. This will redeploy the Jenkins with the updated configuration.
Create a naming rule for process groups
Screenshot shows this rule definition.
To clean up your Kubernetes cluster, execute the
cleanupCluster.sh script in the
Please note that in case of any errors, the install script might leave some files in a inconsistent state, therefore the
setupInfrastructure.sh file can not be run a second time without a cleanup. To prevent any issues with subsequent setup runs, we recommend to fully delete the GitHub organization, the keptn installation folder and checkout the keptn release again. (Some files may have been edited already that are not reverted in case of aborting the setup script.)