Deploying the Operator for Testing

There are several ways to run the Operator in your cluster. Probably the easiest approach is to use a Deployment resource. Another approach is to use the Operator Lifecycle Manager (OLM).

Both methods require that you first build and publish the container image for the Operator.

Log in to Your Cluster as a Cluster Administrator

Both methods use the kubectl command to interact with your Kubernetes cluster from your workstation. Therefore, you need to configure the ~/.kube/config file to grant you a cluster administrator access to the cluster. With an OpenShift cluster, you can use the oc login command.

Run the Operator by Using a Deployment Resource

Set the IMG environment variable to the name of the image that you built and pushed to the container registry. The Operator manager container uses that image.

IMG=quay.io/<some-organization>/quay-api-operator:<tag>
export IMG
make deploy

The make deploy command installs the Custom Resource Definitions (CRDs), creates the quay-api-operator-system Kubernetes namespace, and then creates the Deployment resource that deploys the manager pod.

You can verify that the pod is running by using the kubectl get pods command:

kubectl get pods -n quay-api-operator-system

If the pod fails to start, then review its logs by running the kubectl logs <podname> -n quay-api-operator-system command. These logs are also useful later on, for accessing the Ansible messages, and troubleshooting Quay resources.

For more details, review the Run as a Deployment Inside the Cluster section of the Run the Operator guide.

Undeploy the Operator

To remove the Operator from the cluster, use the undeploy target of the Makefile:

IMG=quay.io/<some-organization>/quay-api-operator:<tag>
export IMG
make undeploy

The command deletes the manager Deployment resource, the CRDs, and the quay-api-operator-system namespace.

Run the Operator by Using OLM

OLM consumes the Operator as a bundle container image that you must prepare.

Bundle the Operator

You create the bundle manifests, and then build and push the image by using the make targets.

In the Makefile, the BUNDLE_IMG variable specifies the name of the bundle image, which is set to quay.io/herve4m/quay-api-operator-bundle:v$(VERSION) by default. Because you do not have write access to that image repository, you must redefine the variable as an environment variable before building the bundle image. Select a repository for which you have write access.

BUNDLE_IMG=quay.io/<some-organization>/quay-api-operator-bundle:<tag>
export BUNDLE_IMG
# Remove a previous local build of the bundle image, if it exists
podman rmi ${BUNDLE_IMG}
# Log in to Quay with your credentials
podman login --username <yourname> quay.io
# Build and then push the image
make bundle bundle-build bundle-push

Install OLM

If you use an OpenShift cluster, the OLM is already installed. Otherwise, use the operator-sdk command to install it.

operator-sdk olm install

Deploy the Operator

Create the quay-api-operator namespace to host the Operator:

kubectl create namespace quay-api-operator

Use the operator-sdk run bundle command to install the Operator in your cluster:

operator-sdk run bundle -n quay-api-operator ${BUNDLE_IMG}

The command create the CatalogSource, OperatorGroup, Subscription, and ClusterServiceVersion resources, and then wait for the Operator to complete its deployment.

Undeploy the Operator

Undeploy the Operator and delete the OLM resources:

kubectl get Subscription -n quay-api-operator | grep quay-api-operator
kubectl delete Subscription -n quay-api-operator quay-api-operator-<version>

kubectl get ClusterServiceVersion | grep quay-api-operator
kubectl delete ClusterServiceVersion quay-api-operator.<version>

kubectl delete OperatorGroup -n quay-api-operator operator-sdk-og
kubectl delete CatalogSource -n quay-api-operator quay-api-operator-catalog

kubectl delete namespace quay-api-operator

For more details, review the OLM Integration Bundle Tutorial and the Deploy Your Operator with OLM section of the Run the Operator guide.