Skip to main content
Version: 2.5.0

Setting up a Kubernetes cluster

Introduction​

This guide is helpful in setting up a development environment intended to be used with odo; this setup is not recommended for a production environment.

odo can be used with ANY Kubernetes cluster. However, this development environment will ensure complete coverage of all features of odo.

Prerequisites​

  • You have a Kubernetes cluster set up (such as minikube)
  • You have admin privileges to the cluster

Important notes: odo will use the default ingress and storage provisioning on your cluster. If they have not been set correctly, see our troubleshooting guide for more details.

Summary​

  • An Ingress controller in order to use odo url create
  • Operator Lifecycle Manager in order to use odo service create
  • (Optional) Service Binding Operator in order to use odo link

Installing an Ingress controller​

Creating an Ingress controller is required to use the odo url create feature.

This can be enabled by installing an Ingress addon as per the Kubernetes documentation such as: the built-in one on minikube or NGINX Ingress.

IMPORTANT: odo cannot specify an Ingress controller and will use the default Ingress controller.

If you are unable to access your components, check that your default Ingress controller has been set correctly.

Minikube​

To install an Ingress controller on a minikube cluster, enable the ingress addon with the following command:

minikube addons enable ingress

NGINX Ingress​

To enable the Ingress feature on a Kubernetes cluster other than minikube, we reccomend to use the NGINX Ingress controller.

On the default installation method, you will need to set NGINX Ingress as your default Ingress controller, so odo may deploy URLs correctly.

Other Ingress controllers​

For a list of all available Ingress controllers see the the Ingress controller documentation.

To learn more about enabling this feature on your cluster, see the Ingress prerequisites on the official kubernetes documentation.

Installing the Operator Lifecycle Manager (OLM)​

Installing the Operator Lifecycle Manager (OLM) is required to use the odo service create feature.

The Operator Lifecycle Manager (OLM) is an open source toolkit to manage Kubernetes native applications, called Operators, in a streamlined and scalable way.

odo utilizes Operators in order to create and link services to applications.

The following command will install OLM cluster-wide as well as create two new namespaces: olm and operators.

curl -sL https://github.com/operator-framework/operator-lifecycle-manager/releases/download/v0.20.0/install.sh | bash -s v0.20.0

Running the script will take some time to install all the necessary resources in the Kubernetes cluster including the OperatorGroup resource.

Note: Check the OLM release page for the latest release.

Installing an Operator​

Installing an Operator allows you to install a service such as Postgres, Redis or DataDog.

To install an operator from the OperatorHub website:

  1. Visit the OperatorHub website.
  2. Search for an Operator of your choice.
  3. Navigate to its detail page.
  4. Click on Install.
  5. Follow the instruction in the installation popup. Please make sure to install the Operator in your desired namespace or cluster-wide, depending on your choice and the Operator capability.
  6. Verify the Operator installation.

Verifying the Operator installation​

Once the Operator is successfully installed on the cluster, you can use odo to verify the Operator installation and see the CRDs associated with it; run the following command:

odo catalog list services

The output may look similar to:

odo catalog list services
Services available through Operators
NAME                                CRDs
datadog-operator.v0.6.0             DatadogAgent, DatadogMetric, DatadogMonitor
service-binding-operator.v0.9.1     ServiceBinding, ServiceBinding

If you do not see your installed Operator in the list, follow the troubleshooting guide to find the issue and debug it.

Troubleshooting the Operator installation​

There are two ways to confirm that the Operator has been installed properly.

The examples you may see in this guide use Datadog Operator and Service Binding Operator.

  1. Verify that its pod started and is in “Running” state.

    kubectl get pods -n operators

    The output may look similar to:

    kubectl get pods -n operators
    NAME                                       READY   STATUS    RESTARTS   AGE
    datadog-operator-manager-5db67c7f4-hgb59   1/1     Running   0          2m13s
    service-binding-operator-c8d7587b8-lxztx   1/1     Running   5          6d23h

  2. Verify that the ClusterServiceVersion (csv) resource is in Succeeded or Installing phase.

    kubectl get csv -n operators

    The output may look similar to:

    kubectl get csv -n operators
    NAME                              DISPLAY                    VERSION   REPLACES                          PHASE
    datadog-operator.v0.6.0           Datadog Operator           0.6.0     datadog-operator.v0.5.0           Succeeded
    service-binding-operator.v0.9.1   Service Binding Operator   0.9.1     service-binding-operator.v0.9.0   Succeeded

    If you see the value under PHASE column to be anything other than Installing or Succeeded, please take a look at the pods in olm namespace and ensure that the pod starting with name operatorhubio-catalog is in Running state:

    kubectl get pods -n olm
    NAME                                READY   STATUS             RESTARTS   AGE
    operatorhubio-catalog-x24dq         0/1     CrashLoopBackOff   6          9m40s

    If you see output like above where the pod is in CrashLoopBackOff state or any other state other than Running, delete the pod:

    kubectl delete pods/<operatorhubio-catalog-name> -n olm

Checking to see if an Operator has been installed​

For this example, we will check the PostgreSQL Operator installation.

Check kubectl get csv to see if your Operator exists:

$ kubectl get csv                         
NAME                      DISPLAY                           VERSION   REPLACES                  PHASE
postgresoperator.v5.0.3   Crunchy Postgres for Kubernetes   5.0.3     postgresoperator.v5.0.2   Succeeded

If the PHASE is something other than Succeeded, you won't see it in odo catalog list services output, and you won't be able to create a working Operator backed service out of it either. You will have to wait patiently until PHASE says Suceeded.

(Optional) Installing the Service Binding Operator​

odo uses Service Binding Operator to provide the odo link feature which helps to connect an odo component to a service or another component.

The Service Binding Operator is optional and is used to provide extra metadata support for odo deployments.

Operators can be installed in a specific namespace or across the cluster-wide.

kubectl create -f https://operatorhub.io/install/service-binding-operator.yaml

Running the command will create the necessary resource in the operators namespace.

If you want to access this resource from other namespaces as well, add your target namespace to .spec.targetNamespaces list in the service-binding-operator.yaml file before running kubectl create.

See Verifying the Operator installation to ensure that the Operator was installed successfully.

Troubleshooting​

Confirming your Ingress Controller functionality​

odo will use the default Ingress Controller. By default, when you install an Ingress Controller such as NGINX Ingress, it will not be set as the default.

You must set it as the default Ingress Controller by modifying the annotation your IngressClass:

kubectl get IngressClass -A
kubectl edit IngressClass/YOUR-INGRESS -n YOUR-NAMESPACE

And add the following annotation:

annotation:
  ingressclass.kubernetes.io/is-default-class: "true"

Confirming your Storage Provisioning functionality​

odo deploys with Persistent Volume Claims. By default, when you install a StorageClass such as GlusterFS, it will not be set as the default.

You must set it as the default storage provisioner by modifying the annotation your StorageClass:

kubectl get StorageClass -A
kubectl edit StorageClass/YOUR-STORAGE-CLASS -n YOUR-NAMESPACE

And add the following annotation:

annotation:
  storageclass.kubernetes.io/is-default-class: "true"