Migrate Kubernetes Applications with Konveyer Crane

Introduction

Occasionally when you’re moving between major version of Kubernetes or Red Hat OpenShift, you’ll want to migrate your applications between clusters. Or if you’re moving between two clouds, you’ll want an easy way to migrate your workloads from one platform to another.

The Crane operator from the open source Konveyer project automates this migration process for you. The Konveyer site offers a selection of helpful projects to administer your cluster. Crane is designed to automate migration from one cluster to another and is surprisingly easy to get working.

This article shows you how we moved a default sample application from a Red Hat OpenShift on AWS (ROSA) to a Red Hat OpenShift on IBM Cloud (ROIC) cluster. To see how it’s done, watch the video or read the steps below.

Install the Crane operator

First, log into the cluster console where your original application is hosted and also log into the console of the destination where you want to migrate your application. In our example, we logged into the OpenShift Service on AWS as our origin console and, in another tab, logged into to the Red Hat OpenShift on IBM Cloud console as our destination console.

From the Operator Hub in both consoles, search for “Crane Operator” and follow the default prompts to install the operator.

Set up your sample application

From your origin cluster, choose the Developer profile and then click +Add to add a project where you will deploy your application into.

Choose a sample app to play with. In our example, we chose the Python application. Name it and then click “Create”. It will pull the source information from GitHub and build an image, deploy the image, and expose it as a PHP endpoint.

You can change back from a Developer profile to the Admin profile in order to see if the operator has been installed correctly.

Create a migration (MIG) controller

Now it’s time to create your migration controller.

1. Go to your m migration cluster (in this example, our IBM console), select the Crane operator, and select Create migration controller. Do the same on the origin cluster (in our example, AWS).

2. Switch to the openshift-migration namespace.

3. Update the host MIG controller.

4. Then you can apply your migration cluster.

• NOTE: Run only this following command on the remote cluster

1. Save your service account secret for the destination cluster.

2. Write this into sa-secret-remote.yaml on your origin cluster:

3. Add your destination cluster:

4. Configure s3 credentials to host migration storage. Included here is the correct access key for my files. You’ll need to have that handy.

5. Configure MIG storage to use s3

6. Create the migration plan. THe plan is essentially saying: “This is what I want”. The plan says what namespace you want to move. In this example, it’s project-a as referenced below.

7. Execute your migration plan.

Watch the magic

Back in your original console, you can find the correct namespace and see that things have moved.

1. Go back to the origin OpenShift console (AWS in our example). Bring up the migration GUI from the openshift-migration namespace. Check through the migration plans which show you the migration or watch the logs to see the migration happening.

2. Or, watch the progress via the CLI if you prefer.

You can also open your destination console (Red Hat OpenShift on IBM Cloud in our example) and see if the new cluster has migrated.

Conclusion

Hopefully walking through these steps has helped you understand the power that Crane can offer you when migrating workloads between clusters. This was only a sample application. With a little work and testing, you should be able to leverage Crane for your applications. If you have any questions or thoughts, come around to #konveyer on the Kubernetes public Slack channel, and the team would me more then willing to help advise you.

Happy migrating your apps!