deploy

Deploying a Development Version Manually

For most tasks, the e2e test shell wrapper should be used to run tests. It leverages the deploy script described below which can also be invoked manually.

The best way to test changes to csi-digitalocean is on a DigitalOcean Kubernetes (DOKS) cluster. However, since DOKS clusters run the DigitalOcean CSI controller as a managed component that cannot be modified by the user, deploying with the example manifests will not work.

To test a development version on a DOKS cluster, do the following:

1. Create a DOKS cluster on the latest version:

   $ doctl k8s cluster create csi-integration-test
   Notice: cluster is provisioning, waiting for cluster to be running
   ..

   Wait for it to finish creating; your kubeconfig will automatically be updated.

2. Build and push a dev version of csi-digitalocean by running VERSION=dev make publish from the root of the repository.

3. Run deploy.sh from this directory, providing a DO API access token for your account (This requires kustomize and kubectl):

   $ DIGITALOCEAN_ACCESS_TOKEN=<token> ./deploy.sh
   Deploying a dev version of the CSI driver to context do-nyc1-csi-integration-test.
   Continue? (yes/no)
   yes

   (You can also pass -y or --yes as a parameter to deploy.sh to skip the prompt.)

Alternative Image Locations

The instructions above assume you have push access to the digitalocean/do-csi-plugin repository on Docker Hub. However, this is not necessary to build and test a dev version of the CSI driver.

To build and publish a test version in your own image repository, do the following from the root of the repository:

DOCKER_REPO=<my-image-repository> VERSION=dev make publish

You can then follow the instructions above, setting the DEV_IMAGE environment variable to your own image location when invoking deploy.sh.