Kf

This is not an officially supported Google product.

Getting started the manual way

Follow the install instructions at https://cloud.google.com/migrate/kf/docs/ to create a GKE cluster, install Kf into it, and deploy an app with the kf CLI.

Deploy a local Kf install to a new cluster

If you need to set up a new development cluster run the following command:

./hack/deploy-dev-release.sh

It will fetch all your local sources and kick off a Cloud Build that builds a version of Kf, creates a GKE cluster and installs the Kf version onto it.

Iterative development

Building the CLI:

$ ./hack/build.sh

Installing Kf server-side components:

With the Operator

Kf is installed through the Operator. Operator code is in the operator folder. To apply current change to the server, run hack/apply-kf-with-operator.sh. This script will build kf with ko, copy the built yaml file into the operator data folder(operator/cmd/manager/kodata/kf/), then build and apply the operator.

Without the Operator

Kf can be installed independently. To do so, you will need to disable the Operator first. One way to do so is to run:

kubectl patch kfsystem kfsystem --type='json' -p="[{'op':'replace','path':'/spec/kf/enabled','value':false}]"

We use ko for rapid development and during the release process to build a full set of kf images and installation YAML. Run the following to stage local changes on a targeted cluster:

$ ./hack/ko-apply.sh

This will build any images required by config/, upload them to the provided registry, and apply the resulting configuration to the current cluster.

Verify the installation of Kf components:

Kf CLI

Kf CLI can be downlowed from official releases or build locally.

To build Kf CLI locally, run hack/build.sh. A executable kf will be generated under bin.

*Kf server side component

Kf has a built-in self diagnostic tool called Kf doctor. Run kf doctor to run through the diagnotics to make sure the Kf server side component and dependencies are properly installed.

Run tests:

All tests can be run using the script hack/test.sh. Integration tests can be skipped by setting the environment variable SKIP_INTEGRATION to true.

Optionally, unit tests can be run separately with script hack/unit-test.sh.

Integration tests can be run with script hack/integration-test.sh. Integrationt tests requires a Kubernetes cluster with Kf server component installed.