Contributing to the CodeFlare Operator

Here are a few things to go over before getting started with CodeFlare Operator development:

Environment setup

The following should be installed in your working environment:

Go 1.22.x
- Download release
- Install Instructions
Operator SDK
GCC

Basic Overview

The main entrypoint for the operator is main.go

Building and Deployment

If changes are made in the api dir, run: make manifests

This will generate new CRDs and associated files

If changes are made to any Go code (like in the controllers dir for example), run: make

This will check and build/compile the modified code

For building and pushing a new version of the operator image:

make image-build -e IMAGE_TAG_BASE=<image-repo/image-name> VERSION=<semver>
make image-push -e IMAGE_TAG_BASE=<image-repo/image-name> VERSION=<semver>

For deploying onto a cluster:

First, either set KUBECONFIG or ensure you are logged into a cluster in your environment
make install
make deploy -e IMG=<image-repo/image-name>

For building and pushing a new version of the bundled operator image:

make bundle-build -e IMAGE_TAG_BASE=<image-repo/image-name> VERSION=<new semver> PREVIOUS_VERSION=<semver to replace>
make bundle-push -e IMAGE_TAG_BASE=<image-repo/image-name> VERSION=<new semver> PREVIOUS_VERSION=<semver to replace>

To create a new openshift-community-operator-release:

make openshift-community-operator-release -e IMAGE_TAG_BASE=<image-repo/image-name> VERSION=<new semver> PREVIOUS_VERSION=<semver to replace> GH_TOKEN=<GitHub token for pushing bundle content to forked repository>

Testing

The CodeFlare Operator currently has unit tests and pre-commit checks

To enable and view pre-commit checks: pre-commit install
To run unit tests, run make test-unit
Note that both are required for CI to pass on pull requests

To write and inspect unit tests:

Unit test functions are defined in suite_test.go (with utils in util/util.go) in the controllers dir
Test cases defined under controllers/testdata

Local debugging with VSCode

Steps outlining how to run the operator locally.

Ensure you are authenticated to your Kubernetes/OpenShift Cluster.
Populate the .vscode/launch.json file with the location of your Kubernetes config file and desired namespace.
In VSCode on the activity bar click Run and Debug or CTRL + SHIFT + D to start a local debugging session of the CodeFlare Operator. The operator should be running as intended.

Example dev workflow

I've made changes to pkg/controllers/raycluster_controller.go and pkg/controllers/raycluster_controller.go. I've written unit tests and would like to test my changes using the unit tests as well as on an OpenShift cluster which I have access to.

Ensure the unit tests you've written are passing and you haven't introduced any regressions
1. run make test-unit
build and push image
1. make image-build -e IMG=<image-repo/image-name:image-tag>
2. make image-push -e IMG=<image-repo/image-name:image-tag>
Login to your OpenShift cluster via oc login --token=... --server=...
deploy ODH/RHOAI if necessary
1. for the latest releases
  1. run make delete-all-in-one -e USE_RHOAI=<true|false> if you would like to run on a fresh ODH/RHOAI deployment
  2. run make all-in-one -e USE_RHOAI=<true|false> to deploy new instance of ODH/RHOAI
2. otherwise follow the dev guides found in ODH and RHOAI (Red Hat internal only)
  1. run make install-nfd-operator, make install-service-mesh-operator, make install-ai-platform-operator, make install-nvidia-operator to ensure all dependent operators are also installed
navigate to the DataScienceCluster resource and set .spec.components.codeflare.managementState to Removed
run make deploy -e IMG=<image-repo/image-name:image-tag>

Your dev image should now be deployed in the cluster

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!