This guide helps you get started developing KubeVela.
- Golang version 1.17+
- Kubernetes version v1.18+ with
~/.kube/config
configured. - ginkgo 1.14.0+ (just for E2E test)
- golangci-lint 1.38.0+, it will install automatically if you run
make
, you can install it manually if the installation is too slow. - kubebuilder v3.1.0+ and you need to manually install the dependency tools for unit test.
- CUE binary v0.3.0+
Install Kubebuilder manually
linux:
wget https://storage.googleapis.com/kubebuilder-tools/kubebuilder-tools-1.21.2-linux-amd64.tar.gz
tar -zxvf kubebuilder-tools-1.21.2-linux-amd64.tar.gz
mkdir -p /usr/local/kubebuilder/bin
sudo mv kubebuilder/bin/* /usr/local/kubebuilder/bin
macOS:
wget https://storage.googleapis.com/kubebuilder-tools/kubebuilder-tools-1.21.2-darwin-amd64.tar.gz
tar -zxvf kubebuilder-tools-1.21.2-darwin-amd64.tar.gz
mkdir -p /usr/local/kubebuilder/bin
sudo mv kubebuilder/bin/* /usr/local/kubebuilder/bin
For other OS or system architecture, please refer to https://storage.googleapis.com/kubebuilder-tools/
You may also be interested with KubeVela's design before diving into its code.
- Clone this project
git clone [email protected]:oam-dev/kubevela.git
KubeVela includes two parts, vela core
and vela cli
.
- The
vela core
is actually a K8s controller, it will watch OAM Spec CRD and deploy resources. - The
vela cli
is a command line tool that can build, run apps(with the help ofvela core
).
For local development, we probably need to build both of them.
- Build Vela CLI
make
After the vela cli built successfully, make
command will create vela
binary to bin/
under the project.
- Configure
vela
binary to System PATH
export PATH=$PATH:/your/path/to/project/kubevela/bin
Then you can use vela
command directly.
- Build Vela Core
make manager
- Run Vela Core
Firstly make sure your cluster has CRDs, below is the command that can help install all CRDs.
make core-install
To ensure you have created vela-system namespace and install definitions of necessary module. you can run the command:
make def-install
And then run locally:
make core-run
This command will run controller locally, it will use your local KubeConfig which means you need to have a k8s cluster locally. If you don't have a one, we suggest that you could setup up a cluster with kind.
When you're developing vela-core
, make sure the controller installed by helm chart is not running.
Otherwise, it will conflict with your local running controller.
You can check and uninstall it by using helm.
helm list -A
helm uninstall -n vela-system kubevela
You can try use your local built binaries follow the documentation.
make test
To execute the unit test of the API module, the mongodb service needs to exist locally.
make unit-test-apiserver
Before e2e test start, make sure you have vela-core running.
make core-run
Start to test.
make e2e-test
Contribute apiserver and velaux
Before start, please make sure you have already started the vela controller environment.
make run-apiserver
By default, the apiserver will serving at "0.0.0.0:8000".
Get the velaux code by:
git clone [email protected]:oam-dev/velaux.git
Configure the apiserver address:
cd velaux
echo "BASE_DOMAIN='http://127.0.0.1:8000'" > .env
Make sure you have installed yarn.
yarn install
yarn start
To execute the e2e test of the API module, the mongodb service needs to exist locally.
# save your config
mv ~/.kube/config ~/.kube/config.save
kind create cluster --image kindest/node:v1.18.15@sha256:5c1b980c4d0e0e8e7eb9f36f7df525d079a96169c8a8f20d8bd108c0d0889cc4 --name worker
kind get kubeconfig --name worker --internal > /tmp/worker.kubeconfig
kind get kubeconfig --name worker > /tmp/worker.client.kubeconfig
# restore your config
mv ~/.kube/config.save ~/.kube/config
make e2e-apiserver-test
- Read our code conventions
- Learn how to Create a pull request