DEVELOPING.md

Developing KubeBlocks

This document covers basic needs to work with KubeBlocks codebase.

• Developing KubeBlocks
  • Setup development environment
    • Bring your own toolbox
      • Install Go
      • Install Make
      • Build KubeBlocks
    • Use VSCode and development container
      • Setup the development container
      • Customize your dev container
        • Use a custom dev container image
        • Connect existing Kubernetes cluster
    • Setup a Kubernetes development environment
      • Docker environment
      • Kubernetes environment
  • Basics
    • Kubebuilder
    • Makefile
    • kbcli
    • Code style
  • Test
    • Envtest
    • Use existing Kubernetes cluster
    • Check test code coverage
  • Debug
    • Install Delve
    • Start a delve debug server
      • Debug specific package
      • Pass arguments
      • Debug envtest
      • Change debug server port
    • Connect the debug server with a frontend client
      • Delve CLI
      • JetBrains GoLand / IntelliJ IDEA with go plugin
      • VSCode

Setup development environment

There are two options for getting an environment up and running for KubeBlocks development.

Bring your own toolbox

To build KubeBlocks on your own host, needs to install the following tools:

• Go (Golang)
• Make

Install Go

Download and install Go 1.20 or later.

Install Make

KubeBlocks uses make for a variety of build and test actions, and needs to be installed as appropriate for your platform:

• Linux
  1. Install the build-essential package:

     sudo apt-get install build-essential

• macOS
  1. Ensure that build tools are installed:

     xcode-select --install

  2. When completed, you should see make and other command line developer tools in /usr/bin.

Build KubeBlocks

When go and make are installed, you can clone the KubeBlocks repository, and build KubeBlocks binaries with the make tool.

• To build for your current local environment:

  make all

• To cross-compile for a different platform, use the GOOS and GOARCH environmental variables:

  make all GOOS=windows GOARCH=amd64

Use VSCode and development container

If you are using Visual Studio Code, you can connect to a development container configured for KubeBlocks development. With development container, you don't need to manually install all of the tools and frameworks needed.

Setup the development container

1. Install VSCode Dev Containers extension
2. Open the KubeBlocks project folder in VSCode
   • VSCode will detect the presence of a dev container definition in the repo and will prompt you to reopen the project in a container:
   • Alternatively, you can open the command palette and use the Remote-Containers: Reopen in Container command:
   • VSCode will pull image and start dev container automatically, once the container is loaded, open an integrated terminal in VS Code and you're ready to develop KubeBlocks in a containerized environment.
3. And you can run make all to build KubeBlocks in the dev container.

Customize your dev container

Use a custom dev container image

The devcontainer.json uses the latest image from ApeCloud Docker hub, you can customize image to suit your need.

1. Edit the docker/Dockerfile-dev dev container image definition, you can change the [Option] configuration or install additional tools.

   # Copy library scripts to execute
   COPY library-scripts/*.sh library-scripts/*.env /tmp/library-scripts/
   
   # [Option] Install zsh
   ARG INSTALL_ZSH="true"
   
   # [Optional] Uncomment the next line to use go get to install anything else you need
   # RUN go get -x <your-dependency-or-tool>
   
   # [Optional] Uncomment this section to install additional OS packages.
   # RUN apt-get update && export DEBIAN_FRONTEND=noninteractive \
   #     && apt-get -y install --no-install-recommends <your-package-list-here>
   
   # [Optional] Uncomment this line to install global node packages.
   # RUN su vscode -c "source /usr/local/share/nvm/nvm.sh && npm install -g <your-package-here>" 2>&

2. Build a new image using updated Dockerfile, and replace the "image" property in devcontainer.json.

   {
       // Update to your dev container image
       "image": "docker.io/apecloud/kubeblocks-dev:latest",
   }
   

3. Rebuild and reopen the workspace in the dev container via the command palette and the Remote-Containers: Rebuild and Reopen in Container command.
4. When you are satisfied with your changes, you can optionally publish your container image to your own registry to speed up rebuilding the container when you only want to make changes to the devcontainer.json configuration in the future. For a docker image named docker.io/xxxx/kubeblocks-dev:

   make build-dev-image DEV_CONTAINER_IMAGE_NAME=docker.io/xxxx/kubeblocks-dev

Connect existing Kubernetes cluster

If you want to reuse an existing Kubernetes config, such as your EKS cluster or local Minikube cluster, you can configure the devcontainer.json copy those settings into the dev container. This requires:

1. Enabling the SYNC_LOCALHOST_KUBECONFIG environment variable
2. Bind mounting the locations of your Kubernetes and Minikube config paths to /home/kubeblocks/.kube-localhost and /home/kubeblocks/.minikube-localhost respectively.
   • You don't need to bind the Minikube path if you're not using it.

   "containerEnv": {
       // Uncomment to overwrite devcontainer .kube/config and .minikube certs with the localhost versions
       // each time the devcontainer starts, if the respective .kube-localhost/config and .minikube-localhost
       // folders respectively are bind mounted to the devcontainer.
       "SYNC_LOCALHOST_KUBECONFIG": "true",
   },
   
   ...
   
   "mounts": [
   // Uncomment to clone local .kube/config into devcontainer
   "type=bind,source=${env:HOME}${env:USERPROFILE}/.kube,target=/home/kubeblocks/.kube-localhost",
   
   // Uncomment to additionally clone minikube certs into devcontainer for use with .kube/config
   "type=bind,source=${env:HOME}${env:USERPROFILE}/.minikube,target=/home/kubeblocks/.minikube-localhost"
   
   ]

⚠ The SYNC_LOCALHOST_KUBECONFIG option only supports providing the dev container with the snapshot configuration from the host and does not support updating the host Kubernetes configuration from the dev container directly.

Setup a Kubernetes development environment

To run KubeBlocks, you needs Docker and a Kubernetes 1.24.1+ cluster for development.

Docker environment

1. Install Docker

   For Linux, you'll have to configure docker to run without sudo for the KubeBlocks build scripts to work. Follow the instructions to manage Docker as a non-root user.

2. Create your Docker Hub account if you don't already have one.

Kubernetes environment

• Kubernetes cluster You can use cloud Kubernetes service, such as EKS GKE AKS, or use local Kubernetes cluster, such as Minikube k3d.
• For development purposes, you will also want to follow the optional steps to install Helm 3.x.

Basics

Kubebuilder

KubeBlocks is using kubebuilder as the operator framework, before your start to code, suggest to read kubebuilder books.

Makefile

KubeBlocks includes a Makefile in the root of the repo. This serves as a high-level interface for common commands. Running make help will produce a list of make targets with descriptions. These targets will be referenced throughout this document.

kbcli

kbcli is a cobra-based command line interface for KubeBlocks which support to the basic interaction commands such as install/uninstall/upgrade KubeBlocks and all kinds of actions for KubeBlocks

The kbcli repository: https://github.com/apecloud/kbcli

Code style

First, read the Go (Golang) coding guidelines again, looking for any style violations. It's easier to remember a style rule once you've violated it.

Run our suite of linters:

make lint

This is not a fast command. On my machine, at the time of writing, it takes about a full minute to run. You can instead run

Test

KubeBlocks uses make for a variety of test actions.

Envtest

Setting up a local control plane and test all packages:

make test

Test specific packages:

# Directory `controllers` contains many packages, it will build and run each package individually.
make test TEST_PACKAGES=./controllers/...

# Test single package
make test-delve TEST_PACKAGES=./controllers/apps/...

Use existing Kubernetes cluster

make test-current-ctx

Instead of setting up a local control plane, this command will point to the control plane of an existing cluster by $HOME/.kube/config.

Check test code coverage

make test cover-report

This command will test all packages and generate a coverage report file cover.html, and open by your default browser automatically.

Debug

We can use Delve for Go debugging.

Install Delve

• Follow the General install instructions.

  Make sure PATH containers $GOPATH/bin which will allow you to run Go binary executables without specifying the absolute path.

• If you are using MacOS, you can install vid HomeBrew with the following command :

  brew install delve

• Check your installation

  dlv version

Start a delve debug server

You can start a delve debug server for running and debugging specific package or envtest.

Debug specific package

make run-delve GO_PACKAGE=./cmd/manager/main.go

Pass arguments

 make run-delve GO_PACKAGE=./cmd/cli/main.go ARGUMENTS='cluster create test-cluster --termination-policy=Halt'

Debug envtest

make test-delve TEST_PACKAGES=./controllers/apps/...

Unlike go test supports multiple packages, Delve needs a single executable to work, it only supports single package.

Change debug server port

You can change debug server port for make run-delve and make test-delve.

make run-delve DEBUG_PORT=2347

After bug server started, it will show the listen address:

API server listening at: [::]:2347

Connect the debug server with a frontend client

Delve support lots of code editors (editor plugins for Delve), you can choose your favorite editor to connect with the listen address for debugging.

This section introduces how to start debugging with the Delve CLI, Goland and VSCode. Please refer to the Delve or editors documentation for more details.

Delve CLI

$ dlv connect 127.0.0.1:2345
Type 'help' for list of commands.
(dlv) break cluster_controller.go:303
(dlv) continue
(dlv) next
(dlv) ......

JetBrains GoLand / IntelliJ IDEA with go plugin

Click Run — Edit configurations, add new debug configuration, select "Go Remote":

Add debug server listen address to Host and Port

And then set breakpoints, and run debug.

VSCode

Create/modify .vscode/launch.json file for connecting debug server:

{
    "version": "0.2.0",
    "configurations": [
        {
            "name": "debug envtest",
            "type": "go",
            "request": "attach",
            "mode": "remote",
            "remotePath": "",
            "port": 2346,
            "host": "127.0.0.1",
            "showLog": true,
            "trace": "log",
            "logOutput": "rpc"
        }
    ]
}

Set breakpoints and run Start Debugging.