github.com/argoproj/argo-cd/v2@v2.10.9/docs/developer-guide/toolchain-guide.md (about) 1 # Development toolchain 2 3 ## Preface 4 5 !!!note "Before you start" 6 The Argo CD project continuously grows, both in terms of features and community size. It gets adopted by more and more organisations which entrust Argo CD to handle their critical production workloads. Thus, we need to take great care with any changes that affect compatibility, performance, scalability, stability and security of Argo CD. For this reason, every new feature or larger enhancement must be properly designed and discussed before it gets accepted into the code base. 7 8 We do welcome and encourage everyone to participate in the Argo CD project, but please understand that we can't accept each and every contribution from the community, for various reasons. If you want to submit code for a great new feature or enhancement, we kindly ask you to take a look at the 9 [code contribution guide](code-contributions.md#) before you start to write code or submit a PR. 10 11 We want to make contributing to Argo CD as simple and smooth as possible. 12 13 This guide shall help you in setting up your build & test environment, so that you can start developing and testing bug fixes and feature enhancements without having to make too much effort in setting up a local toolchain. 14 15 If you want to submit a PR, please read this document carefully, as it contains important information guiding you through our PR quality gates. 16 17 As is the case with the development process, this document is under constant change. If you notice any error, or if you think this document is out-of-date, or if you think it is missing something: Feel free to submit a PR or submit a bug to our GitHub issue tracker. 18 19 If you need guidance with submitting a PR, or have any other questions regarding development of Argo CD, do not hesitate to [join our Slack](https://argoproj.github.io/community/join-slack) and get in touch with us in the `#argo-contributors` channel! 20 21 ## Before you start 22 23 You will need at least the following things in your toolchain in order to develop and test Argo CD locally: 24 25 * A Kubernetes cluster. You won't need a fully blown multi-master, multi-node cluster, but you will need something like K3S, Minikube or microk8s. You will also need a working Kubernetes client (`kubectl`) configuration in your development environment. The configuration must reside in `~/.kube/config` and the API server URL must point to the IP address of your local machine (or VM), and **not** to `localhost` or `127.0.0.1` if you are using the virtualized development toolchain (see below) 26 27 * You will also need a working Docker runtime environment, to be able to build and run images. The Docker version must be 17.05.0 or higher, to support multi-stage builds. 28 29 * Obviously, you will need a `git` client for pulling source code and pushing back your changes. 30 31 * Last but not least, you will need a Go SDK and related tools (such as GNU `make`) installed and working on your development environment. The minimum required Go version for building and testing Argo CD is **v1.17**. 32 33 * We will assume that your Go workspace is at `~/go`. 34 35 !!! note 36 **Attention minikube users**: By default, minikube will create Kubernetes client configuration that uses authentication data from files. This is incompatible with the virtualized toolchain. So if you intend to use the virtualized toolchain, you have to embed this authentication data into the client configuration. To do so, start minikube using `minikube start --embed-certs`. Please also note that minikube using the Docker driver is currently not supported with the virtualized toolchain, because the Docker driver exposes the API server on 127.0.0.1 hard-coded. If in doubt, run `make verify-kube-connect` to find out. 37 38 ## Submitting PRs 39 40 ### Continuous Integration process 41 42 When you submit a PR against Argo CD's GitHub repository, a couple of CI checks will be run automatically to ensure your changes will build fine and meet certain quality standards. Your contribution needs to pass those checks in order to be merged into the repository. 43 44 !!!note 45 46 Please make sure that you always create PRs from a branch that is up-to-date with the latest changes from Argo CD's master branch. Depending on how long it takes for the maintainers to review and merge your PR, it might be necessary to pull in latest changes into your branch again. 47 48 Please understand that we, as an Open Source project, have limited capacities for reviewing and merging PRs to Argo CD. We will do our best to review your PR and give you feedback as soon as possible, but please bear with us if it takes a little longer as expected. 49 50 The following read will help you to submit a PR that meets the standards of our CI tests: 51 52 ### Title of the PR 53 54 Please use a meaningful and concise title for your PR. This will help us to pick PRs for review quickly, and the PR title will also end up in the Changelog. 55 56 We use [PR title checker](https://github.com/marketplace/actions/pr-title-checker) to categorize your PR into one of the following categories: 57 58 * `fix` - Your PR contains one or more code bug fixes 59 * `feat` - Your PR contains a new feature 60 * `docs` - Your PR improves the documentation 61 * `chore` - Your PR improves any internals of Argo CD, such as the build process, unit tests, etc 62 63 Please prefix the title of your PR with one of the valid categories. For example, if you chose the title your PR `Add documentation for GitHub SSO integration`, please use `docs: Add documentation for GitHub SSO integration` instead. 64 65 ### PR template checklist 66 67 Upon opening a PR, the details will contain a checklist from a template. Please read the checklist, and tick those marks that apply to you. 68 69 ### Automated builds & tests 70 71 After you have submitted your PR, and whenever you push new commits to that branch, GitHub will run a number of Continuous Integration checks against your code. It will execute the following actions, and each of them has to pass: 72 73 * Build the Go code (`make build`) 74 * Generate API glue code and manifests (`make codegen`) 75 * Run a Go linter on the code (`make lint`) 76 * Run the unit tests (`make test`) 77 * Run the End-to-End tests (`make test-e2e`) 78 * Build and lint the UI code (`make lint-ui`) 79 * Build the `argocd` CLI (`make cli`) 80 81 If any of these tests in the CI pipeline fail, it means that some of your contribution is considered faulty (or a test might be flaky, see below). 82 83 ### Code test coverage 84 85 We use [CodeCov](https://codecov.io) in our CI pipeline to check for test coverage, and once you submit your PR, it will run and report on the coverage difference as a comment within your PR. If the difference is too high in the negative, i.e. your submission introduced a significant drop in code coverage, the CI check will fail. 86 87 Whenever you develop a new feature or submit a bug fix, please also write appropriate unit tests for it. If you write a completely new module, please aim for at least 80% of coverage. 88 If you want to see how much coverage just a specific module (i.e. your new one) has, you can set the `TEST_MODULE` to the (fully qualified) name of that module with `make test`, i.e.: 89 90 ```bash 91 make test TEST_MODULE=github.com/argoproj/argo-cd/server/cache 92 ... 93 ok github.com/argoproj/argo-cd/server/cache 0.029s coverage: 89.3% of statements 94 ``` 95 96 ## Local vs Virtualized toolchain 97 98 Argo CD provides a fully virtualized development and testing toolchain using Docker images. It is recommended to use those images, as they provide the same runtime environment as the final product and it is much easier to keep up-to-date with changes to the toolchain and dependencies. But as using Docker comes with a slight performance penalty, you might want to setup a local toolchain. 99 100 Most relevant targets for the build & test cycles in the `Makefile` provide two variants, one of them suffixed with `-local`. For example, `make test` will run unit tests in the Docker container, `make test-local` will run it natively on your local system. 101 102 If you are going to use the virtualized toolchain, please bear in mind the following things: 103 104 * Your Kubernetes API server must listen on the interface of your local machine or VM, and not on `127.0.0.1` only. 105 * Your Kubernetes client configuration (`~/.kube/config`) must not use an API URL that points to `localhost` or `127.0.0.1`. 106 107 You can test whether the virtualized toolchain has access to your Kubernetes cluster by running `make verify-kube-connect` (*after* you have setup your development environment, as described below), which will run `kubectl version` inside the Docker container used for running all tests. 108 109 The Docker container for the virtualized toolchain will use the following local mounts from your workstation, and possibly modify its contents: 110 111 * `~/go/src` - Your Go workspace's source directory (modifications expected) 112 * `~/.cache/go-build` - Your Go build cache (modifications expected) 113 * `~/.kube` - Your Kubernetes client configuration (no modifications) 114 * `/tmp` - Your system's temp directory (modifications expected) 115 116 ## Setting up your development environment 117 118 The following steps are required no matter whether you chose to use a virtualized or a local toolchain. 119 120 !!!note "Docker privileges" 121 If you opt in to use the virtualized toolchain, you will need to have the 122 appropriate privileges to interact with the Docker daemon. It is not 123 recommended to work as the root user, and if your user does not have the 124 permissions to talk to the Docker user, but you have `sudo` setup on your 125 system, you can set the environment variable `SUDO` to `sudo` in order to 126 have the build scripts make any calls to the `docker` CLI using sudo, 127 without affecting the other parts of the build scripts (which should be 128 executed with your normal user privileges). 129 130 You can either set this before calling `make`, like so for example: 131 132 ``` 133 SUDO=sudo make sometarget 134 ``` 135 136 Or you can opt to export this permanently to your environment, for example 137 ``` 138 export SUDO=sudo 139 ``` 140 141 ### Clone the Argo CD repository from your personal fork on GitHub 142 143 * `mkdir -p ~/go/src/github.com/argoproj` 144 * `cd ~/go/src/github.com/argoproj` 145 * `git clone https://github.com/yourghuser/argo-cd` 146 * `cd argo-cd` 147 148 ### Optional: Setup an additional Git remote 149 150 While everyone has their own Git workflow, the author of this document recommends to create a remote called `upstream` in your local copy pointing to the original Argo CD repository. This way, you can easily keep your local branches up-to-date by merging in latest changes from the Argo CD repository, i.e. by doing a `git pull upstream master` in your locally checked out branch. To create the remote, run `git remote add upstream https://github.com/argoproj/argo-cd` 151 152 ### Install the must-have requirements 153 154 Make sure you fulfill the pre-requisites above and run some preliminary tests. Neither of them should report an error. 155 156 * Run `kubectl version` 157 * Run `docker version` 158 * Run `go version` 159 160 ### Build the required Docker image 161 162 Build the required Docker image by running `make test-tools-image`. This image offers the environment of the virtualized toolchain. 163 164 The `Dockerfile` used to build these images can be found at `test/container/Dockerfile`. 165 166 ### Test connection from build container to your K8s cluster 167 168 Run `make verify-kube-connect`, it should execute without error. 169 170 If you receive an error similar to the following: 171 172 ``` 173 The connection to the server 127.0.0.1:6443 was refused - did you specify the right host or port? 174 make: *** [Makefile:386: verify-kube-connect] Error 1 175 ``` 176 177 you should edit your `~/.kube/config` and modify the `server` option to point to your correct K8s API (as described above). 178 179 ### Using k3d 180 181 [k3d](https://github.com/rancher/k3d) is a lightweight wrapper to run [k3s](https://github.com/rancher/k3s), a minimal Kubernetes distribution, in docker. Because it's running in a docker container, you're dealing with docker's internal networking rules when using k3d. A typical Kubernetes cluster running on your local machine is part of the same network that you're on, so you can access it using **kubectl**. However, a Kubernetes cluster running within a docker container (in this case, the one launched by make) cannot access 0.0.0.0 from inside the container itself, when 0.0.0.0 is a network resource outside the container itself (and/or the container's network). This is the cost of a fully self-contained, disposable Kubernetes cluster. The following steps should help with a successful `make verify-kube-connect` execution. 182 183 1. Find your host IP by executing `ifconfig` on Mac/Linux and `ipconfig` on Windows. For most users, the following command works to find the IP address. 184 185 * For Mac: 186 187 ``` 188 IP=`ifconfig en0 | grep inet | grep -v inet6 | awk '{print $2}'` 189 echo $IP 190 ``` 191 192 * For Linux: 193 194 ``` 195 IP=`ifconfig eth0 | grep inet | grep -v inet6 | awk '{print $2}'` 196 echo $IP 197 ``` 198 199 Keep in mind that this IP is dynamically assigned by the router so if your router restarts for any reason, your IP might change. 200 201 2. Edit your ~/.kube/config and replace 0.0.0.0 with the above IP address. 202 203 3. Execute a `kubectl version` to make sure you can still connect to the Kubernetes API server via this new IP. Run `make verify-kube-connect` and check if it works. 204 205 4. Finally, so that you don't have to keep updating your kube-config whenever you spin up a new k3d cluster, add `--api-port $IP:6550` to your **k3d cluster create** command, where $IP is the value from step 1. An example command is provided here: 206 207 ``` 208 k3d cluster create my-cluster --wait --k3s-arg '--disable=traefik@server:*' --api-port $IP:6550 -p 443:443@loadbalancer 209 ``` 210 211 !!!note 212 For k3d versions less than v5.0.0, the example command flags `--k3s-arg` and `'--disable=traefik@server:*'` should change to `--k3s-server-arg` and `'--disable=traefik'`, respectively. 213 214 ## The development cycle 215 216 When you have developed and possibly manually tested the code you want to contribute, you should ensure that everything will build correctly. Commit your changes to the local copy of your Git branch and perform the following steps: 217 218 ### Pull in all build dependencies 219 220 As build dependencies change over time, you have to synchronize your development environment with the current specification. In order to pull in all required dependencies, issue: 221 222 * `make dep-ui` 223 224 Argo CD recently migrated to Go modules. Usually, dependencies will be downloaded on build time, but the Makefile provides two targets to download and vendor all dependencies: 225 226 * `make mod-download` will download all required Go modules and 227 * `make mod-vendor` will vendor those dependencies into the Argo CD source tree 228 229 ### Generate API glue code and other assets 230 231 Argo CD relies on Google's [Protocol Buffers](https://developers.google.com/protocol-buffers) for its API, and this makes heavy use of auto-generated glue code and stubs. Whenever you touched parts of the API code, you must re-generate the auto generated code. 232 233 * Run `make codegen`, this might take a while 234 * Check if something has changed by running `git status` or `git diff` 235 * Commit any possible changes to your local Git branch, an appropriate commit message would be `Changes from codegen`, for example. 236 237 !!!note 238 There are a few non-obvious assets that are auto-generated. You should not change the autogenerated assets, as they will be overwritten by a subsequent run of `make codegen`. Instead, change their source files. Prominent examples of non-obvious auto-generated code are `swagger.json` or the installation manifest YAMLs. 239 240 ### Build your code and run unit tests 241 242 After the code glue has been generated, your code should build and the unit tests should run without any errors. Execute the following statements: 243 244 * `make build` 245 * `make test` 246 247 These steps are non-modifying, so there's no need to check for changes afterwards. 248 249 ### Lint your code base 250 251 In order to keep a consistent code style in our source tree, your code must be well-formed in accordance to some widely accepted rules, which are applied by a Linter. 252 253 The Linter might make some automatic changes to your code, such as indentation fixes. Some other errors reported by the Linter have to be fixed manually. 254 255 * Run `make lint` and observe any errors reported by the Linter 256 * Fix any of the errors reported and commit to your local branch 257 * Finally, after the Linter reports no errors anymore, run `git status` or `git diff` to check for any changes made automatically by Lint 258 * If there were automatic changes, commit them to your local branch 259 260 If you touched UI code, you should also run the Yarn linter on it: 261 262 * Run `make lint-ui` 263 * Fix any of the errors reported by it 264 265 ## Contributing to Argo CD UI 266 267 Argo CD, along with Argo Workflows, uses shared React components from [Argo UI](https://github.com/argoproj/argo-ui). Examples of some of these components include buttons, containers, form controls, 268 and others. Although you can make changes to these files and run them locally, in order to have these changes added to the Argo CD repo, you will need to follow these steps. 269 270 1. Fork and clone the [Argo UI repository](https://github.com/argoproj/argo-ui). 271 272 2. `cd` into your `argo-ui` directory, and then run `yarn install`. 273 274 3. Make your file changes. 275 276 4. Run `yarn start` to start a [storybook](https://storybook.js.org/) dev server and view the components in your browser. Make sure all your changes work as expected. 277 278 5. Use [yarn link](https://classic.yarnpkg.com/en/docs/cli/link/) to link Argo UI package to your Argo CD repository. (Commands below assume that `argo-ui` and `argo-cd` are both located within the same parent folder) 279 280 * `cd argo-ui` 281 * `yarn link` 282 * `cd ../argo-cd/ui` 283 * `yarn link argo-ui` 284 285 Once `argo-ui` package has been successfully linked, test out changes in your local development environment. 286 287 6. Commit changes and open a PR to [Argo UI](https://github.com/argoproj/argo-ui). 288 289 7. Once your PR has been merged in Argo UI, `cd` into your `argo-cd/ui` folder and run `yarn add git+https://github.com/argoproj/argo-ui.git`. This will update the commit SHA in the `ui/yarn.lock` file to use the lastest master commit for argo-ui. 290 291 8. Submit changes to `ui/yarn.lock`in a PR to Argo CD. 292 293 ## Setting up a local toolchain 294 295 For development, you can either use the fully virtualized toolchain provided as Docker images, or you can set up the toolchain on your local development machine. Due to the dynamic nature of requirements, you might want to stay with the virtualized environment. 296 297 ### Install required dependencies and build-tools 298 299 !!!note 300 The installations instructions are valid for Linux hosts only. Mac instructions will follow shortly. 301 302 For installing the tools required to build and test Argo CD on your local system, we provide convenient installer scripts. By default, they will install binaries to `/usr/local/bin` on your system, which might require `root` privileges. 303 304 You can change the target location by setting the `BIN` environment before running the installer scripts. For example, you can install the binaries into `~/go/bin` (which should then be the first component in your `PATH` environment, i.e. `export PATH=~/go/bin:$PATH`): 305 306 ```shell 307 make BIN=~/go/bin install-tools-local 308 ``` 309 310 Additionally, you have to install at least the following tools via your OS's package manager (this list might not be always up-to-date): 311 312 * Git LFS plugin 313 * GnuPG version 2 314 315 ### Install Go dependencies 316 317 You need to pull in all required Go dependencies. To do so, run 318 319 * `make mod-download-local` 320 * `make mod-vendor-local` 321 322 ### Test your build toolchain 323 324 The first thing you can do to test whether your build toolchain is setup correctly is by generating the glue code for the API and after that, run a normal build: 325 326 * `make codegen-local` 327 * `make build-local` 328 329 This should return without any error. 330 331 ### Run unit-tests 332 333 The next thing is to make sure that unit tests are running correctly on your system. These will require that all dependencies, such as Helm, Kustomize, Git, GnuPG, etc are correctly installed and fully functioning: 334 335 * `make test-local` 336 337 ### Run end-to-end tests 338 339 The final step is running the End-to-End testsuite, which makes sure that your Kubernetes dependencies are working properly. This will involve starting all the Argo CD components locally on your computer. The end-to-end tests consists of two parts: a server component, and a client component. 340 341 * First, start the End-to-End server: `make start-e2e-local`. This will spawn a number of processes and services on your system. 342 * When all components have started, run `make test-e2e-local` to run the end-to-end tests against your local services. 343 344 For more information about End-to-End tests, refer to the [End-to-End test documentation](test-e2e.md).