We welcome contributions in the form of issues and pull requests. We view the contributions and the process as the same for github and external contributors.
IMPORTANT: Building your own runner is critical for the dev inner loop process when contributing changes. However, only runners built and distributed by GitHub (releases) are supported in production. Be aware that workflows and orchestrations run service side with the runner being a remote process to run steps. For that reason, the service can pull the runner forward so customizations can be lost.
Log issues for both bugs and enhancement requests. Logging issues are important for the open community.
Issues in this repository should be for the runner application. Note that the VM and virtual machine images (including the developer toolsets) installed on the actions hosted machine pools are located in this repository
We ask that before significant effort is put into code changes, that we have agreement on taking the change before time is invested in code changes.
- Create a feature request. Once agreed we will take the enhancement
- Create an ADR to agree on the details of the change.
An ADR is an Architectural Decision Record. This allows consensus on the direction forward and also serves as a record of the change and motivation. Read more here
Git for Windows and Linux Install Here (needed for dev sh script)
If you just want to get from building the sourcecode to using it to execute an action, you will need:
- The url of your repository
- A runner registration token. You can find it at
https://github.com/{your-repo}/settings/actions/runners/new
git clone https://github.com/actions/runner
cd runner/src
./dev.(sh/cmd) layout # the runner that built from source is in {root}/_layout
cd ../_layout
./config.(sh/cmd) --url https://github.com/{your-repo} --token ABCABCABCABCABCABCABCABCABCAB # accept default name, labels and work folder
./run.(sh/cmd)
If you trigger a job now, you can see the runner execute it.
Tip: Make sure your job can run on this runner. The easiest way is to set runs-on: self-hosted
in the workflow file.
If you're using VS Code, you can follow these steps instead.
Navigate to the src
directory and run the following command:
Commands:
layout
(l
): Run first time to create a full runner layout in{root}/_layout
build
(b
): Build everything and update runner layout foldertest
(t
): Build runner binaries and run unit tests
git clone https://github.com/actions/runner
cd runner
cd ./src
./dev.(sh/cmd) layout # the runner that built from source is in {root}/_layout
<make code changes>
./dev.(sh/cmd) build # {root}/_layout will get updated
./dev.(sh/cmd) test # run all unit tests before git commit/push
Let's break that down.
git clone https://github.com/actions/runner
cd runner
If you want to push your changes to a remote, it is recommended you fork the repository and use that fork as your origin instead of https://github.com/actions/runner
.
This command will build all projects, then copies them and other dependencies into a folder called _layout
. The binaries in this folder are then used for running, debugging the runner.
cd ./src # execute the script from this folder
./dev.(sh/cmd) layout # the runner that built from source is in {root}/_layout
If you make code changes after this point, use the argument build
to build your code in the src
folder to keep your _layout
folder up to date.
cd ./src
./dev.(sh/cmd) build # {root}/_layout will get updated
This command runs the suite of unit tests in the project
cd ./src
./dev.(sh/cmd) test # run all unit tests before git commit/push
If you want to manually test your runner and run actions from a real repository, you'll have to configure it before running it.
cd runner/_layout
./config.(sh/cmd) # configure your custom runner
You will need your the name of your repository and a runner registration token. Check Quickstart if you don't know how to get this token.
These can also be passed down as arguments to config.(sh/cmd)
:
cd runner/_layout
./config.(sh/cmd) --url https://github.com/{your-repo} --token ABCABCABCABCABCABCABCABCABCAB
All that's left to do is to start the runner:
cd runner/_layout
./run.(sh/cmd) # run your custom runner
cd runner/_layout/_diag
ls
cat (Runner/Worker)_TIMESTAMP.log # view your log file
Using Visual Studio Code Using Visual Studio
We use the .NET Foundation and CoreCLR style guidelines located here