- Antrea Network Policy
- Traffic Encryption (Ipsec / WireGuard)
- Securing Control Plane
- OS-specific Known Issues
- OVS Pipeline
- Feature Gates
- Network Flow Visibility
- Traceflow Guide
- NoEncap and Hybrid Traffic Modes
- Egress Guide
- NodePortLocal Guide
- Antrea API Groups
- Antrea API Reference
Thank you for taking the time out to contribute to project Antrea!
This guide will walk you through the process of making your first commit and how to effectively get it merged upstream.
- Getting Started
- Issue and PR Management
To get started, let’s ensure you have completed the following prerequisites for contributing to project Antrea:
- Read and observe the code of conduct.
- Check out the Architecture document for the Antrea architecture and design.
- Set up necessary accounts.
- Set up your development environment
Now that you’re setup, skip ahead to learn how to contribute.
At minimum, you need the following accounts for effective participation:
- Github: Committing any change requires you to have a github account.
- Slack: Join the Kubernetes Slack and look for our #antrea channel.
- Google Group: Join our mailing list.
There are multiple ways in which you can contribute, either by contributing code in the form of new features or bug-fixes or non-code contributions like helping with code reviews, triaging of bugs, documentation updates, filing new issues or writing blogs/manuals etc.
In order to help you get your hands “dirty”, there is a list of starter issues from which you can choose.
Developers work in their own forked copy of the repository and when ready, submit pull requests to have their changes considered and merged into the project’s repository.
Fork your own copy of the repository to your GitHub account by clicking on
Forkbutton on Antrea’s GitHub repository.
Clone the forked repository on your local setup.
git clone https://github.com/$user/antrea
Add a remote upstream to track upstream Antrea repository.
git remote add upstream https://github.com/antrea-io/antrea
Never push to upstream remote
git remote set-url --push upstream no_push
Create a topic branch.
git checkout -b branchName
Make changes and commit it locally. Make sure that your commit is signed.
git add <modifiedFile> git commit -s
Keeping branch in sync with upstream.
git checkout branchName git fetch upstream git rebase upstream/main
Push local branch to your forked repository.
git push -f $remoteBranchName branchName
Create a Pull request on GitHub. Visit your fork at
Compare & Pull Requestbutton next to your
Once you have opened a Pull Request (PR), reviewers will be assigned to your PR and they may provide review comments which you need to address. Commit changes made in response to review comments to the same branch on your fork. Once a PR is ready to merge, squash any fix review feedback, typo and merged sorts of commits.
To make it easier for reviewers to review your PR, consider the following:
- Follow the golang coding conventions.
- Format your code with
make golangci-fix; if the linters flag an issue that cannot be fixed automatically, an error message will be displayed so you can address the issue.
- Follow git commit guidelines.
- Follow logging guidelines.
If your PR fixes a bug or implements a new feature, add the appropriate test cases to our automated test suite to guarantee enough coverage. A PR that makes significant code changes without contributing new test cases will be flagged by reviewers and will not be accepted.
Getting your PR verified by CI
It is a requirement to get your PR verified with CI checks before it gets merged. Also, it helps to find possible bugs before the review work starts. Once you create a PR, or you push new commits, CI checks at the bottom of a PR page will be refreshed. Checks include Github Action ones and Jenkins ones. Github Action ones will be triggered automatically when you push to the head branch of the PR but Jenkins ones need to be triggered manually with comments. Please note that if you are a first-time contributor, the Github workflows need approval from someone with write access to the repo. It’s a Github security mechanism.
Here are the trigger phrases for individual checks:
/test-integration: Integration tests
/test-e2e: Linux IPv4 e2e tests
/test-conformance: Linux IPv4 conformance tests
/test-networkpolicy: Linux IPv4 networkpolicy tests
/test-all-features-conformance: Linux IPv4 conformance tests with all features enabled
/test-windows-e2e: Windows IPv4 e2e tests
/test-windows-conformance: Windows IPv4 conformance tests
/test-windows-networkpolicy: Windows IPv4 networkpolicy tests
/test-windows-proxyall-e2e: Windows IPv4 e2e tests with proxyAll enabled
/test-ipv6-e2e: Linux dual stack e2e tests
/test-ipv6-conformance: Linux dual stack conformance tests
/test-ipv6-networkpolicy: dLinux ual stack networkpolicy tests
/test-ipv6-only-e2e: Linux IPv6 only e2e tests
/test-ipv6-only-conformance: Linux IPv6 only conformance tests
/test-ipv6-only-networkpolicy: Linux IPv6 only networkpolicy tests
/test-flexible-ipam-e2e: Flexible IPAM e2e tests
Here are the trigger phrases for groups of checks:
/test-all: Linux IPv4 tests
/test-windows-all: Windows IPv4 tests, including e2e tests with proxyAll enabled
/test-ipv6-all: Linux dual stack tests
/test-ipv6-only-all: Linux IPv6 only tests
Besides, you can skip a check with
/skip-e2e: skip Linux IPv4
e2e tests. Integration check cannot be skipped.
Skipping a check should be used only when the change doesn’t influence the specific function. For example:
- doc change: skip all checks
- comment change: skip all checks
- test/e2e/* change: skip conformance and networkpolicy checks
- *_windows.go change: skip Linux checks
For more information about the tests we run as part of CI, please refer to ci/README.md.
Cherry-picks to release branches
If your PR fixes a critical bug, it may need to be backported to older release branches which are still maintained. If this is the case, one of the Antrea maintainers will let you know once your PR is approved. Please refer to the documentation on cherry-picks for more information.
For symbol names and documentation, do not introduce new usage of harmful language such as ‘master / slave’ (or ‘slave’ independent of ‘master’) and ‘blacklist / whitelist’. For more information about what constitutes harmful language and for a reference word replacement list, please refer to the Inclusive Naming Initiative.
We are committed to removing all harmful language from the project. If you detect existing usage of harmful language in code or documentation, please report the issue to us or open a Pull Request to address it directly. Thanks!
Building and testing your change
To build the Antrea Docker image together with all Antrea bits, you can simply do:
- Checkout your feature branch and
The second step will compile the Antrea code in a
golang container, and build
Ubuntu 20.04 Docker image that includes all the generated binaries.
must be installed on your local machine in advance.
Alternatively, you can build the Antrea code in your local Go environment. The
Antrea project uses the
Go modules support which was introduced in Go 1.11. It
facilitates dependency tracking and no longer requires projects to live inside
To develop locally, you can follow these steps:
- Install Go 1.17
- Checkout your feature branch and
- To build all Go files and install them under
- To run all Go unit tests, run
- To build the Antrea Ubuntu Docker image separately with the binaries generated in step 2, run
Reverting a commit
Create a branch in your forked repo
git checkout -b revertName
Sync the branch with upstream
git fetch upstream git rebase upstream/main
Create a revert based on the SHA of the commit. The commit needs to be signed.
git revert -s SHA
Push this new commit.
git push $remoteRevertName revertName
Create a Pull Request on GitHub. Visit your fork at
Compare & Pull Requestbutton next to your
Sign-off Your Work
As a CNCF project, Antrea must enforce the
Developer Certificate of
Origin (DCO) on all Pull Requests. We
require that for all commits constituting the Pull Request, the commit message
Signed-off-by line with an email address that matches the commit
author. By adding this line to their commit messages, contributors sign-off
that they adhere to the requirements of the DCO.
Git provides the
-s command-line option to append the required line
automatically to the commit message:
git commit -s -m 'This is my commit message'
For an existing commit, you can also use this option with
git commit -s --amend
If more than one person works on something it’s possible for more than one person to sign-off on it. For example:
Signed-off-by: Some Developer email@example.com Signed-off-by: Another Developer firstname.lastname@example.org
We use the
DCO Github App to enforce that all
commits in a Pull Request include the required
Signed-off-by line. If this is
not the case, the app will report a failed status for the Pull Request and it
will be blocked from being merged.
Compared to our earlier CLA, DCO tends to make the experience simpler for new contributors. If you are contributing as an employee, there is no need for your employer to sign anything; the DCO assumes you are authorized to submit contributions (it’s your responsibility to check with your employer).
Issue and PR Management
We use labels and workflows (some manual, some automated with GitHub Actions) to help us manage triage, prioritize, and track issue progress. For a detailed discussion, see docs/issue-management.md.
Filing An Issue
Help is always appreciated. If you find something that needs fixing, please file an issue here. Please ensure that the issue is self explanatory and has enough information for an assignee to get started.
Before picking up a task, go through the existing issues and make sure that your change is not already being worked on. If it does not exist, please create a new issue and discuss it with other members.
For simple contributions to Antrea, please ensure that this minimum set of labels are included on your issue:
- kind – common ones are
kind/design. For an overview of the different types of issues that can be submitted, see Issue and PR Kinds. The kind of issue will determine the issue workflow.
- area (optional) – if you know the area the issue belongs in, you can assign it.
Otherwise, another community member will label the issue during triage. The
area label will identify the area of interest an issue or PR belongs in and
will ensure the appropriate reviewers shepherd the issue or PR through to its
closure. For an overview of areas, see the
- size (optional) – if you have an idea of the size (lines of code, complexity, effort) of the issue, you can label it using a size label. The size can be updated during backlog grooming by contributors. This estimate is used to guide the number of features selected for a milestone.
All other labels will be assigned during issue triage.
Once an issue has been submitted, the CI (GitHub actions) or a human will
automatically review the submitted issue or PR to ensure that it has all relevant
information. If information is lacking or there is another problem with the
submitted issue, an appropriate
triage/<?> label will be applied.
After an issue has been triaged, the maintainers can prioritize the issue with
Once an issue has been submitted, categorized, triaged, and prioritized it
is marked as
ready-to-work. A ready-to-work issue should have labels
indicating assigned areas, prioritization, and should not have any remaining
Issue and PR Kinds
kind label to describe the kind of issue or PR you are submitting. Valid
kind/api-change– for api changes
kind/bug– for filing a bug
kind/cleanup– for code cleanup and organization
kind/deprecation– for deprecating a feature
kind/design– for proposing a design or architectural change
kind/documentation– for updating documentation
kind/failing-test– for reporting a failed test (may create with automation in future)
kind/feature– for proposing a feature
kind/support– to request support. You may also get support by using our Slack channel for interactive help. If you have not set up the appropriate accounts, please follow the instructions in accounts setup.
For more details on how we manage issues, please read our Issue Management doc.