development.md

Development Guide

This document is the canonical source of truth for things like supported toolchain versions for building Kubernetes.

Please submit an issue on Github if you

• Notice a requirement that this doc does not capture.
• Find a different doc that specifies requirements (the doc should instead link here).

Development branch requirements will change over time, but release branch requirements are frozen.

Pre submit flight checks

Determine whether your issue or pull request is improving Kubernetes' architecture or whether it's simply fixing a bug.

If you need a diagram, add it. SEPARATE the description of the problem (e.g. Y is a critical component that is too slow for an SLA that we care about) from the solution (e.g. make X faster).

Some of these checks were less common in Kubernetes' earlier days. Now that we have over 1000 contributors, each issue should be filed with care. No issue should take more than 5 minutes to check for sanity (even the busiest of reviewers can spare 5 minutes to review a patch that is thoughtfully justified).

Is this just a simple bug fix?

Simple bug patches are easy to review since test coverage is submitted with the patch. Bug fixes don't usually require a lot of extra testing, but please update the unit tests so they catch the bug!

Is this an architecture improvement?

Some examples of "Architecture" improvements include:

• Adding a new feature or making a feature more configurable or modular.
• Improving test coverage.
• Decoupling logic or creation of new utilities.
• Making code more resilient (sleeps, backoffs, reducing flakiness, etc.).

These sorts of improvements are easily evaluated, especially when they decrease lines of code without breaking functionality. That said, please explain exactly what you are 'cleaning up' in your Pull Request so as not to waste a reviewer's time.

If you're making code more resilient, include tests that demonstrate the new resilient behavior. For example: if your patch causes a controller to better handle inconsistent data, make a mock object which returns incorrect data a few times and verify the controller's new behaviour.

Is this a performance improvement ?

Performance bug reports MUST include data that demonstrates the bug. Without data, the issue will be closed. You can measure performance using kubemark, scheduler_perf, go benchmark tests, or e2e tests on a real cluster with metric plots.

Examples of how NOT to suggest a performance bug (these lead to a long review process and waste cycles):

• We should be doing X instead of Y because it might lead to better performance.
• Doing X instead of Y would reduce calls to Z.

The above statements have no value to a reviewer because neither is backed by data. Writing issues like this lands your PR in a no-man's-land and waste your reviewers' time.

Examples of possible performance improvements include (remember, you MUST document the improvement with data):

• Improving a caching implementation.
• Reducing calls to functions which are O(n^2)
• Reducing dependence on API server requests.
• Changing the value of default parameters for processes, or making those values 'smarter'.
• Parallelizing a calculation that needs to run on a large set of node/pod objects.

These issues should always be submitted with (in decreasing order or value):

• A golang Benchmark test.
• A visual depiction of reduced metric load on a cluster (measurable using metrics/ endpoints and grafana).
• A hand-instrumented timing test (i.e. adding some logs into the controller manager).

Here are some examples of properly submitted performance issues. If you are new to kubernetes and thinking about filing a performance optimization, re-read one or all of these before you get started.

• kubernetes/kubernetes#18266 (apiserver)
• kubernetes/kubernetes#32833 (node)
• kubernetes/kubernetes#31795 (scheduler)

Since performance improvements can be empirically measured, you should follow the "scientific method" of creating a hypothesis, collecting data, and then revising your hypothesis. The above issues do this transparently, using figures and data rather then conjecture. Notice that the problem is analyzed and a correct solution is created before a single line of code is reviewed.

Building Kubernetes with Docker

Official releases are built using Docker containers. To build Kubernetes using Docker please follow these instructions.

Building Kubernetes on a local OS/shell environment

Kubernetes development helper scripts assume an up-to-date GNU tools environment. Recent Linux distros should work out-of-the-box.

Mac OS X ships with outdated BSD-based tools. We recommend installing OS X GNU tools.

etcd

Kubernetes maintains state in etcd, a distributed key store.

Please install it locally to run local integration tests.

Go

Kubernetes is written in Go. If you don't have a Go development environment, please set one up.

Kubernetes	requires Go
1.0 - 1.2	1.4.2
1.3, 1.4	1.6
1.5, 1.6	1.7 - 1.7.5
1.7+	1.8.1

Ensure your GOPATH and PATH have been configured in accordance with the Go environment instructions.

Upgrading Go

Upgrading Go requires specific modification of some scripts and container images.

• The image for cross compiling in build/build-image/cross. The VERSION file and Dockerfile.
• Update dockerized-e2e-runner.sh to run a kubekins-e2e with the desired Go version. This requires pushing the e2e and test images that are FROM the desired Go version.
• The cross tag KUBE_BUILD_IMAGE_CROSS_TAG in build/common.sh.

Dependency management

Kubernetes build/test scripts use godep to manage dependencies.

go get -u github.com/tools/godep

The Godep version that Kubernetes is using is listed in Godep/Godep.json (in the kubernetes repo root). See what version you are running with this command:

godep version

Developers planning to manage dependencies in the vendor/ tree may want to explore alternative environment setups. See using godep to manage dependencies.

Build with Bazel/Gazel

Building with Bazel is currently experimental. For more information, see Build with Bazel.

Workflow

1 Fork in the cloud

1. Visit https://github.com/kubernetes/kubernetes
2. Click Fork button (top right) to establish a cloud-based fork.

2 Clone fork to local storage

Per Go's workspace instructions, place Kubernetes' code on your GOPATH using the following cloning procedure.

Define a local working directory:

# If your GOPATH has multiple paths, pick
# just one and use it instead of $GOPATH here.
# You must follow exactly this pattern,
# neither `$GOPATH/src/github.com/${your github profile name/`
# nor any other pattern will work.
working_dir=$GOPATH/src/k8s.io

If you already do Go development on github, the k8s.io directory will be a sibling to your existing github.com directory.

Set user to match your github profile name:

user={your github profile name}

Both $working_dir and $user are mentioned in the figure above.

Create your clone:

mkdir -p $working_dir
cd $working_dir
git clone https://github.com/$user/kubernetes.git
# or: git clone [email protected]:$user/kubernetes.git

cd $working_dir/kubernetes
git remote add upstream https://github.com/kubernetes/kubernetes.git
# or: git remote add upstream [email protected]:kubernetes/kubernetes.git

# Never push to upstream master
git remote set-url --push upstream no_push

# Confirm that your remotes make sense:
git remote -v

3 Branch

Get your local master up to date:

cd $working_dir/kubernetes
git fetch upstream
git checkout master
git rebase upstream/master

Branch from it:

git checkout -b myfeature

Then edit code on the myfeature branch.

Build

Note: If you are using CDPATH, you must either start it with a leading colon, or unset the variable. The make rules and scripts to build require the current directory to come first on the CD search path in order to properly navigate between directories.

cd $working_dir/kubernetes
make

To remove the limit on the number of errors the Go compiler reports (default limit is 10 errors):

make GOGCFLAGS="-e"

To build with optimizations disabled (enables use of source debug tools):

make GOGCFLAGS="-N -l"

To build binaries for all platforms:

make cross

Install etcd

hack/install-etcd.sh

Test

cd $working_dir/kubernetes

# Run all the presubmission verification. Then, run a specific update script (hack/update-*.sh)
# for each failed verification. For example:
#   hack/update-gofmt.sh (to make sure all files are correctly formatted, usually needed when you add new files)
#   hack/update-bazel.sh (to update bazel build related files, usually needed when you add or remove imports)
make verify

# Alternatively, run all update scripts to avoid fixing verification failures one by one.
make update

# Run every unit test
make test

# Run package tests verbosely
make test WHAT=./pkg/api/helper GOFLAGS=-v

# Run integration tests, requires etcd
# For more info, visit https://github.com/kubernetes/community/blob/master/contributors/devel/testing.md#integration-tests
make test-integration

# Run e2e tests by building test binaries, turn up a test cluster, run all tests, and tear the cluster down
# Equivalent to: go run hack/e2e.go -- -v --build --up --test --down
# Note: running all e2e tests takes a LONG time! To run specific e2e tests, visit:
# https://github.com/kubernetes/community/blob/master/contributors/devel/e2e-tests.md#building-kubernetes-and-running-the-tests
make test-e2e

See the testing guide and end-to-end tests for additional information and scenarios.

Run make help for additional information on these make targets.

4 Keep your branch in sync

# While on your myfeature branch
git fetch upstream
git rebase upstream/master

5 Commit

Commit your changes.

git commit

Likely you go back and edit/build/test some more then commit --amend in a few cycles.

6 Push

When ready to review (or just to establish an offsite backup or your work), push your branch to your fork on github.com:

git push -f ${your_remote_name} myfeature

7 Create a pull request

1. Visit your fork at https://github.com/$user/kubernetes
2. Click the Compare & Pull Request button next to your myfeature branch.
3. Check out the pull request process for more details.

If you have upstream write access, please refrain from using the GitHub UI for creating PRs, because GitHub will create the PR branch inside the main repository rather than inside your fork.

Get a code review

Once your pull request has been opened it will be assigned to one or more reviewers. Those reviewers will do a thorough code review, looking for correctness, bugs, opportunities for improvement, documentation and comments, and style.

Commit changes made in response to review comments to the same branch on your fork.

Very small PRs are easy to review. Very large PRs are very difficult to review. At the assigned reviewer's discretion, a PR may be switched to use Reviewable instead. Once a PR is switched to Reviewable, please ONLY send or reply to comments through Reviewable. Mixing code review tools can be very confusing.

See Faster Reviews for some thoughts on how to streamline the review process.

Squash and Merge

Upon merge (by either you or your reviewer), all commits left on the review branch should represent meaningful milestones or units of work. Use commits to add clarity to the development and review process.

Before merging a PR, squash any fix review feedback, typo, and rebased sorts of commits.

It is not imperative that every commit in a PR compile and pass tests independently, but it is worth striving for.

For mass automated fixups (e.g. automated doc formatting), use one or more commits for the changes to tooling and a final commit to apply the fixup en masse. This makes reviews easier.