This guide explains how to set up your environment for developing on Helm and Tiller.
- The latest version of Go
- The latest version of Glide
- A Kubernetes cluster w/ kubectl (optional)
- The gRPC toolchain
We use Make to build our programs. The simplest way to get started is:
$ make bootstrap build
NOTE: This will fail if not running from the path
k8s.io should not be a symlink or
build will not find the relevant
This will build both Helm and Tiller.
make bootstrap will attempt to
install certain tools if they are missing.
To run all of the tests (without running the tests for
To run Helm and Tiller locally, you can run
- Helm and Tiller are known to run on macOS and most Linuxes, including Alpine.
- Tiller must have access to a Kubernetes cluster. It learns about the
cluster by examining the Kube config files that
Man pages and Markdown documentation are already pre-built in
docs/. You may
regenerate documentation using
To expose the Helm man pages to your
man client, you can put the files in your
$ export MANPATH=$GOPATH/src/k8s.io/helm/docs/man:$MANPATH $ man helm
gRPC and Protobuf
Helm and Tiller communicate using gRPC. To get started with gRPC, you will need to…
protocfor compiling protobuf files. Releases are here
- Run Helm’s
make bootstrapto generate the
protoc-gen-goplugin and place it in
Note that you need to be on protobuf 3.2.0 (
protoc --version). The
protoc-gen-go is tied to the version of gRPC used in
Kubernetes. So the plugin is maintained locally.
While the gRPC and ProtoBuf specs remain silent on indentation, we require that the indentation style matches the Go format specification. Namely, protocol buffers should use tab-based indentation and rpc declarations should follow the style of Go function declarations.
The Helm API (HAPI)
We use gRPC as an API layer. See
pkg/proto/hapi for the generated Go code,
_proto for the protocol buffer definitions.
To regenerate the Go files from the protobuf source,
To build Docker images, use
Pre-build images are already available in the official Kubernetes Helm GCR registry.
Running a Local Cluster
For development, we highly recommend using the
developer-oriented distribution. Once this is installed, you can use
helm init to install into the cluster.
For developing on Tiller, it is sometimes more expedient to run Tiller locally instead of packaging it into an image and running it in-cluster. You can do this by telling the Helm client to us a local instance.
$ make build $ bin/tiller
And to configure the Helm client, use the
--host flag or export the
$ export HELM_HOST=localhost:44134 $ helm install foo
(Note that you do not need to use
helm init when you are running Tiller directly)
Tiller should run on any >= 1.3 Kubernetes cluster.
We welcome contributions. This project has set up some guidelines in order to ensure that (a) code quality remains high, (b) the project remains consistent, and © contributions follow the open source legal requirements. Our intent is not to burden contributors, but to build elegant and high-quality open source code so that our users will benefit.
Make sure you have read and understood the main CONTRIBUTING guide:
Structure of the Code
The code for the Helm project is organized as follows:
- The individual programs are located in
cmd/. Code inside of
cmd/is not designed for library re-use.
- Shared libraries are stored in
- The raw ProtoBuf files are stored in
hapistands for the Helm Application Programming Interface).
- The Go files generated from the
protodefinitions are stored in
scripts/directory contains a number of utility scripts. Most of these are used by the CI/CD pipeline.
rootfs/folder is used for Docker-specific files.
docs/folder is used for documentation and examples.
Go dependencies are managed with
Glide and stored in the
We use Git for our version control system. The
master branch is the
home of the current development candidate. Releases are tagged.
We accept changes to the code via GitHub Pull Requests (PRs). One workflow for doing this is as follows:
- Go to your
- Fork that repository into your GitHub account
- Add your repository as a remote for
- Create a new working branch (
git checkout -b feat/my-feature) and do your work on that branch.
- When you are ready for us to review, push your branch to GitHub, and then open a new pull request with us.
For Git commit messages, we follow the Semantic Commit Messages:
fix(helm): add --foo flag to 'helm install' When 'helm install --foo bar' is run, this will print "foo" in the output regardless of the outcome of the installation. Closes #1234
Common commit types:
- fix: Fix a bug or error
- feat: Add a new feature
- docs: Change documentation
- test: Improve testing
- ref: refactor existing code
- helm: The Helm CLI
- tiller: The Tiller server
- proto: Protobuf definitions
- pkg/lint: The lint package. Follow a similar convention for any package
*: two or more scopes
We follow the Go coding style standards very closely. Typically, running
go fmt will make your code beautiful for you.
We also typically follow the conventions recommended by
go lint and
make test-style to test the style conformance.
Because this project is largely Go code, we format our Protobuf files as closely to Go as possible. There are currently no real formatting rules or guidelines for Protobuf, but as they emerge, we may opt to follow those instead.
Standards: - Tabs for indentation, not spaces. - Spacing rules follow Go conventions (curly braces at line end, spaces around operators).
- Files should specify their package with
option go_package = "...";
- Comments should translate into good Go code comments (since
copies comments into the destination source code file).
- RPC functions are defined in the same file as their request/response
- Deprecated RPCs, messages, and fields are marked deprecated in the comments (
DEPRECATED updates a foo.).