4.7 KiB
Developing for NGINX Ingress Controller
This document explains how to get started with developing for NGINX Ingress controller. It includes how to build, test, and release ingress controllers.
Quick Start
Getting the code
The code must be checked out as a subdirectory of k8s.io, and not github.com.
mkdir -p $GOPATH/src/k8s.io
cd $GOPATH/src/k8s.io
# Replace "$YOUR_GITHUB_USERNAME" below with your github username
git clone https://github.com/$YOUR_GITHUB_USERNAME/ingress-nginx.git
cd ingress-nginx
Initial developer environment build
$ make dev-env
Updating the deployment
The nginx controller container image can be rebuilt using:
$ ARCH=amd64 TAG=dev REGISTRY=$USER/ingress-controller make build image
The image will only be used by pods created after the rebuild. To delete old pods which will cause new ones to spin up:
$ kubectl get pods -n ingress-nginx
$ kubectl delete pod -n ingress-nginx ingress-nginx-controller-<unique-pod-id>
Dependencies
The build uses dependencies in the vendor directory, which
must be installed before building a binary/image. Occasionally, you
might need to update the dependencies.
This guide requires you to install go 1.13 or newer.
This will automatically save the dependencies to the vendor/ directory.
$ go get
$ make dep-ensure
Building
All ingress controllers are built through a Makefile. Depending on your requirements you can build a raw server binary, a local container image, or push an image to a remote repository.
In order to use your local Docker, you may need to set the following environment variables:
# "gcloud docker" (default) or "docker"
$ export DOCKER=<docker>
# "quay.io/kubernetes-ingress-controller" (default), "index.docker.io", or your own registry
$ export REGISTRY=<your-docker-registry>
To find the registry simply run: docker system info | grep Registry
Building the e2e test image
The e2e test image can also be built through the Makefile.
$ make -C test/e2e-image build
Then you can load the docker image using kind:
$ kind load docker-image --name="ingress-nginx-dev" nginx-ingress-controller:e2e
Nginx Controller
Build a raw server binary
$ make build
TODO: add more specific instructions needed for raw server binary.
Build a local container image
$ TAG=<tag> REGISTRY=$USER/ingress-controller make image
Deploying
There are several ways to deploy the ingress controller onto a cluster. Please check the deployment guide
Testing
To run unit-tests, just run
$ cd $GOPATH/src/k8s.io/ingress-nginx
$ make test
If you have access to a Kubernetes cluster, you can also run e2e tests using ginkgo.
$ cd $GOPATH/src/k8s.io/ingress-nginx
$ KIND_CLUSTER_NAME="ingress-nginx-test" make kind-e2e-test
To set focus to a particular set of tests, a FOCUS flag can be set.
KIND_CLUSTER_NAME="ingress-nginx-test" FOCUS="no-auth-locations" make kind-e2e-test
NOTE: if your e2e pod keeps hanging in an ImagePullBackoff, make sure you've made your e2e nginx-ingress-controller image available to minikube as explained in the Building the e2e test image section
To run unit-tests for lua code locally, run:
$ cd $GOPATH/src/k8s.io/ingress-nginx
$ ./rootfs/etc/nginx/lua/test/up.sh
$ make lua-test
Lua tests are located in $GOPATH/src/k8s.io/ingress-nginx/rootfs/etc/nginx/lua/test. When creating a new test file it must follow the naming convention <mytest>_test.lua or it will be ignored.
Releasing
All Makefiles will produce a release binary, as shown above. To publish this
to a wider Kubernetes user base, push the image to a container registry, like
gcr.io. All release images are hosted under gcr.io/google_containers and
tagged according to a semver scheme.
An example release might look like:
$ make release
Please follow these guidelines to cut a release:
- Update the release page with a short description of the major changes that correspond to a given image tag.
- Cut a release branch, if appropriate. Release branches follow the format of
controller-release-version. Typically, pre-releases are cut from HEAD. All major feature work is done in HEAD. Specific bug fixes are cherry-picked into a release branch. - If you're not confident about the stability of the code, tag it as alpha or beta. Typically, a release branch should have stable code.