flux2-kustomize-helm-example

For this example we assume a scenario with two clusters: staging and production. The end goal is to leverage Flux and Kustomize to manage both clusters while minimizing duplicated declarations.

We will configure Flux to install, test and upgrade a demo app using HelmRepository and HelmRelease custom resources. Flux will monitor the Helm repository, and it will automatically upgrade the Helm releases to their latest chart version based on semver ranges.

Prerequisites

You will need a Kubernetes cluster version 1.16 or newer and kubectl version 1.18. For a quick local test, you can use Kubernetes kind. Any other Kubernetes setup will work as well though.

In order to follow the guide you'll need a GitHub account and a personal access token that can create repositories (check all permissions under repo).

Install the Flux CLI on MacOS and Linux using Homebrew:

brew install fluxcd/tap/flux

Or install the CLI by downloading precompiled binaries using a Bash script:

curl -s https://fluxcd.io/install.sh | sudo bash

Repository structure

The Git repository contains the following top directories:

• apps dir contains Helm releases with a custom configuration per cluster
• infrastructure dir contains common infra tools such as NGINX ingress controller and Helm repository definitions
• clusters dir contains the Flux configuration per cluster

├── apps
│   ├── base
│   ├── production 
│   └── staging
├── infrastructure
│   ├── nginx
│   ├── redis
│   └── sources
└── clusters
    ├── production
    └── staging

The apps configuration is structured into:

• apps/base/ dir contains namespaces and Helm release definitions
• apps/production/ dir contains the production Helm release values
• apps/staging/ dir contains the staging values

./apps/
├── base
│   └── podinfo
│       ├── kustomization.yaml
│       ├── namespace.yaml
│       └── release.yaml
├── production
│   ├── kustomization.yaml
│   └── podinfo-patch.yaml
└── staging
    ├── kustomization.yaml
    └── podinfo-patch.yaml

In apps/base/podinfo/ dir we have a HelmRelease with common values for both clusters:

apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
  name: podinfo
  namespace: podinfo
spec:
  releaseName: podinfo
  chart:
    spec:
      chart: podinfo
      sourceRef:
        kind: HelmRepository
        name: podinfo
        namespace: flux-system
  interval: 5m
  values:
    cache: redis-master.redis:6379
    ingress:
      enabled: true
      annotations:
        kubernetes.io/ingress.class: nginx

In apps/staging/ dir we have a Kustomize patch with the staging specific values:

apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
  name: podinfo
spec:
  chart:
    spec:
      version: ">=1.0.0-alpha"
  test:
    enable: true
  values:
    ingress:
      hosts:
        - host: podinfo.staging

Note that with version: ">=1.0.0-alpha" we configure Flux to automatically upgrade the HelmRelease to the latest chart version including alpha, beta and pre-releases.

In apps/production/ dir we have a Kustomize patch with the production specific values:

apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
  name: podinfo
  namespace: podinfo
spec:
  chart:
    spec:
      version: ">=1.0.0"
  values:
    ingress:
      hosts:
        - host: podinfo.production

Note that with version: ">=1.0.0" we configure Flux to automatically upgrade the HelmRelease to the latest stable chart version (alpha, beta and pre-releases will be ignored).

Infrastructure:

./infrastructure/
├── nginx
│   ├── kustomization.yaml
│   ├── namespace.yaml
│   └── release.yaml
├── redis
│   ├── kustomization.yaml
│   ├── namespace.yaml
│   └── release.yaml
└── sources
    ├── bitnami.yaml
    ├── kustomization.yaml
    └── podinfo.yaml

In infrastructure/sources/ dir we have the Helm repositories definitions:

apiVersion: source.toolkit.fluxcd.io/v1beta2
kind: HelmRepository
metadata:
  name: podinfo
spec:
  interval: 5m
  url: https://stefanprodan.github.io/podinfo
---
apiVersion: source.toolkit.fluxcd.io/v1beta2
kind: HelmRepository
metadata:
  name: bitnami
spec:
  interval: 30m
  url: https://charts.bitnami.com/bitnami

Note that with interval: 5m we configure Flux to pull the Helm repository index every five minutes. If the index contains a new chart version that matches a HelmRelease semver range, Flux will upgrade the release.

Bootstrap staging and production

The clusters dir contains the Flux configuration:

./clusters/
├── production
│   ├── apps.yaml
│   └── infrastructure.yaml
└── staging
    ├── apps.yaml
    └── infrastructure.yaml

In clusters/staging/ dir we have the Kustomization definitions:

apiVersion: kustomize.toolkit.fluxcd.io/v1beta2
kind: Kustomization
metadata:
  name: apps
  namespace: flux-system
spec:
  interval: 10m0s
  dependsOn:
    - name: infrastructure
  sourceRef:
    kind: GitRepository
    name: flux-system
  path: ./apps/staging
  prune: true
  wait: true
---
apiVersion: kustomize.toolkit.fluxcd.io/v1beta2
kind: Kustomization
metadata:
  name: infrastructure
  namespace: flux-system
spec:
  interval: 10m0s
  sourceRef:
    kind: GitRepository
    name: flux-system
  path: ./infrastructure
  prune: true

Note that with path: ./apps/staging we configure Flux to sync the staging Kustomize overlay and with dependsOn we tell Flux to create the infrastructure items before deploying the apps.

Fork this repository on your personal GitHub account and export your GitHub access token, username and repo name:

export GITHUB_TOKEN=<your-token>
export GITHUB_USER=<your-username>
export GITHUB_REPO=<repository-name>

Verify that your staging cluster satisfies the prerequisites with:

flux check --pre

Set the kubectl context to your staging cluster and bootstrap Flux:

flux bootstrap github \
    --context=staging \
    --owner=${GITHUB_USER} \
    --repository=${GITHUB_REPO} \
    --branch=main \
    --personal \
    --path=clusters/staging

The bootstrap command commits the manifests for the Flux components in clusters/staging/flux-system dir and creates a deploy key with read-only access on GitHub, so it can pull changes inside the cluster.

Watch for the Helm releases being install on staging:

$ watch flux get helmreleases --all-namespaces 
NAMESPACE	NAME   	REVISION	SUSPENDED	READY	MESSAGE                          
nginx    	nginx  	5.6.14  	False    	True 	release reconciliation succeeded	
podinfo  	podinfo	5.0.3   	False    	True 	release reconciliation succeeded	
redis    	redis  	11.3.4  	False    	True 	release reconciliation succeeded

Verify that the demo app can be accessed via ingress:

$ kubectl -n nginx port-forward svc/nginx-ingress-controller 8080:80 &

$ curl -H "Host: podinfo.staging" http://localhost:8080
{
  "hostname": "podinfo-59489db7b5-lmwpn",
  "version": "5.0.3"
}

Bootstrap Flux on production by setting the context and path to your production cluster:

flux bootstrap github \
    --context=production \
    --owner=${GITHUB_USER} \
    --repository=${GITHUB_REPO} \
    --branch=main \
    --personal \
    --path=clusters/production

Watch the production reconciliation:

$ flux get kustomizations --watch
NAME          	REVISION                                        READY
apps          	main/797cd90cc8e81feb30cfe471a5186b86daf2758d	True
flux-system   	main/797cd90cc8e81feb30cfe471a5186b86daf2758d	True
infrastructure	main/797cd90cc8e81feb30cfe471a5186b86daf2758d	True

Encrypt Kubernetes secrets

In order to store secrets safely in a Git repository, you can use Mozilla's SOPS CLI to encrypt Kubernetes secrets with OpenPGP or KMS.

Install gnupg and sops:

brew install gnupg sops

Generate a GPG key for Flux without specifying a passphrase and retrieve the GPG key ID:

$ gpg --full-generate-key
Email address: [email protected]

$ gpg --list-secret-keys [email protected]
sec   rsa3072 2020-09-06 [SC]
      1F3D1CED2F865F5E59CA564553241F147E7C5FA4

Create a Kubernetes secret on your clusters with the private key:

gpg --export-secret-keys \
--armor 1F3D1CED2F865F5E59CA564553241F147E7C5FA4 |
kubectl create secret generic sops-gpg \
--namespace=flux-system \
--from-file=sops.asc=/dev/stdin

Generate a Kubernetes secret manifest and encrypt the secret's data field with sops:

kubectl -n redis create secret generic redis-auth \
--from-literal=password=change-me \
--dry-run=client \
-o yaml > infrastructure/redis/redis-auth.yaml

sops --encrypt \
--pgp=1F3D1CED2F865F5E59CA564553241F147E7C5FA4 \
--encrypted-regex '^(data|stringData)$' \
--in-place infrastructure/redis/redis-auth.yaml

Add the secret to infrastructure/redis/kustomization.yaml:

apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: redis
resources:
  - namespace.yaml
  - release.yaml
  - redis-auth.yaml

Enable decryption on your clusters by editing the infrastructure.yaml files:

apiVersion: kustomize.toolkit.fluxcd.io/v1beta2
kind: Kustomization
metadata:
  name: infrastructure
  namespace: flux-system
spec:
  # content omitted for brevity
  decryption:
    provider: sops
    secretRef:
      name: sops-gpg

Export the public key so anyone with access to the repository can encrypt secrets but not decrypt them:

gpg --export -a [email protected] > public.key

Push the changes to the main branch:

git add -A && git commit -m "add encrypted secret" && git push

Verify that the secret has been created in the redis namespace on both clusters:

kubectl --context staging -n redis get secrets
kubectl --context production -n redis get secrets

You can use Kubernetes secrets to provide values for your Helm releases:

apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
  name: redis
spec:
  # content omitted for brevity
  values:
    usePassword: true
  valuesFrom:
  - kind: Secret
    name: redis-auth
    valuesKey: password
    targetPath: password

Find out more about Helm releases values overrides in the docs.

Add clusters

If you want to add a cluster to your fleet, first clone your repo locally:

git clone https://github.com/${GITHUB_USER}/${GITHUB_REPO}.git
cd ${GITHUB_REPO}

Create a dir inside clusters with your cluster name:

mkdir -p clusters/dev

Copy the sync manifests from staging:

cp clusters/staging/infrastructure.yaml clusters/dev
cp clusters/staging/apps.yaml clusters/dev

You could create a dev overlay inside apps, make sure to change the spec.path inside clusters/dev/apps.yaml to path: ./apps/dev.

Push the changes to the main branch:

git add -A && git commit -m "add dev cluster" && git push

Set the kubectl context and path to your dev cluster and bootstrap Flux:

flux bootstrap github \
    --context=dev \
    --owner=${GITHUB_USER} \
    --repository=${GITHUB_REPO} \
    --branch=main \
    --personal \
    --path=clusters/dev

Identical environments

If you want to spin up an identical environment, you can bootstrap a cluster e.g. production-clone and reuse the production definitions.

Bootstrap the production-clone cluster:

flux bootstrap github \
    --context=production-clone \
    --owner=${GITHUB_USER} \
    --repository=${GITHUB_REPO} \
    --branch=main \
    --personal \
    --path=clusters/production-clone

Pull the changes locally:

git pull origin main

Create a kustomization.yaml inside the clusters/production-clone dir:

apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
  - flux-system
  - ../production/infrastructure.yaml
  - ../production/apps.yaml

Note that besides the flux-system kustomize overlay, we also include the infrastructure and apps manifests from the production dir.

Push the changes to the main branch:

git add -A && git commit -m "add production clone" && git push

Tell Flux to deploy the production workloads on the production-clone cluster:

flux reconcile kustomization flux-system \
    --context=production-clone \
    --with-source 

Testing

Any change to the Kubernetes manifests or to the repository structure should be validated in CI before a pull requests is merged into the main branch and synced on the cluster.

This repository contains the following GitHub CI workflows:

• the test workflow validates the Kubernetes manifests and Kustomize overlays with kubeconform
• the e2e workflow starts a Kubernetes cluster in CI and tests the staging setup by running Flux in Kubernetes Kind