Using Flux with Kustomize

Scenario and Goals

The following example makes use of Flux's experimental feature for factoring out manifests using Kustomize (and other such tooling, in theory).

For this example we assume an scenario with two clusters, staging and production. The goal is to levarage the full functionality of Flux (including automatic releases and supporting all fluxctl commands) to manage both clusters while minimizing duplicated declarations.

staging and production are almost identical, they both deploy a podinfo service. However, we have different requirments for each cluster:

  1. We want automated deployments for staging but not for production since we want a rubber-stamp every change. However, we want to still be able to make the changes with fluxctl.
  2. Since we expect production to have a higher load than staging, we want a higher replica range there.

How to run the example

In order to run this example, you need to:

  1. Deploy Flux version 1.13.0 or newer.

  2. Make sure to pass the flag --manifest-generation=true to fluxd, in its container spec.

  3. Fork this repository and add the fork's URL as the --git-url flag for the fluxd container.

  4. Pick an environment to run (staging or production) and ask Flux to use that environment by passing flag --git-path=staging or --git-path=production

  5. As usual, you need to make sure that the ssh key shown by fluxctl identity is added to the your github fork.

How does this example work?

├── .flux.yaml
├── base
│   ├── demo-ns.yaml
│   ├── kustomization.yaml
│   ├── podinfo-dep.yaml
│   ├── podinfo-hpa.yaml
│   └── podinfo-svc.yaml
├── staging
│   ├── flux-patch.yaml
│   └── kustomization.yaml
└── production
    ├── flux-patch.yaml
    ├── kustomization.yaml
    └── replicas-patch.yaml
  • base contains the base manifests. The resources to be deployed in staging and production are almost identical to the ones described here.
  • the staging and production directores make use of base, with a few patches, to generate the final manifests for each environment:
    • staging/kustomization.yaml and production/kustomization.yaml are Kustomize config files which indicate how to apply the patches.
    • staging/flux-patch.yaml and production/flux-patch.yaml contain environment-specific Flux annotations and the container images to be deployed in each environment.
    • production/replicas-patch.yaml increases the number of replicas of podinfo in production.
  • .flux.yaml is used by Flux to generate and update manifests. Its commands are run in the directory (staging or production). In this particular case, .flux.yaml tells Flux to generate manifests running kustomize build and update policy annotations and container images by editing flux-patch.yaml, which will implicitly applied to the manifests generated with kustomize build.

I want to know more

Check the documentation of .flux.yaml files