furry-train

furry-train

Simple web app with a status endpoint

Used mainly to test the setup of a k3s cluster on raspberryPi 4 (Ubuntu server 64bit)

Checkable at https://api.shipperizer.org/api/v0/status

shipperizer@xxx 9:36:57 π  http https://api.shipperizer.org/api/v0/status
HTTP/1.1 200 OK
Content-Length: 25
Content-Type: text/plain; charset=utf-8
Date: Fri, 09 Apr 2021 08:43:37 GMT
Vary: Accept-Encoding

{
    "message": "Purple Bro"
}

Build and deploy

Build setup is for multiarch support, a requirements for this is buildx

For skaffold integration i followed the suggestion here as ther eis no direct integration between skaffold and buildx

Images are pushed to ghcr.io/shipperizer/furry-train-web-app, k3s cluster has a secret allowing it to pull them, see the snippet below in deployments.yaml

containers:
- image: ghcr.io/shipperizer/furry-train-web-app
  name: furry-train-web-app
  envFrom:
    - configMapRef:
        name: furry-train-web-app
  name: furry-train-web-app
  ports:
  - name: http
    containerPort: 8000
imagePullSecrets:
- name: regcred-github

Kaniko

For kaniko builds, use the --profile kaniko modifier on skaffold, for this you will need an Opaque secret:

 echo '{"auths":{"ghcr.io":{"auth":"****************"}}}' | kubectl create secret generic regcred-github-kaniko --from-file=config.json=/dev/stdin

the profile is targeted at building on an arm64 cluster only, if you need to use a different arch change initImage and image values

ArgoCD

ArgoCD is used (together with ArgoCD image updater) to keep application up-to-date

see the argocd-app.yaml for extra informations local setup will be described eventually in here, step by step

Future improvements

Plan is to setup argocd and argocd image updater to pull images and update the deployment files, skaffold will be used only for local/manual deployment