Erich Blume f3319c753c Update deploy-k8s-service doc with ProxyGroup ingress pattern (#166 )

## Summary
- Updated the "Configure Ingress" section to use the current ProxyGroup pattern (`proxy-group: "ingress"`, `defaultBackend`, `tls.hosts`)
- Replaced the old per-ingress proxy example that used `rules:` with `host:` (which breaks ProxyGroup routing)
- Added key points explaining why `defaultBackend` is required and what each annotation does
- Updated checklist to mention ProxyGroup

## Test plan
- [ ] Review rendered doc for accuracy against existing ingress manifests

Reviewed-on: https://forge.ops.eblu.me/eblume/blumeops/pulls/166

2026-02-11 21:10:42 -08:00

3.3 KiB

Raw Blame History

title

modified

Deploy a Kubernetes Service

Quick reference for deploying a new service to BlumeOps Kubernetes via ArgoCD. See adding-a-service for detailed explanations.

Create Manifests

argocd/manifests/<service>/
├── deployment.yaml
├── service.yaml
└── ingress-tailscale.yaml

Namespace should match service name. Use registry.ops.eblu.me for images.

Create ArgoCD Application

# argocd/apps/<service>.yaml
apiVersion: argoproj.io/v1alpha1
kind: Application
metadata:
  name: <service>
  namespace: argocd
spec:
  project: default
  source:
    repoURL: ssh://forgejo@indri.tail8d86e.ts.net:2200/eblume/blumeops.git
    targetRevision: main
    path: argocd/manifests/<service>
  destination:
    server: https://kubernetes.default.svc
    namespace: <service>
  syncPolicy:
    syncOptions:
    - CreateNamespace=true

Configure Ingress

Add a tailscale-operator routed through the ProxyGroup with Homepage annotations:

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: <service>-tailscale
  namespace: <service>
  annotations:
    tailscale.com/proxy-class: "default"
    tailscale.com/proxy-group: "ingress"
    gethomepage.dev/enabled: "true"
    gethomepage.dev/name: "Service Name"
    gethomepage.dev/group: "Apps"
    gethomepage.dev/icon: "<service>.png"
    gethomepage.dev/href: "https://<service>.ops.eblu.me"
    gethomepage.dev/pod-selector: "app=<service>"
spec:
  ingressClassName: tailscale
  defaultBackend:
    service:
      name: <service>
      port:
        number: 80
  tls:
    - hosts:
        - <service>

Key points:

proxy-group: "ingress" routes through the shared ProxyGroup instead of spawning a per-ingress proxy
Do not use rules: with host: — the ProxyGroup proxy receives the FQDN as Host header (e.g. <service>.tail8d86e.ts.net), so a short host: <service> won't match. Use defaultBackend instead.
tls.hosts sets the MagicDNS hostname (becomes <service>.tail8d86e.ts.net)

Add Caddy Route (if needed)

If other pods need to access the service, add to ansible/roles/caddy/defaults/main.yml:

caddy_services:
  - name: <service>
    upstream: "https://<service>.tail8d86e.ts.net"

Then: mise run provision-indri -- --tags caddy

See routing for when Caddy is needed.

Deploy

# Sync apps to pick up new Application
argocd app sync apps

# Test on feature branch first
argocd app set <service> --revision <branch>
argocd app sync <service>

# Verify
kubectl --context=minikube-indri -n <service> get pods
kubectl --context=minikube-indri -n <service> logs -f deployment/<service>

# After PR merge, reset to main
argocd app set <service> --revision main
argocd app sync <service>

Checklist

Manifests in argocd/manifests/<service>/
Application in argocd/apps/<service>.yaml
Tailscale Ingress via ProxyGroup with Homepage annotations
Caddy route (if pod-to-service access needed)
Tested on feature branch
PR reviewed and merged
Reset to main branch

adding-a-service - Full tutorial with explanations
apps - ArgoCD application registry
routing - Service routing options

3.3 KiB Raw Blame History