Local Kubernetes

Why Develop with Kubernetes

  • Devcontainer + k3d matches the production target, so fewer surprises when shipping.
  • One platform to learn and operate; we avoid maintaining divergent dev/prod stacks.
  • Local clusters let you exercise the same manifests, operators, and ingress as production.
  • Easier to debug: use k9s/kubectl locally with the same permissions and objects you expect in prod.

Run Kubernetes inside the devcontainer with k3d so your local environment matches how we deploy to real clusters. We develop on Kubernetes because production runs there too—keeping dev and prod aligned means fewer surprises and less duplicated infra to manage.

Bootstrap the cluster

Open a terminal inside the devcontainer and recreate the local cluster with the bundled recipe:

just dev-init

This deletes any previous k3d-nails cluster, creates a fresh one with a single agent node, maps the app/Postgres NodePorts to the host (30000-30001), and patches your kubeconfig via just get-config so kubectl works from inside the container.

Install the Nails stack

With the cluster up, install the platform operators and sample StackApp:

just dev-setup

The Justfile in nails-devcontainer/Justfile runs stack init followed by stack install --manifest stack.dev.yaml, giving you a ready-to-use namespace plus database credentials.

Inspect the cluster

Use k9s (preinstalled in the devcontainer) to browse pods, services, and logs after the install completes:

k9s

You can also spot-check with kubectl get pods -A. Re-run just dev-init any time you want to reset the cluster to a clean state.