A second ago your dev loop was instant — now every code change costs you 2–5 minutes.
Sound familiar? You finished a service, it runs great on your laptop, all that's left is to "build the container and deploy." And then it begins: docker build → docker push → kubectl apply → wait for the Pod to come up → check the logs. Instead of uvicorn --reload in a fraction of a second, you wait minutes for every little tweak. And on top of that, bugs you literally cannot see on a laptop: OOMKilled, CPU throttling, a NetworkPolicy cutting your traffic, a failing readiness probe, RBAC denying access. Code that "worked locally" — and dies in the cluster for reasons you never even considered.
This is the first article in a new series on local development for Kubernetes. I dig into where TWO distinct pains come from — the slow loop AND the low-fidelity environment — and why shift-left isn't "test earlier," it's "test in a real cluster." Using the end-to-end example myapp (FastAPI + PostgreSQL), I show what we'll build by the end of the series: a local k3d cluster, real manifests, and Tilt with live-update that brings iteration back down to seconds.
If you're getting your first service ready to deploy to Kubernetes — start here. https://dorokhovich.com/blog/local-k8s-inner-dev-loop?utm_source=devto&utm_medium=syndication&utm_campaign=local-k8s-inner-dev-loop
Top comments (0)