Ephemeral CI Runners
Deploys now run as throwaway pods inside the cluster instead of on hosted runners reaching in from outside — one pod per job, scaled to zero when idle, so CI costs nothing between deploys.
Deploys now run as throwaway pods inside the cluster instead of on hosted runners reaching in from outside — one pod per job, scaled to zero when idle, so CI costs nothing between deploys.
A tour of the platform underneath this site — compute, networking, TLS, storage, observability, and the security posture — on a small but real multi-workload Kubernetes cluster.
How the cluster API is protected while still letting CI deploy to it — the Control Plane ACL dance, supply-chain hardening, and where secrets live.
What happens on every push to main: open the cluster ACL, build the Hugo site, push the image, apply the manifest, roll out, and close the ACL again.
An overview of how this static site is built and served from a small Linode Kubernetes cluster — and the other apps that now share the same repo and pipeline.
How traffic flows from the internet down to the nginx pod serving this page — NodeBalancer, Traefik, and a deliberately minimal container image.