Writing
Notes on production Kubernetes for teams that can't justify a platform team. Every piece maps to one of the five ways it actually hurts: cost, deploys, ownership, security, upgrades.
-
ownership
The Kubernetes checklist for teams without a platform teamThe minimum viable discipline for production — Git as the only door, the rollout basics, the boring security baseline, tested restores, scheduled upgrades — and the enterprise baggage to refuse.
-
triage
The first 10 things I'd check in a messy Kubernetes clusterHanded an undocumented cluster? A 90-minute, read-only triage: version support, Git-vs-reality diff, the RBAC five-minute audit, the bus-factor interview. Three artifacts out.
-
cost
8 Kubernetes cost leaks small teams missHalf of teams saw costs go up after adopting Kubernetes. The leaks are in the defaults: requests copied from tutorials, logs kept forever, one load balancer per service, and a bill nobody owns.
-
deploys
Why Kubernetes rolling updates still give you 503sSIGTERM and load-balancer removal happen in parallel, not in sequence. The four stacked causes of deploy-time errors, their fixes, and the fifteen-minute test everyone skips.
-
audit
The Kubernetes audit worksheetAssess your own cluster in half a day. Eight sections, every command read-only, ending in a risk map, a cost opportunity in currency per month, and a ten-item remediation backlog.