Skip to content

$ ls ./blog

Blog

Field notes from production Kubernetes and from fixes merged upstream: real output, real numbers, what broke and why.

RSS feed

Your cluster says 3/3. It will still go dark with one zone.

Manifests describe intent; the cluster holds placement. How kubectl survive-zone proves which workloads stop serving when a zone dies, and why a perfectly spread deployment can still go down.

  • Kubernetes
  • Scheduling
  • Reliability

· 4 min read