Kubernetes Upgrades Without Downtime

The piece outlines a low-risk way to upgrade Kubernetes without downtime: review release notes, test in a production-like non-prod cluster, scan for deprecated APIs, and add PodDisruptionBudgets.

It recommends upgrading control plane components first, then moving nodes in small batches while watching key signals like the API server and autoscaler.

Rather than leaning on rollback tactics, the article argues for catching risky changes early and scheduling upgrades only in a controlled weekday window.

The core message is that smooth Kubernetes upgrades depend on disciplined operations, not just a one-time checklist.