Kubernetes · DevOps

Kubernetes rollouts without panic at 2 a.m.

By Seungmin Oh ·

Hero image for article: Kubernetes rollouts without panic at 2 a.m.

Docker & Kubernetes Practicum includes a deliberate bad rollout lab: learners deploy a broken readiness probe and watch traffic drain incorrectly. The debrief covers maxSurge, maxUnavailable, and when to pause rollouts.

Seungmin Oh shares runbooks from education workloads — smaller than enterprise scale but honest about probe tuning. We stress observability: if you cannot see pod readiness, you cannot trust a rollout.

Helm values per environment are reviewed like application code. Typos in `values-staging.yaml` have caused more incidents than application bugs in our alumni reports.

Graduates leave with a rollback checklist taped metaphorically to their repo README — a habit hiring managers notice.

← Back to blog