allans.dev

Why Most Kubernetes Migrations Fail on Day Two

Mon, 22 Jun 2026 00:00:00 +0000

Most Kubernetes migrations don’t fail during the migration. They fail about two weeks after, when the first real incident hits and the team realizes nobody actually decided how things work day-to-day on the new platform.

The pattern

It usually goes like this:

A team spends months getting workloads running on Kubernetes.
Day one goes fine — traffic is cut over, dashboards are green.
Two to three weeks later, a pod starts crash-looping at 2am, or a node gets cordoned during a deploy, or a HorizontalPodAutoscaler does something nobody expected under load.
Nobody can answer “who owns this” or “what’s the rollback” with confidence, because those questions were never answered before go-live.

The cluster isn’t the hard part. The hard part is everything that has to be true around the cluster for it to be operable by a team that didn’t build it.

About

Mon, 01 Jan 0001 00:00:00 +0000

Background

I’m a Senior Platform/SRE/DevOps Engineer with hands-on experience running Kubernetes, CI/CD, and cloud infrastructure in production. I spent 4.5 years at Globoplay, one of Latin America’s largest streaming platforms, working on infrastructure supporting high-traffic, high-availability workloads. I now work independently, partnering remotely with engineering teams across the US and other markets on platform and infrastructure work.

My core stack is Kubernetes, Docker, Terraform, Python, Bash, and CI/CD (GitLab CI, GitHub Actions) — tools I use daily and trust under pressure. I work regularly with Argo CD, HashiCorp Vault, and AWS KMS for GitOps and secrets management, and I’m building depth in Helm, Dynatrace, Backstage, and data layers like MongoDB and Redis. I’m also actively learning Istio and Go, and I don’t pretend otherwise — I’d rather be clear about what’s production-tested versus what’s still developing.

Contact

Mon, 01 Jan 0001 00:00:00 +0000

The fastest ways to reach me:

LinkedIn: linkedin.com/in/as-enginer
Email: allangoeng@gmail.com

No contact form here — email or LinkedIn works best.

Projects

Mon, 01 Jan 0001 00:00:00 +0000

A selection of work related to Kubernetes, bare metal cloud, and CI/CD platform engineering will be listed here.

This page is a placeholder — case studies and write-ups are in progress.