Seven years.
Dozens of clusters.
One playbook.

I've managed more than a hundred Kubernetes clusters in production. I've been the person at 2AM with a pager, staring at kubectl describe output, trying to figure out why checkout is down and the release is blocked.

After years of running twenty commands to understand one problem — mentally mapping service dependencies, guessing at blast radius, praying the scheduler would change its mind — I started building tools to make the invisible visible. The posts on this site are a byproduct of that work.

Every post comes from a real incident. No theory. No "according to the docs." Just what actually happened, what I tried, what failed, and what finally worked.

"Pending" doesn't mean broken. It means the scheduler is looking at every node and saying nope, nope, nope.

What I build

Kubilitics — a Kubernetes operating system I've been working on for the last two years. Cluster health scores, topology visualization, risk ranking, blast-radius simulation. It's the dashboard I wanted at 2AM.

troubleshoot-kubernetes-like-a-pro — an open-source repo with 35 real K8s failure scenarios. Deploy a broken config, debug it, apply the fix. The lab behind every post.