Festus Owumi walks through his project of building a lightweight version of Kubernetes in Go. He removed etcd (replacing it with in-memory storage), skipped containers entirely, dropped authentication, and focused purely on the control plane mechanics. Through this process, he demonstrates how the reconciliation loop, API server concurrency handling, and scheduling logic actually work at their most basic level.

You will learn:

How the reconciliation loop works - The core concept of desired state vs current state that drives all Kubernetes operations
Why the API server is the gateway to etcd - How Kubernetes prevents race conditions using optimistic concurrency control and why centralized validation matters
What the scheduler actually does - Beyond simple round-robin assignment, understanding node affinity, resource requirements, and the complex scoring algorithms that determine pod placement
The complete pod lifecycle - Step-by-step walkthrough from kubectl command to running pod, showing how independent components work together like an orchestra

Sponsor

This episode is sponsored by StormForge by CloudBolt — automatically rightsize your Kubernetes workloads with ML-powered optimization

More info

Find all the links and info for this episode here: https://ku.bz/pf5kK9lQF
Interested in sponsoring an episode? Learn more.

Building Kubernetes (a lite version) from scratch in Go, with Owumi Festus

KubeFM

Building Kubernetes (a lite version) from scratch in Go, with Owumi Festus