Booklets

A low-effort onramp to Kubernetes

Most Kubernetes documentation is written for people who already run Kubernetes. That’s the gap these booklets are here to close.

If your team is considering Kubernetes — wondering whether the shape of it fits your workloads, your team size, your budget, and your data-sovereignty constraints — start with the introduction booklet. It explains the moving parts in plain English, before any kubectl appears, and is short enough to read in an afternoon.

If you’re ready to run a cluster yourself, the setup booklet walks you from a fresh Hetzner account to a working Jōkamachi cluster, with no prior Kubernetes experience assumed.

If you’re evaluating whether to trust us with production, the architecture booklet is the one to read first — it covers control plane, worker pools, multi-tenancy, and the deliberately boring networking choices that hold a Jōkamachi cluster together.

If you’re comparing the numbers, the performance booklet puts hard numbers behind the technical decisions: when Hetzner’s load balancer is the right call and when a custom NAT gateway pulls ahead, what you actually lose to shared CPU cores versus dedicated, when performance cores are worth paying for. Read this one alongside Architecture if you’re sizing a cluster against a target SLO.

The KAOS booklet is about what goes into agentic cluster management — and starts, unusually, with a chapter titled You don’t need KAOS! We mean it: real operating experience comes from running the cluster yourself, with kubectl in one hand and an LLM CLI in the other. KAOS is what you reach for when you’ve built up enough experience to want to codify it and let an agent act on it under policy. The booklet walks the progression from “are there any incidents?” to “have I seen this before?” to “should this go through GitOps?” and arrives at KAOS as a destination, not a starting point.

The Kubenix booklet covers the GitOps side of the same story — what’s underneath the cluster’s typed primitives, why we use them, and how they make horizontal property testing feasible (which, in turn, is what makes durable agentic changes safe enough to automate).

Why booklets, not blog posts

Each booklet is opinionated, end-to-end, and designed to be read in a sitting. We update them in place — there is no “best of 2023” archive to wade past, no comments thread to read first, no search bar required.

The complete catalogue is below.