Skip to main content

Next steps

Inspect 1.36

You have a cluster and a running app. From here, the path splits into understanding the platform you are now running and growing what you run on it.

Understand the platform#

The section explains what you just used: how a cluster stack bundles the node image, topology, and add-ons into one tested unit, how the management and workload clusters divide responsibility, and how the whole system stays declarative instead of drifting. Read it before you put anything you depend on in production.

Know how upgrades work#

Syself Autopilot upgrades clusters by replacing nodes one at a time, not by patching them in place. covers how a rollout proceeds, what protects your workloads during one, and what to set up before your first upgrade lands.

Work through the security checklist#

The node under your cluster is hardened by default: an integrity-checked base image, a host firewall on from the first boot, a seccomp profile (a filter that blocks a container from making dangerous system calls) for every container, and AppArmor (a Linux feature that restricts what a program on the node can do) on non-privileged containers. Your pods are not hardened yet. On a fresh cluster any pod can reach any other, and namespaces start unrestricted. Work through the and on your namespaces before you put real traffic on them.

Run a real service instead of renting one#

The Deployment and Service from the previous page are a demo. If you want to stop paying a cloud vendor for a managed database, cache, or message queue, run it yourself on the cluster with an operator that handles replication and failover for you. sets up PostgreSQL with CloudNativePG as a starting point; the same pattern applies to other stateful services.

Growing beyond one cluster#

Running separate dev and staging clusters alongside production, and operating a fleet of clusters for multiple clients, are both common next steps. We are still writing up the dedicated guides for both; until then, the section and the how-to guides cover the building blocks you need for either.

You're set up#

You went from nothing to a highly available Kubernetes cluster with a real app running on it, end to end. Everything above builds on that: pick the section that matches what you are about to do next.

Just evaluating? The cluster keeps billing until you remove it. When you are done, to stop the meter.

Related#

Deploy your first app

Deploy a sample application and expose it with a Hetzner Load Balancer using kubectl. See the external IP appear and access your app from the internet.

Overview

Syself Autopilot runs managed upstream Kubernetes on your own Hetzner servers. The two-cluster model, cluster stacks, what Syself operates, what stays yours, and where to read next.