Skip to content
longbridge docs
Open Console →

Quickstart

Go from nothing to a working cluster with one joined node in about five minutes — create a control plane, grab the kubeconfig, attach a machine, and deploy.

This walks you from zero to a running cluster with one node attached. Budget about five minutes, most of which is a machine booting.

You’ll need: an early-access account, the lb CLI, kubectl, and one machine you can install Talos Linux on (a spare PC, a cloud VM, or a VM on your laptop). The machine only needs outbound internet — no public IP, no port-forwarding.

1. Create a control plane

Give it a name and a size. Sizes are described in Sizes & pricing; small is the right place to start.

lb create cluster home --size small --region eu-hel

In a few seconds you have a managed control plane. Pull its kubeconfig:

lb kubeconfig home > ~/.kube/longbridge-home
export KUBECONFIG=~/.kube/longbridge-home
kubectl get nodes

You’ll see an empty cluster — no nodes yet. That’s next.

2. Join a node

Ask the control plane for a short-lived join token:

lb node token home

Boot your machine from the Talos image and apply a worker config that points at your cluster’s endpoint with that token. The CLI prints the exact config for your cluster:

lb node config home --token <token> > worker.yaml
talosctl apply-config --insecure --nodes <machine-ip> --file worker.yaml

The node dials out to the control plane and registers itself — NAT and home networks just work. Within a minute:

kubectl get nodes
# NAME        STATUS   ROLES    AGE   VERSION
# home-1      Ready    <none>   40s   v1.33.x

New to attaching machines? Join a node has the per-platform detail for homelab, Hetzner, and AWS.

3. Deploy something

It’s a normal cluster. Deploy with the tools you already use:

kubectl create deployment hello --image=nginx
kubectl expose deployment hello --port 80
kubectl get pods -w

That’s it — a managed control plane, your own node, one kubeconfig.

Where to next