Your first server¶

This walks the full lifecycle with the hcloud CLI: register a key, create a server, watch the action, reach it, and delete it. It works against either backend: on the stub the server is a bookkeeping record; on a libvirt host it's a real cloud-init VM you can SSH into.

Point the CLI at your endpoint first:

export HCLOUD_TOKEN=…                              # your token
export HCLOUD_ENDPOINT=https://api.example.com/v1  # or http://localhost:8000/v1

1. Register an SSH key¶

Your key is installed for root on the VM (matching Hetzner).

hcloud ssh-key create --name me --public-key-from-file ~/.ssh/id_ed25519.pub

2. Create a server¶

hcloud server create --name web1 --type small --image debian-13 --ssh-key me

Like Hetzner, this returns immediately with an action; the VM boots in the background. The CLI polls the action to completion for you. See How it works for the model.

3. Find its address¶

hcloud server list
hcloud server ip -6 web1        # → e.g. 2001:db8:abcd:100:0:1:0:1

Each server gets a unique public IPv6 (see IPv6 networking).

4. SSH in¶

ssh root@$(hcloud server ip -6 web1)

macOS

A bare IPv6 literal may need ssh -6 <addr>, or add AddressFamily inet6 to ~/.ssh/config; a client-side getaddrinfo quirk. If your machine has no IPv6 at all, jump through the host: ssh -J root@<host> root@<vm-ipv6>.

5. Power actions & cleanup¶

hcloud server reboot web1
hcloud server poweroff web1
hcloud server poweron web1
hcloud server delete web1

Power operations are idempotent (powering on a running server is a success no-op), and rebuild --image <name> re-images from the catalog.

Stub vs. real

Against the stub backend, steps 3–4 return an allocated address but there's no VM to reach. To boot a real, reachable VM, deploy to a Linux host: Deploying to a host.