Adopt your servers

Adoption is PaaSbox’s flagship move: you hand over Hetzner servers you already own, PaaSbox reinstalls them in place and runs them as Kubernetes worker nodes — and never deletes them, so the grandfathered price you locked in before the 2026 increase stays locked in. Adopt, don’t create.

Which servers qualify

• Grandfathered means ordered before 15 June 2026, 08:00 CEST — Hetzner’s price-adjustment cutoff. A server ordered before then keeps its old rate for as long as that exact instance lives and is never rescaled to another type.
• Any Hetzner Cloud server in a project you’ve connected can be adopted; the console lists your servers and marks the grandfathered ones, with the monthly saving each would preserve.
• Adopting newer (post-increase) servers works too — there’s just no price to rescue; you’d normally use a Managed pool for those instead.

Prerequisites: a connected Hetzner project (read/write API token) — that’s it. Adoption itself is optional and doesn’t need a card; running a cluster with the adopted servers does.

What adoption does — read this first

Adoption wipes the server

Each adopted server is reinstalled (OS rebuild). All data on it is destroyed. Migrate your workloads off first. The server is never deleted, so its grandfathered price survives. The console makes you type WIPE to confirm — there is no one-click path past this.

Rebuild-in-place is precisely why the price survives: Hetzner bills rescales and new orders at current prices, but a rebuild keeps the same server instance — same ID, same rate, empty disk. Adoption is an OS reinstall, not a certified data-destruction service — if you need certified erasure, handle that before adopting.

The flow

1. Select servers — from your connected project’s server list, multi-select the ones to adopt. The console shows type, location, architecture, and the projected monthly saving per server.

2. Type WIPE — the confirmation restates what happens: reinstall, data destroyed, never deleted.

3. Watch the phases — each server steps through Enrolled → Wiping → Ready. A rebuild typically takes a few minutes per server, and they proceed in parallel.

4. Use the fleet — Ready servers form your fleet, grouped by server type and location. Create a Pooled node pool over them in any cluster, grow it by adopting more, or release servers back to standalone use.

The price-safety matrix

Only two Hetzner operations forfeit a grandfathered rate:

Action	Price effect
Rescale (change_type)	✗ re-priced — Hetzner bills rescales at current rates
Delete + recreate	✗ re-priced — a new server is a new order

Everything else is price-safe: power on/off and reboot, rename, labels, rebuild / OS reinstall (this is how PaaSbox re-provisions without losing the price), rescue mode, snapshots and backups, mounting ISOs, public-IP assign/unassign, private-network attach/detach, floating IPs, protection toggles, and moving the server to another project. Billing continues while a server is powered off — you own the box.

Two consequences worth knowing:

• PaaSbox only ever uses price-safe operations on adopted servers. Upgrades, node repairs, and reprovisioning are rebuilds in place — never delete-and-recreate, never a rescale.
• On-demand Managed nodes are ordinary cattle: they’re deleted and recreated freely (for example around hibernation), always at current prices. Your adopted servers are unaffected.

Never-delete is structural

“We never delete your server” isn’t a policy promise to trust — it’s how the system is built:

• The platform component that manages adopted servers has no delete path for them. Releasing a server from a cluster, deleting the cluster, unenrolling the server — all of these stop management of the machine; none of them destroy it. Even in payment dunning, pooled servers are never deleted.
• Hetzner delete protection is enabled on adopted servers as a second, independent guard at the infrastructure level.
• The read/write API token you connect is needed for Managed pools; the pooled path’s never-delete guarantee does not depend on token scope.

What it’s worth

A single grandfathered cpx42 keeps ~€52/month off your bill; four of them, ~€209/month — every month, surviving every upgrade. The grandfathered pricing page has the full price table and the savings estimator, and the FAQ covers what happens to adopted servers when you delete a cluster (they return to your fleet, price intact).