FAQ
Do I need paasbox’s clusters at all?
Section titled “Do I need paasbox’s clusters at all?”No. paasbox is first an open-source CLI: paasbox garden up gives you your own Gardener landscape in your own Hetzner project, for free, and you create clusters from it yourself. See your own Gardener. The paid clusters on this page and the rest of these docs exist for people who do not want to run a landscape. I run them on my own landscape, best effort.
Is the console live? Can I start today?
Section titled “Is the console live? Can I start today?”Yes. Sign up at console.paasbox.com. The
quickstart takes you from token to kubectl get nodes in about ten minutes.
What is the difference between a test cluster and a production cluster?
Section titled “What is the difference between a test cluster and a production cluster?”The control plane. A test cluster (€49/month) runs one etcd and one API server. If the platform node underneath restarts, management pauses for a few minutes and cluster state is restored from backups taken every five minutes. A production cluster (€99/month) runs three etcd members and three API servers over three platform nodes, so a platform node can fail without a pause, and it carries a monitoring stack. Everything else is the same: backups, kubeconfig, node pools, private clusters, your own identity provider. Your apps keep running on your workers either way. Full comparison: what’s included.
Which regions can I use?
Section titled “Which regions can I use?”nbg1 (Nuremberg) is live. fsn1 and hel1 show “Register interest” and open at sustained
demand. See regions for the current list and how to choose.
Do I keep full cluster-admin?
Section titled “Do I keep full cluster-admin?”Yes. paasbox issues a real admin kubeconfig for your cluster. You
deploy and manage your workloads yourself with kubectl, Helm or Flux. The console handles
cluster lifecycle, fleet and billing; it is not an in-cluster UI.
Do you resell Hetzner compute?
Section titled “Do you resell Hetzner compute?”No. Your worker nodes, load balancers and volumes live in your own Hetzner Cloud project, and Hetzner bills you for them directly. paasbox is software that provisions and operates your cluster on your own cloud. It never resells infrastructure or compute.
Is this multi-cloud?
Section titled “Is this multi-cloud?”No, Hetzner Cloud only. The whole point, your servers and your grandfathered prices, is Hetzner-specific.
Why does the Hetzner token need read/write?
Section titled “Why does the Hetzner token need read/write?”On-demand (“managed”) node pools create and delete servers in your project, and Hetzner API tokens cannot exclude delete, so read/write is required for that path. The grandfathered (“pooled”) path never deletes. That guarantee is a code invariant plus Hetzner delete protection, not a matter of token scope. We say this plainly when you connect your account, and security & trust covers how the token is stored and what it can reach.
What happens to a server when I “adopt” it?
Section titled “What happens to a server when I “adopt” it?”Adoption brings the server under Kubernetes by rebuilding it: an OS reinstall, which wipes the disk. Move your workloads off first; the console makes you confirm this explicitly. After adoption, the server is rebuilt in place and never deleted, which is how it keeps its grandfathered price. Full flow and the price-safety table: adopt your servers.
Can I mix grandfathered and on-demand nodes?
Section titled “Can I mix grandfathered and on-demand nodes?”Yes. A cluster can have a pooled pool over your adopted servers, your cheap baseline, and a managed pool of on-demand nodes for load peaks. The autoscaler prefers your pooled nodes and only adds on-demand nodes when the pool is full. How the two behave: node pools.
What happens if paasbox shuts down?
Section titled “What happens if paasbox shuts down?”You get six months’ notice and migration help at no charge, and you never needed us to leave in the
first place: your cluster’s backups and keys are escrowed in a bucket in your own project, and the
open-source paasbox CLI (paasbox garden) restores your control plane there while your workloads
keep serving. Details, including the recreation path without Gardener: leaving paasbox.
What happens if I delete a cluster?
Section titled “What happens if I delete a cluster?”The cluster is torn down and the per-cluster charge stops, prorated. Your adopted servers are not deleted. They are your property, keep their grandfathered rate, and are ready for another cluster.
Do you back up my data?
Section titled “Do you back up my data?”We operate and back up the control plane: your cluster’s etcd state, continuously. Your application data, databases, volumes and object storage, is yours to back up, the same as on any cluster where you hold admin. The exact split is in backups & availability.
Is there a free trial?
Section titled “Is there a free trial?”Not as such. There is a €5 signup credit (one per team), with a card required up front. Nothing is charged until you start a cluster, and on the hourly plan the credit covers about 33 awake hours of a test cluster. Hibernate when you are done and the meter reads €0. That is the cheapest way to try the paid clusters under real conditions.
What happens if I hit my spend cap?
Section titled “What happens if I hit my spend cap?”You are warned as you approach it, and at the limit your awake clusters are hibernated, never deleted. State, volumes and adopted servers all survive. Raise the cap, or wait for the month to roll over, and wake them. Details in billing & spend controls.
Which Kubernetes versions can I run?
Section titled “Which Kubernetes versions can I run?”New clusters offer the two current upstream minor versions, 1.36 and 1.35 (as of 2026-09-06). You pick the version when composing the cluster. Version upgrades are rolled out by the platform; adopted servers are upgraded by rebuild in place, so their price is never at risk.
Can I use paasbox without grandfathered servers?
Section titled “Can I use paasbox without grandfathered servers?”Yes. A cluster with only managed (on-demand) pools is a normal setup: you still get the operated control plane, autoscaling pools, hibernation and the flat per-cluster fee. Adopting grandfathered servers is the largest saving, not a requirement.