Limits & fair use
The flat fee covers a standard footprint. This page says what that is, what else is bounded and why, and what happens when you reach a limit.
The included footprint
Section titled “The included footprint”| Per cluster | Included |
|---|---|
| Worker nodes | up to 10 |
| Pods | about 500 |
| Attached volumes | about 50 |
The same for a test and a production cluster. These are the sizes at which the control plane we run for you stays inside its standard footprint; pods and volumes are what actually drive its size, node count matters mostly through the pods it carries.
Larger clusters are welcome. Above the footprint we look at the cluster together and add a footprint step to the fee — the same three numbers again. The console will show your footprint and the step price with it; until then, a larger cluster is a short conversation, not a blocker.
Limits that exist for technical reasons
Section titled “Limits that exist for technical reasons”| Limit | Value | Why |
|---|---|---|
| Pods per node | 10 to 500, default 110 | kubelet setting; more pods need a wider node range |
| Volumes per node | 16 | Hetzner attaches at most 16 volumes to one server |
| Placement group size | 10 servers | Hetzner’s spread placement group limit |
| Minimum pool size | 1 | the smallest pool is one node; hibernate the cluster or remove the pool instead |
| Zones per cluster | 1 | control plane and nodes live in one Hetzner location |
| IPv4 addresses | Hetzner’s per-project quota | private clusters use one or two addresses per project, whatever the fleet size |
Fair use on the account
Section titled “Fair use on the account”- Clusters per team — a soft cap before a quick review, to protect platform capacity. New accounts start with three clusters until the first invoice clears.
- Spend caps — €200 a month on a new account, raised to €1,000 and €5,000 with billing history. At the cap, awake clusters are hibernated, never deleted. See billing & spend controls.
- Adopted servers — bounded by the fleet you enroll; there is no platform limit.
What happens at a limit
Section titled “What happens at a limit”The composer refuses the specific over-limit change and says why. A running cluster is never touched by a limit: nothing is scaled down, hibernated or deleted because you grew. Where a review is needed, we contact you; where a Hetzner quota is the limit, the console shows the Hetzner error and the request goes to Hetzner support.