Autoscaling for Hetzner Cloud

Add autoscaling and self-healing to the Hetzner servers you already run. No Kubernetes. No migration. Your traffic never touches us.

Connect read-only14-day Observe trial · read-only token

group: webrange 2–6 × cx33

example timeline

web-1
healthy
web-2
healthy
web-3
healthy

02:59
no actionall 3 nodes healthy; p95 CPU 31%

01Read-only 14-day Observe trial, enforced by Hetzner's own read-only token.
02No node receives traffic before it passes the load balancer health check.
03Your account, your bill, no per-server fee.
04Works with plain Docker or cloud-init, Coolify, and any webhook or CI.
05Replaces a dead node at 03:00 and adds one at 15:00, and explains both.

How it works

Four modes. Start read-only, move up when the report convinces you.

The mode decides whether Hetscale may call the Hetzner API. It never changes how a decision is made, or how it is explained.

[observe] web-2 unhealthy for 3 min
[observe] would: create web-4 beside web-2
[observe] would cost: +1 node-hour, billed by Hetzner
[observe] no API write; token is read-only

The trial mode. Hetscale holds a Hetzner token with read permission only, so it cannot create or delete anything even by mistake. After 14 days you get a report of every decision it would have taken.

[rehearsal] create web-4 from template
[rehearsal] deploy via adapter → wait for LB health
[rehearsal] web-4 active → drain → delete
[rehearsal] passed; Live unlocked

A supervised dry run before Live: Hetscale creates one node from your template, deploys through your adapter, waits for the load balancer health check, drains it and deletes it while you watch. Success unlocks Live; failure produces a report and never leaves a node behind.

[propose] create web-5 (cx33)
[propose] reason: CPU 78% > 75% for 3 min
[propose] waiting for approval…
[propose] approved → executing

Each action is proposed with its reason and waits for approval in the dashboard or by email link; unapproved proposals expire.

[live] web-2 unhealthy for 3 min
[live] create web-4 beside web-2
[live] web-4 active → LB target
[live] web-2 kept for diagnosis; 3 healthy

Hetscale creates, health-checks, attaches, drains and deletes nodes within the minimum and maximum you set. Moving back to a safer mode takes effect immediately.

Calculator

What does a fixed fleet cost you at night?

Server type€8.49/mo eachServers today4Load at night, share of peak35%

fleet with Hetscale: 2–4 × cx33
saving, per month: €7–10

Assumes 8–12 low-load hours a day, a baseline of at least 2 nodes, and Hetzner's net list price at Falkenstein, 2026-09-17. The full calculator takes your real load curve.

Open the calculator

Fit

When to pick something else

Your workload keeps state on the node: databases, local queues, sticky sessions.
You already run Kubernetes and its autoscaler does the job.
Your servers are not on Hetzner Cloud.
Load is flat. One server at a fixed size is cheaper than any control plane.

Kubernetes for many services or complex scheduling; Cloud 66 for greenfield with one vendor; Laravel Cloud for zero-ops on AWS; over-provisioning when traffic is stable.

Find your starting point: the onboarding cases