A node group is paused

A paused node group keeps its servers running and taking traffic, but Hetscale decides nothing and starts nothing new for it. Except under an emergency stop, a server that was already leaving still finishes, so it stops costing money. The State card on the node group’s Overview names the reason and leads to its heading on this page; Needs you on the dashboard names it too, with a “?” that leads there. Whatever the reason, a member resumes a paused node group with Resume, beside Controls on its Overview, except under an emergency stop, and Hetscale then decides for the group again from the next minute.

Paused by a member

A member chose Pause in the group’s Controls, for example while a rollout of your own runs. Resume it with Resume, beside Controls.

Emergency stop

A member chose Emergency stop: no server is created or deleted for the group until the stop is lifted, and its servers keep running and taking traffic. Resume is not offered while the stop stands; choose Lift the emergency stop, which also runs the group again.

New servers never became ready

Three new servers in a row failed to become ready, at the deploy or at the load balancer’s health check, so Hetscale stopped instead of burning through servers. The group’s timeline shows where each one stopped, and A new server never becomes ready lists the usual causes with the fix for each. Fix the cause, then Resume, which starts the count again. Each Rehearsal starts it again too, and new servers that fail while healing waits out a fault every server shares (What if every server fails at once?) do not count.

Hetzner answered with an error

Hetzner refused a call in a way that trying again does not fix: it no longer accepts the project’s token, the node group’s load balancer is gone from Hetzner, the snapshot of the current release is gone, or Hetzner gave another such answer to a create, a routing or a delete. The group’s timeline names the error. Replace the project’s token when Hetzner no longer accepts it (Replace a project’s token); when the snapshot is gone and the node group deploys from the seed’s snapshot, choose Take a new release on its Deployment page; then Resume. A server protected against deletion does not pause the group: only its removal fails, with an alert. What Hetscale does with each Hetzner answer is on Hetzner API errors and what Hetscale does.

Adapter version not supported

The adapter met a version of your deployment tool that it does not support and paused instead of guessing. No adapter checks a version today: a seed Coolify deploys is cloned like any other seed, and Hetscale calls no Coolify API (Coolify). The planned mode in which Coolify deploys each new server will check Coolify’s version; when it pauses a group, move to a supported version, then Resume.

Hetzner account limit reached

Hetzner refused a new server because a limit of your Hetzner account was reached, for example its Primary IPs (public IPv4 and IPv6 addresses, load balancers included); you also get a mail. When Hetzner’s answer names the limit, the pause on the node group’s timeline and the mail name it, such as its servers, vCPUs, dedicated vCPUs or Primary IPs; a limit Hetscale has no words for is shown under Hetzner’s own name. New Hetzner accounts start with low server and vCPU limits; the Limits page of the Hetzner Console shows yours. Ask Hetzner to raise the limit in the Console, or free what it counts, then Resume. Servers without public addresses take no Primary IP (Servers without public addresses), and the quickstart explains that limit before Go Live. Which limits a node group meets, and how to raise them, is on Hetzner limits that shape a node group.

The plan ended

This is not a pause. When your organization’s plan lapses and its grace ends, node groups keep their state and Hetscale only observes: it creates, drains and deletes nothing, and removes nothing. The billing page says so, and you get a mail. Choose a plan on the billing page, and writing resumes once it applies. The grace is described on Modes.

Restored from the Archive

A Live node group comes back from the Archive paused, because its servers may have changed in Hetzner while nothing read them. Check its servers, then Resume.

The project’s token was replaced

The project’s token had stopped working and a member replaced it, so every Live node group of the project came back paused: Hetzner may have changed while nothing read it. Check the group’s servers, then Resume.