A node group is servers short

A node group is “Short of capacity” when Hetzner has no stock of its server type anywhere Hetscale may create it and the fallback chain is used up, so it runs fewer servers than it should, counting those on their way. The servers that run keep running; the missing ones wait. That shortfall is what Needs you, the “Capacity gap”, the mails and the incident report count. Your plan, a euro ceiling, the cost guard and your organization’s Node cap can also hold a new server; those holds show on the node group’s timeline, not as a shortfall.

What does the panel show?

• Needs you on the dashboard lists the group as “Short of capacity”, with how many servers it is short.
• Capacity & cost shows “Stock of in ”, with “Stock needs a look” when it does, the “Capacity gap”, servers short times minutes over the period, and every create attempt with its outcome, such as “Out of stock (412)” or “Skipped: out of stock”.
• You get a mail when a shortfall starts, if it lasts and when it ends, and an incident report on the timeline once it is over.

Is the server type out of stock?

When Hetzner has no stock of the type where Hetscale tries to create a server, Hetscale tries the other locations of the load balancer’s network zone, then, in a fallback mode, the types you allowed. When all of them are out, the group waits: Hetscale checks every minute and adds servers as soon as stock returns. If Hetzner no longer sells the type in the zone, no stock returns: the alert says so, and only a fallback mode or a new seed adds servers. To let other types fill in, choose “Until the group’s type is back, try the types I pick, in order” or “Until the group’s type is back, use the cheapest fitting type in stock” under “When the type is out of stock” in the group’s Settings. Capacity & cost also shows what one more server always on, a higher Min, would have covered and what it costs. The whole chain is on When your Hetzner server type is sold out.

Did Hetzner refuse a server for an account limit?

Then the group is not short but paused, with “Hetzner account limit reached”. Ask Hetzner to raise the limit in the Console, then resume the group: A node group is paused.

Is your plan holding new servers?

A new server above your plan’s capacity waits when it would pass the next plan’s capacity (on Team, its own capacity plus half), when your organization’s profile says “Never scale above my plan’s capacity”, when the month’s burst budget is spent and its seven days have passed, or after a downgrade, until your servers fit the new plan. A replacement for a failed server and a Rehearsal never wait for the plan, and the group’s timeline records the holds. Move up a plan on the billing page, or turn that profile setting off. How capacity and the burst budget work is on Pricing.

Has the euro ceiling been reached?

With a Euro ceiling in the group’s Settings, once the servers Hetscale created this month have cost that much at list prices, Hetscale creates no more until the month ends, replacements included; running servers keep running. The timeline records the hold and you get an alert. Raise the ceiling or clear it.

The cost guard, Cost guard · servers created within a window, also holds new servers, replacements included, until its window frees a slot; the timeline records it and you get an alert. Your organization’s Node cap, set on its profile, is the most servers Hetscale may run for you across all node groups and projects, and new servers wait at it too.