Constant error leads to reaching rate-limit

[guettli]

/kind bug

failed to create HCloud server my-wl-cluster-control-plane-5n58s: server location disabled (resource_unavailable, 8df3d4a02ed0770d42deaaa585873d74)

This makes the controller call Reconcile again and again. Which leads to reaching the rate-limit.

Related code line in createServer() in server.go

	server, err := s.scope.HCloudClient.CreateServer(ctx, opts)

This limit was reached:

exceeded hcloud rate limit with calling function "ListLoadBalancers"

In this case it was the first machine in the cluster. I guess this does not happen for other machines.

Related: Use custom Exponential backoff - to avoid rate-limit · Issue #1757 · syself/cluster-api-provider-hetzner

[Archisman-Mridha]

I'm facing the same issue currently. The LB gets created, but not the 1st server : it complains server location disabled.

Am using CAPH v1.0.7.

What does that server location disabled error mean btw? Do you know what's causing it?

[guettli]

@Archisman-Mridha see: Hetzner Cloud Changelog

[Archisman-Mridha]

Identified the issue. CAX11 servers were not available in the FSN1 region.

Thanks for your time 🫡.

[guettli]

In the future the server type and the location will be in the error message, so that things are easier to understand.

But currently, Reconcile still return an error which make the controller try again and again every fast, so that the rate limit will be reached soon.