- Docs
- Cluster management
- Troubleshooting
Troubleshooting
Diagnose a cluster that will not come up, an unhealthy API server or etcd, expired certificates, and stuck upgrades
- Cluster creation fails Diagnose why your Syself Autopilot cluster is not coming up, from missing Hetzner secrets and quota limits to stuck machines.
- No servers available Resolve cluster degradation from Hetzner Cloud capacity limits. Move control planes and workers to bare-metal servers to avoid provisioning failures.
- Failed or stuck upgrades A stack or Kubernetes upgrade is not completing, so find what is blocking the rollout and how to recover.
- Control-plane and etcd recovery The API server is flapping or an etcd member is unhealthy, so restore quorum without making it worse.
- Expired certificates kubectl reports an expired certificate, so find out whether your kubeconfig or the API server certificate ran out, and renew the right one.