Troubleshooting

Diagnostics for your own clusters and workloads.

Clusters — a cluster that does not come up, workers that never join, a control plane you cannot reach.

Workloads — storage that will not bind, Services with no address, and the failures that look like Kubernetes bugs but are platform boundaries.

Before you start

Check which cluster you are talking to. Your Cluster object, addresses, quota and backups live in your Environment on the platform's cluster. Your workloads live in your own cluster, behind its own kubeconfig. Most "this command shows nothing" is one of those two.

Know where your visibility ends. Storage backends, provider-network routing and the platform's own components are not things you can see into. A diagnosis that runs out of evidence there is finished, not incomplete — hand it over with the object names and UIDs you have.

For the platform side — the under cluster, the operator, the provider networks — see Platform Administration → Troubleshooting.