> ## Documentation Index
> Fetch the complete documentation index at: https://docs.spitzli.cloud/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect your Kubernetes cluster

> Enroll an existing approved cluster and manage permitted workloads.

This integration requires operator enablement, an available cluster slot and approved namespaces. You must control the existing Kubernetes or k3s cluster. Registration does not install Kubernetes or migrate workloads.

**1. Confirm permission**

Ask your operator to confirm cluster access, the cluster-count limit and allowed namespaces. Permission to manage existing applications is separate.

**2. Register the cluster**

Where enabled, open **Your infrastructure → Connect own cluster**. Enter the requested name, cluster type, location and location evidence. A declared region alone is not verification.

**3. Enroll its agent**

Open the protected enrollment flow and follow the generated instructions on the intended cluster. If enrollment expires, request a new one. Keep setup credentials out of chat and source control.

**4. Check inventory**

Wait for the agent to connect. Review the observation time, completeness and unavailable resource types before acting on the inventory.

## Workload actions

| Resource | Actions available when authorized |
| - | - |
| Deployment, StatefulSet | Scale, stop, start, restart and logs |
| DaemonSet | Restart and logs |
| Pod, Job | Logs |
| CronJob | Suspend and resume |

Start restores the replica count recorded by a previous Spitzli Cloud stop. Controller-managed resources or unsupported action combinations can be blocked. The integration does not expose arbitrary manifest editing or a shell.

System namespaces remain protected. Changes require a fresh inventory observation and current resource identity. If the state is stale, refresh it and review the workload before retrying.

## Follow operations and disconnect

Follow the returned operation to its final result. A busy or uncertain operation can block another write to that workload; an operator must reconcile uncertain results before changes resume.

Revoking the agent ends access through Spitzli Cloud. It does not delete the cluster or stop independently running workloads. Keep provider access, backups and recovery procedures with the infrastructure owner.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.