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

# Kubernetes Provider

> Manage Kubernetes workloads and configuration on GKE clusters

The Kubernetes provider manages core Kubernetes resources — Deployments, StatefulSets, Services, ConfigMaps, and Secrets — on GKE clusters using your GCP credentials.

## Setup

The Kubernetes provider connects to GKE clusters via the `gcp/gke` resource. All Kubernetes resources require a `cluster` dependency pointing to a GKE cluster.

## Resources

| Resource                                       | Type                     | Description                                                   |
| ---------------------------------------------- | ------------------------ | ------------------------------------------------------------- |
| [Deployment](/providers/kubernetes/deployment) | `kubernetes/deployment`  | Stateless workloads with rolling updates                      |
| StatefulSet                                    | `kubernetes/statefulset` | Stateful workloads with persistent storage                    |
| Service                                        | `kubernetes/service`     | Network exposure for pods (ClusterIP, LoadBalancer, NodePort) |
| ConfigMap                                      | `kubernetes/configmap`   | Configuration key-value storage                               |
| Secret                                         | `kubernetes/secret`      | Sensitive data storage with base64 encoding                   |

## Architecture

Kubernetes resources depend on a GKE cluster:

```
gcp/gke (cluster)
  ├─▶ kubernetes/deployment
  ├─▶ kubernetes/statefulset
  ├─▶ kubernetes/service
  ├─▶ kubernetes/configmap
  └─▶ kubernetes/secret
```

## Notes

* All Kubernetes resources use server-side apply for idempotent create/update operations.
* The `cluster` and `namespace` fields are immutable on all resources — changing them requires recreation.
* Workload resources (Deployment, StatefulSet) wait for all replicas to be ready before reporting success.