# Kubernetes Inventory

The Kubernetes inventory lets Plakar Control Plane connect to a Kubernetes
cluster and discover supported resources. Once connected, the discovered
resources become available for management directly from Plakar Control Plane.

## Supported Resources

| Resource                     | Source | Store | Destination |
| ---------------------------- | ------ | ----- | ----------- |
| PersistentVolumeClaims (PVC) | Yes    | No    | Yes         |
| KubeVirt virtual machines    | Yes    | No    | Yes         |

Virtual machines are discovered only when KubeVirt is installed in the cluster.
On clusters without KubeVirt, the inventory discovers PVCs only.

## Authentication

Kubernetes inventories authenticate using a kubeconfig file. The kubeconfig
determines which cluster Plakar Control Plane connects to and which permissions
it has, so the associated user or service account must have read access to
`PersistentVolumeClaims` in the namespaces you want to discover. To discover
virtual machines, it also needs read access to KubeVirt `VirtualMachines` in
those namespaces. See
[Kubernetes Service Accounts and RBAC](../../../guides/kubernetes/kubernetes-rbac#inventory)
for a least-privilege role.

When creating an Kubernetes inventory in Plakar Control Plane, you must provide:

- **Name**: a name for the inventory.
- **Kubeconfig**: paste the contents of your kubeconfig file directly, or select
  your kubeconfig file to use. You can leave this field empty if Plakar Control
  Plane runs inside the cluster, in which case it uses the in-cluster
  configuration instead. The kubeconfig can also be loaded from a
  [secret provider](../../secret-providers).
- **Namespaces**: namespaces to discover resources from. Leave it empty for the
  inventory to discover resources in every namespace in the cluster.

## Creating a Kubernetes Inventory

When creating a new Kubernetes inventory, provide the inventory name and either
paste in a kubeconfig or select a kubeconfig file. If Plakar Control Plane runs
inside the cluster you want to inventory, you can leave the kubeconfig field
empty and it will use the in-cluster configuration.

After creating the inventory, trigger a synchronization to discover and load
PVCs and virtual machines from the configured cluster. You can run
synchronization again at any time to refresh the inventory, for example after
creating new PVCs or virtual machines.

![Creating Kubernetes inventory](../images/create-kubernetes-inventory.png)