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

# Using Static Volumes

> Create a PV for existing backend storage and make it available to workloads in a namespace through a PVC.

Static volumes are suitable for scenarios where backend storage already exists or where storage resources must be prepared in advance by an administrator. You need to create a PV for the backend storage first, and then create a PVC to bind to the PV.

## Review Before Proceeding

* A backend storage volume is ready for use by the CKS cluster.
* Information required to create a PV has been obtained, including capacity, access mode, CSI driver name, and a unique volume identifier. The specific fields and values are provided by the storage backend or storage administrator.

## Create PV

1. Log in to the [GPU Cloud console](https://console.scitix.ai/gpu-cloud/home) and go to the **Clusters** page.

2. On the **Clusters** page, click the name of the target cluster to access the cluster workspace.

3. In the left navigation pane, go to the **Storage** page, switch to the **PV** tab, and click **Create**.

   ![Create PV](https://mintcdn.com/scitix-adfc65c2/zAestxkyOFggIEE8/gpu-cloud/cluster/media/cluster-create-pv.png?fit=max\&auto=format\&n=zAestxkyOFggIEE8\&q=85\&s=25a10245c3299f11d0a5a806b63d50bf)

4. Set **Name** and complete the PV configuration in **Spec (YAML)**.

   | Configuration Item | Description |
   | - | - |
   | `capacity.storage` | Sets the capacity provided by the PV. |
   | `accessModes` | Sets the access modes supported by the volume. They must be compatible with subsequent PVC requests. |
   | `persistentVolumeReclaimPolicy` | Sets how the PV is reclaimed after the PVC is released. Select `Retain` if you need to retain the data. |
   | `storageClassName` | Sets the storage class value used to filter and match PVCs. When using static volumes, the values in the PV and PVC should be consistent; this field does not automatically create a PV. |
   | `csi.driver` | Enter the actual CSI driver name. |
   | `csi.volumeHandle` | Enter the unique identifier of the backend volume. |

   The following YAML shows the basic structure of a CSI PV. Replace the placeholders with the actual values provided by the storage backend, and add other fields as required by the storage plugin.

   ```yaml theme={null}
   metadata:
     labels: {}
     annotations: {}
   spec:
     capacity:
       storage: 100Gi
     accessModes:
       - ReadWriteMany
     persistentVolumeReclaimPolicy: Retain
     storageClassName: manual
     volumeMode: Filesystem
     csi:
       driver: <CSI_DRIVER_NAME>
       volumeHandle: <UNIQUE_VOLUME_ID>
   ```

5. Click **Create**.

## Create a PVC and Bind It to a PV

1. On the **Storage** page, switch to the **PVC** tab and click **Create**.

2. Configure the PVC.

   | Parameter | Description |
   | - | - |
   | **Name** | Set the PVC name. Workloads reference the PVC by this name. |
   | **Namespace** | Select the namespace to which the PVC belongs. |
   | **Spec (YAML)** | Enter the PVC `metadata` and `spec`. Do not enter `apiVersion` or `kind`. |

   `volumeName` must be set to the name of the PV created in the previous step. The requested capacity must not exceed the PV capacity, and `accessModes`, `storageClassName`, and `volumeMode` must be compatible with the PV.

   ```yaml theme={null}
   metadata:
     labels: {}
     annotations: {}
   spec:
     accessModes:
       - ReadWriteMany
     resources:
       requests:
         storage: 100Gi
     storageClassName: manual
     volumeName: training-data-pv
     volumeMode: Filesystem
   ```

3. Click **Create**.

4. Return to the **PVC** tab and verify that the PVC status has changed to `Bound`. Then, on the **PV** tab, verify that the target PV is bound to this PVC.

## Next Actions

After the PVC status becomes `Bound`, you can [mount persistent storage to a workload](https://docs.scitix.ai/gpu-cloud/cluster/mount-persistent-storage-to-workload.md). When a PVC or PV is no longer needed, see [Clean up persistent storage resources](https://docs.scitix.ai/gpu-cloud/cluster/cleanup-persistent-storage-resources.md).

## Frequently Asked Questions

> **What should I do if a PVC cannot be bound to the specified PV?**
>
> Ensure that `volumeName` exactly matches the PV name, and verify that the requested capacity, access modes, `storageClassName`, and `volumeMode` are compatible with the PV. The target PV must also be available for binding and must not already be bound to another PVC.


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