> ## 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.

# Clean up persistent storage resources

> Delete unused PVCs and PVs, and handle backend data according to the reclaim policy for dynamically or statically provisioned volumes.

When persistent storage used by a workload is no longer needed, you can clean up the PVC and PV. Before cleanup, make sure that the workload has stopped using the PVC, and determine whether the backend data will be retained or deleted based on the reclaim policy.

## Pre-Action Confirmation

* The workloads using the target PVC have been stopped, or the corresponding `volumeMounts` and `volumes` configurations have been removed from the workload YAML.
* It has been confirmed that the target PVC, PV, and backend data are no longer needed, or the data has been backed up.

> [!WARNING]
>
> After a PVC is deleted, workloads can no longer access data through that PVC. When the reclaim policy is `Delete`, the CSI driver may also delete the PV, backend storage, and data. If you need to retain the data, first confirm the reclaim policy and data backup status.

## Understand the Impact of Cleanup

| Volume Type | Cleanup Target | Impact |
| - | - | - |
| Dynamic volume | Delete the PVC. | After the PVC is deleted, the handling of the dynamically provisioned PV and backend data depends on the reclaim policy and CSI driver behavior. |
| Static volume | Delete the PVC first, then handle the PV as needed. | After the PVC is deleted, the PV is unbound from the PVC. Whether to retain the PV and backend data depends on the reclaim policy and the administrator's data disposal requirements. |

## Clean up dynamic storage volumes

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

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 and switch to the **PVC** tab.

4. Locate the target PVC and click **Delete**.

5. Verify the PVC information and confirm deletion.

6. Switch to the **PV** tab and verify the processing result for the corresponding PV.

   * If the PV has been deleted, the PV object has been fully cleaned up.
   * If the PV is retained, confirm how the backend data should be handled before processing the PV and the data.

## Clean Up Static Volumes

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

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

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

4. Locate the target PVC and click **Delete**.

5. Switch to the **PV** tab and check the status and reclaim policy of the original PV.

6. Process PVs according to the data disposal method:

   * If the data needs to be retained, retain the PV or have an administrator reprovision it.
   * When the PV object is no longer needed, ensure that the backend data is no longer needed or has been backed up, then delete the target PV.

## Frequently Asked Questions

> **Why does the PV still exist after the PVC is deleted?**
>
> Whether a PV is automatically deleted depends on the reclaim policy and storage plugin behavior. When the reclaim policy is `Retain`, the PV and backend data are retained and usually require further action by an administrator. When the reclaim policy is `Delete`, Kubernetes requests that the storage plugin handle the corresponding storage; the actual outcome depends on the plugin's capabilities and configuration.


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