Skip to content

Managing Storage Volumes

Overview
Storage volumes, also known as Persistent Volume Claims (PVCs), provide workloads with persistent storage that outlasts the lifecycle of individual containers or pods. Typical examples include Jupyter Notebooks retaining user data, databases storing files, or shared workspaces in a multi-user environment.

Why Use Storage Volumes (PVCs)?

  • Persistence: Data remains intact even if the workload restarts or the node reboots.
  • Scalability: Volumes can be sized and resized (if supported) to match growing or shrinking requirements.
  • Flexibility: Different storage classes (e.g., NFS, block storage) can be used to meet specific performance or availability needs.

Access the Storage Volume Dashboard

  1. Log In
    • Sign in to the GPU Cluster Manager as a user.

  2. Open “Storage” Dashboard

    • In the left navigation bar, select the Storage Icon Storage Icon.
    • The lone Volumes tab displays a table of existing volume:

    Storage Volume Dashboard

Dashboard Elements

The dashboard provides a clear overview of all your storage volumes in a table format, along with tools to manage them.

Volume Management Controls

  • Search Volumes Field: Easily find a specific volume by typing its name into this search bar.
  • All Projects Dropdown: Filter the list of volumes to show only those belonging to a specific project you select.
  • + New Volume Button: Click this button to begin the process of creating a brand new storage volume for your projects.

Volume List Table

Each row in the table represents one of your storage volumes and displays the following key information:

  • Name: The unique name you've given to your storage volume (e.g., vol-blue).

  • Project: The project that this volume is associated with (e.g., proj-blue).

  • Status: The current state of the volume (e.g., Pending, Ready, Bound). A volume must be Ready or Bound to be used.

  • Volume: This column typically shows an internal identifier or related Kubernetes PersistentVolume name, linking it to the underlying storage resource.

  • Request Storage: The amount of storage space you initially requested for this volume (e.g., 10GiB).

  • Capacity: The actual size of the storage volume that has been provisioned (e.g., 0B in this example, indicating it's not yet fully provisioned for the Pending status).

  • Access Mode: How the volume can be mounted and accessed by your workloads (e.g., local-path).

  • Storage Class: The type of storage infrastructure used for this volume (e.g., <unset> or a specific storage class name).

  • Volume Attributes Class: Additional attributes or a classification for the volume.

  • Volume Mode: Specifies whether the volume is for a Filesystem (typical for most uses) or Block storage.

  • Age: How long the storage volume has existed since its creation (e.g., 4m 39s).

  • Actions: Icons to perform quick actions on the volume:

  • Edit Icon (Pen): Allows you to modify certain properties of the volume.

  • Delete Icon (Trash Can): Permanently removes the storage volume. Use with caution, as this will delete all data on the volume.

Pagination Controls

  • At the bottom right, you'll find navigation arrows and a selector (e.g., 1 / 10 / page) to browse through pages of volumes if you have many, and to adjust how many volumes are displayed per page.

Dashboard Benefits

  • Create Storage: Easily provision new persistent volumes for your AI workloads.
  • Monitor Volume Status: Keep an eye on whether your volumes are ready, pending, or experiencing issues.
  • Review Volume Details: Check the requested size, actual capacity, and access modes of your storage.
  • Manage Volumes: Edit or delete volumes as needed.

Key Operations

  1. Create a Volume

    • Define the Name, Project association, Storage Class, Requested Storage, and Access Mode.
    • This is typically done via the + New Volume dialog.
    • See the detailed Creating a Volume documentation for step-by-step instructions.

  2. Delete a Volume

    • Remove a PVC that is no longer required to free up storage capacity.
    • Ensure that no workloads are actively using the volume to prevent data loss or application errors.
    • Check the Deleting a Volume section for more details and best practices.

Best Practices

  • Align volume size and performance characteristics with your workloads to optimize cost and resource usage.
  • Use the appropriate Access Mode (e.g., ReadWriteOnce vs. ReadWriteMany) to match how your workload will access the data.
  • Monitor disk usage and storage performance regularly to ensure your projects have enough capacity and performance headroom.