Skip to main content

Deploy a Pod

The Pods page provides GPU container rentals and lists all Pods you are currently using, along with their running status, start time, usage duration, and creation time. Deployment takes four steps.

Step 1: Select Resource​

Click Create Pod and choose a suitable GPU resource from the resource cards; you can filter by Brand, Model, or GPU Mode.

Each resource card shows: resource name, specification and compute resources (GPU model, vCPU, vRAM), billing method, GPU mode (shared/dedicated; no label means CPU-only), and a real-time stock indicator:

  • In Stock, green: available to rent
  • Limited Stock, yellow: available to rent, but the resource may be taken by others if you do not deploy promptly
  • Low Stock, red: available to rent, but stock is tight
  • Out of Stock, black: card is grayed out and cannot be rented

If you need Shared Memory, toggle it on and enter a size.

Throughout deployment, the Summary panel on the right shows the pricing (estimated hourly/monthly rate), resource specification, shared memory, image, and storage in real time.

Step 2: Deployment Settings​

  1. Enter a Pod name (lowercase letters, digits, and - only; it must start with a letter and end with a letter or digit).
  2. Choose an image template from one of two sources:
    • Official Templates: image templates published by the administrator
    • Custom Image: supports Docker Hub, GitHub Container Registry, NGC, GCP Artifact, and AWS ECR; see Images and Custom Images
  3. Advanced Settings: if an official template has preset deployment parameters, they are filled in automatically and can be modified:
    • Ports: port name and port value
    • Environment Variables (optional): enter key-value pairs, or click the key icon to use a Secret as an environment variable
    • Start Command (optional)
    • Liveness Probe command (optional)

Step 3: Mount Storage (optional)​

Select an existing volume from the mount list, or click Create Volume to create one on the spot (see Storage for the settings). After selecting a volume, the system fills in the Mount Path automatically.

caution

Billing starts as soon as a volume is enabled (capacity × usage duration), and a volume can only be expanded after creation, never shrunk.

Step 4: Summary and deploy​

The summary page consolidates the resource, image, deployment parameters, storage, and estimated rental cost. After confirming everything, click Deploy.

Back on the Pods page, deployment is complete once the status changes from Pending to Running. For ongoing management, see View and Manage Pods.