Skip to content

Stopping and Starting Workspaces

Workspaces can be stopped to free up resources temporarily and started again when you’re ready to resume work. This feature is especially useful for intermittent workloads where you need to pause processing but want to keep your environment configuration intact.

1. Stopping a Workspace

  1. Open the Workspaces Dashboard

    • In the left navigation menu, select Workspaces.
    • Find the Workspace you wish to stop in the list.

    Workspace Dashboard

  2. Click “Stop”

    • Within the Actions section at the bottom of the Workspace card, look for the Stop Icon icon to Stop the desired Workspace.
    • A pop-up screen will appear to Confirm your decision. The platform will gracefully shut down the Workspace pod or container, freeing compute resources.

  3. Verify Stopped Status

    • The Workspace’s status in the dashboard changes to something like “Stopped” or “Paused.”
    • All running processes in that Workspace are terminated; however, any Checkpointing or persistent volumes attached to this Workspace will retain their data.

Note: If you haven’t enabled Checkpointing, ephemeral session data in memory will be lost. Ensure you’ve saved any critical work to a persistent volume (PVC) before stopping.

2. Starting a Workspace

  1. Locate the Stopped Workspace

    • Return to the Workspaces dashboard.
    • Filter by Stopped status or simply search for a Workspace card that displays a Stopped status.

  2. Click “Resume”

    • In the Actions section of the Workspace card, select the Resume Icon Resume Icon.
    • The platform will allocate the previously configured resources (GPU, CPU, Memory) and bring the environment back online.

  3. Wait for Initialization

    • The Workspace status transitions to Not Ready indicating the environment is being re-provisioned.
    • Once it shows Ready, you can reconnect, launch Jupyter/VS Code, and continue from where you left off—particularly if checkpointing or persistent storage is in place.

3. Best Practices

  • Checkpointing: Enable checkpointing for critical sessions so you can resume exactly where you left off.
  • Persistent Volumes: Store vital data and notebooks in persistent volumes to avoid data loss when the workspace is stopped.
  • Resource Optimization: Stopping Workspaces you aren’t actively using helps free GPU/CPU/Memory resources for other workloads.

4. Troubleshooting

  1. Workspace Won’t Stop
  2. Check if any finalization processes are active (e.g., saving logs). In some cases, you may need to force-stop if the environment is unresponsive.
  3. Workspace Won’t Start
  4. Ensure the cluster has enough free resources to re-provision the workspace (GPU, CPU, memory).
  5. Check for scheduling errors in the platform logs if the environment remains in “Pending” state.

With these steps, you can effectively pause and resume your Workspace sessions as needed, maximizing resource usage and ensuring critical data remains protected.