Skip to main content

Managing Instances

Once the instance is created, you can navigate to Manage Instances to access:

  • Overview: User can see the Instance, Plan and Connection Details under the Overview tab

  • Node Events: Entry logs for seeing the instance specific action like Starting Container, Pulling Image

NodeEvents

  • Monitoring: The Monitoring tab provides real-time insights into workspace, storage, and memory usage. It also tracks GPU utilization, memory consumption, temperature, and power usage to help users monitor the performance of their instance

  • Workspace size: You can view the current disk size and modify it based on your requirements. To update, adjust the disk size value and click the Update button.

  • Associated Datasets: In this section, users can attach datasets to the instance in both mounted and unmounted modes. After updating a dataset, the instance must be restarted for the changes to take effect.

  • Shared File-System: It is built to deliver simple, scalable, and highly available file storage, allowing multiple instances to access the same file system simultaneously for seamless shared storage. For this the SFS must be mounted to a path.

  • Network and Security: Under the Network & Security tab, users can configure SSH keys, attach or detach a reserved IP, manage Security Groups, and attach custom scripts to the Node. When a Security Group is added, the corresponding custom ports are automatically configured. The attached script ensures seamless execution by installing and configuring all required dependencies for optimal performance.

  • Images: Here, you can save the instance image to the container registry. Saving an image allows you to preserve the current environment (packages, configurations, and code) so it can be reused, replicated, or restored later.

Instance Actions​

  • Launch App: This provides an interactive workspace where you can manage files, write and execute code, and create documentation. Example Use Case- If a user wants to verify the GPU resources of their launched H100 series instance, they can run the command "!nvidia-smi" inside a Python3 kernel.

LaunchApp

  • Save Image: The Save Image option lets you capture the current state of your instance (packages, configs, and code) for reuse and reproducibility. To save an image, you must have a container registry created beforehand

  • Restart: Restarting clears the current session while preserving the saved data and settings, ensuring instance runs with the updated state.

  • Update Plan: After clicking the Update Plan button, the user is redirected to the Update Node Resource page, where they can choose the desired GPU/CPU plan for their Instance Image. Once they click Update, the instance will enter a Waiting status, and shortly after, it will be updated with the new resources and return to Running status.

Note

For updating the Instance node should be in the stopped state. Click on the Stop Instance from Action button.

Delete: When instance is deleted, all the resource associated with it will be deleted, including the worskpace(disk).