Debugging Operator - NVIDIA Docs

NVIDIA Docs Hub NVIDIA Clara Clara Holoscan Deploy 0.8.1 EA Debugging Operator

Remote Debugging Python Application with VSCode

Visual Studio Code (VSCode) supports Remote Debugging through Python extension.

Using the feature, bundled segmentation operators are configured to support debugging while developing/testing operators locally.

(sdk folder)/clara-reference-app/Operators/ai includes debug_app_docker.sh script to support debugging AI operator with VSCode.

Copy
Copied!

            
            USAGE: ./debug_app_docker.sh [-p <number>] [app_name] [model_name]

  -p <number>           Remote debug port number to use (default: 3000)
  -h                    Help message

Args
  app_name   - application name (e.g., app_livertumor)
  model_name - model name (e.g., segmentation_liver_v1)

Examples
  ./debug_app_docker.sh                                                              # Livertumor AI (default)
  ./debug_app_docker.sh app_braintumor segmentation_mri_brain_tumors_br16_t1c2tc_v1  # Braintumor AI
  ./debug_app_docker.sh app_vnet                                                     # Organ Segmentation AI

Executing ./debug_app_docker.sh would launch a VSCode window that app.py file is open. Breakpoints can be toggled by clicking on the editor margin or using F9 on the current line (see https://code.visualstudio.com/docs/editor/debugging#_breakpoints).

Since launch.json file (in .vscode subfolder) is already configured, you can start your debug session with F5. Please see Debug actions for more information.

Pressing Ctrl + C in the terminal or closing the VSCode window would terminate the app and dependent Docker containers/network.