remote debugging

• Adding print statements to local code or the remote code directly can be rough.
• Take advantage of Python’s support for remote application debugging.

Guide

• Install debugpy on the remote container. This can be used directly on the command-line. It’s especially helpful when running a test script.
• Create a launch.json file in a .vscode folder in your workspace

{
    "version": "0.2.0",
    "configurations": [
        {
            "name": "Python Debugger: Remote Attach",
            "type": "debugpy",
            "request": "attach",
            "connect": {
                "host": "localhost",
                "port": 5678
            },
            "justMyCode": false,
            "pathMappings": [
                {
                    "localRoot": "${workspaceFolder}/vllm",
                    "remoteRoot": "${workspace/path}/vllm"
                }
            ]
        }
    ]
}

• Port-forward your container
• Run your script using debugpy

python3 -m debugpy --listen localhost:5678 --wait-for-client -m vllm.entrypoints.openai.api_server {args}

---

Resources:
VS Code Python Debugger Extension
DebugPy