[FIXED] How to publish flask web app which runs C++ subprocess

Issue

Current Development Build Architecture:

• Single-Page Flask Web App: c:/code/flask_app/app.py
• C++ Program: c:/code/flask_app/main.exe

I have a fully functional Flask web application that involves a straightforward HTTP POST request from a client to a server (main.js -> app.py). Additionally, it utilizes the Socket.IO WebSocket library to trigger the execution of a subprocess (main.exe), located in the root directory of the Flask app (app.py), based on user-generated data.

During development, this setup worked smoothly without any issues. However, I have encountered several challenges when attempting to deploy the application using Gunicorn on a local Ubuntu/Linux server:

1. Compatibility of main.exe with Linux:

I have observed that the Linux server fails to execute main.exe. I suspect this issue might require either rebuilding main.exe to make it compatible with the Linux operating system or identifying a suitable alternative solution.

2. Socket.IO and Gunicorn Interaction:

Upon deploying the application using Gunicorn, I encountered recurring Socket.IO 400 (BAD REQUEST) errors. I am uncertain whether these errors are caused by the nature of main.exe or if they are a result of Socket.IO's compatibility with Gunicorn.

3. Performance Concerns with Gunicorn:

Additionally, I have noticed that the application experiences severe performance degradation when launched via Gunicorn. I am curious to know if the Socket.IO errors contribute to this laggy behavior or if there might be other factors at play.

Given the above challenges, I would appreciate any guidance or insights on the best and most straightforward approach to publish my web application, which I have developed over the past five weeks. I am open to suggestions on addressing the compatibility issues with main.exe, resolving Socket.IO-related concerns, or optimizing the application's performance on Gunicorn.

Solution

Compatibility of main.exe with Linux

That’s exactly where your problem begins from. The machine generated code in the main.exe cannot run on Linux OS because it was compiled for Windows OS.

Also while recompiling for Linux, watch out for any platform dependent code for windows. Such might not be compatible with Linux or might produce very underperforming code (this is from my experience).

Performance Concerns with Gunicorn

Gunicorn has limitations in its load balancing algorithm, gunicorn can only be used with one worker process so that might lead to bottlenecks also.

Answered By - Chukwujiobi Canon