Image generation with FastAPI and Vue.js 3.x

In this repo you can find how to use ONNX-Runtime docker with FastAPI (as backend) and Vue.js (as frond-end).

There are several main modules:

• Celery broker aka Celery broker (RabbitMQ) (RabbitMQ), docker - rabbitmq;
• Celery backend aka Celery backend (Redis) (Redis), docker - redis;
• Celery worker aka Celery workers (Celery), docker - worker;
• model module aka Model module (ONNX-Runtime), docker - server_model;
• backend aka Server application (FastAPI), docker - server_backend;
• front-end aka UserClient (Vue.js), docker - web_module;

Each module is a docker with own settings and config file. You can observe docker-compose.yml file for more details.

Here is simple diagram for this system:

How to start

Simple solution (aka production )

Run docker-compose file docker-compose.yml with command:

docker-compose -f docker-compose.yml up

If you want test it locally - run docker as it is. If you want to start real server, then you need to modify some variables:

• In server/server_web/app.py, on line 11, you should append your IP address with port of front-end (by default its 21665), aka http://192.186.12.2:21665 Thats it!

After that you need to open two ports: 21665 (for front-end) and 21666 (for backend).

Hard solution (aka develop mode)

For this case, you need start several things before application.

Step 1. Start Celery broker and backend

First is to run broker (RabbitMQ) and backend (Redis):

• Redis: docker run -d -p 6379:6379 redis

• RabbitMQ: docker run -d -p 5672:5672 rabbitmq

Step 2. Start Celery workers

After that, you need to start Celery itself.

Go to folder server/server_web and run workers with Celery:

Linux:

celery -A routers.celery_task_app.worker worker -l info

Windows:

celery -A routers.celery_task_app.worker worker -l info -P gevent

Problem with Celery on Windows

Solution 1

While trying to start Celery on windows, I have some problems with that, but found some solution in the Internet in order to start worker on Windows try next:

Install: pip install eventlet

After that, Celery can be started via command: celery -A routers.celery_task_app.worker worker --pool=eventlet

but there is some issue with this eventlet

Solution 2 (recommended)

Instal: pip install gevent

Run Celery with command: celery -A routers.celery_task_app.worker worker -l info -P gevent

Tested on Windows 10

Step 3. Start Server application

In folder server/server_web update your python environment with requirements.txt file:

pip install -r requirements.txt

In same folder run FastAPI server with command:

uvicorn app:app --host 0.0.0.0 --port 21666

Step 4. Start UserClient

In folder client, run next command:

npm run serve