今天分享一份代码,使用 Celery、RabbitMQ 和 MongoDB 实现一个异步任务工作流,你可以修改 task.py 来实现你自己的异步任务。
架构图如下:
其中 Celery 来执行异步任务,RabbitMQ 作为消息队列,MongoDB 存储任务执行结果,FastAPI 提供 Web 接口。
以上所有模块均可使用 Docker 一键部署。
下面为 Demo 使用方法:
1、确保本机已安装 Docker、Git
2、下载源代码:
git clone https://github.com/aarunjith/async-demo.git
3、部署并启动:
cd async-demo
docker compose up --build
4、启动一个异步任务:
$ curl -X POST http://localhost:8080/process
任务会发送到消息队列,同时会立即返回一个任务 id:
❯ curl -X POST http://localhost:8080/process
{"status":"PENDING","id":"a129c666-7b5b-45f7-ba54-9d7b96a1fe58","error":""}%
5、查询任务状态:
curl -X POST http://localhost:8080/check_progress/<task_id>
任务完成后的返回结果如下:
❯ curl -X POST http://localhost:8080/check_progress/a129c666-7b5b-45f7-ba54-9d7b96a1fe58
{"status":"SUCEESS","data":"\"hello\""}%
代码目录结构如下:
其中 app.py 如下:
from fastapi import FastAPI
from celery.result import AsyncResult
from tasks import start_processing
from loguru import logger
from pymongo import MongoClient
import uvicorn
# Lets create a connection to our backend where celery stores the results
client = MongoClient("mongodb://mongodb:27017")
# Default database and collection names that Celery create
db = client['task_results']
coll = db["celery_taskmeta"]
app = FastAPI()
@app.post('/process')
async def process_text_file():
'''
Process endpoint to trigger the start of a process
'''
try:
result = start_processing.delay()
logger.info(f'Started processing the task with id {result.id}')
return {
"status": result.state,
'id': result.id,
'error': ''
}
except Exception as e:
logger.info(f'Task Execution failed: {e}')
return {
"status": "FAILURE",
'id': None,
'error': e
}
@app.post('/check_progress/{task_id}')
async def check_async_progress(task_id: str):
'''
Endpoint to check the task progress and fetch the results if the task is
complete.
'''
try:
result = AsyncResult(task_id)
if result.ready():
data = coll.find({'_id': task_id})[0]
return {'status': 'SUCEESS', 'data': data['result']}
else:
return {"status": result.state, "error": ''}
except Exception as e:
data = coll.find({'_id': task_id})[0]
if data:
return {'status': 'SUCEESS', 'data': data['result']}
return {'status': 'Task ID invalid', 'error': e}
if __name__ == "__main__":
uvicorn.run("app:app", host='0.0.0.0', port='8080')
如果要实现自己的任务队列,就修改 task.py 来添加自己的异步任务,可以整合到自己的项目中。