Question

我写了一个基于flask的深度学习服务，为了提高效率，许多会话和模型都加载到全局变量中。某些预测较慢，因此我将gevent的WSGIServer用于非阻塞访问。但是担心这会导致某些会话同时影响TensorFlow，从而导致错误的结果。那么有什么方法可以确保线程安全？

import os
from flask import Flask, request
from flask_cors import *
import json

import numpy as np

import tensorflow as tf

from gevent import monkey
from gevent.pywsgi import WSGIServer
monkey.patch_all()

app = Flask(__name__)
CORS(app, supports_credentials=True)


# load model and create global prediction session
    # some codes



@app.route('/service1', methods=['POST'])
def service:
    # some codes using the global session


if __name__ == '__main__':
    app.config['JSON_AS_ASCII'] = False
    http_server = WSGIServer(('0.0.0.0', 5000), app, keyfile='*****.key',
                             certfile='******.pem')
    http_server.serve_forever()

如何使用gevent的WSGIServer确保全局变量的安全性

0 个答案: