Request对象连接池小于最优值(8)问题求助

我在开发Telegram机器人时遇到如下问题：Request对象的连接池小于最优值(8)。我尝试创建带有更大连接池的Session对象但无效，问题仍存在，已检查所需依赖库，附上相关代码（已移除机器人令牌），恳请提供其他解决思路。

import os
import requests
import telegram
from telegram.ext import Updater, CommandHandler, MessageHandler, Filters

# Replace YOUR_TOKEN_HERE with your actual bot token
bot = telegram.Bot(token='')

# Set up the Updater and Dispatcher
updater = Updater(bot=bot, use_context=True)
dispatcher = updater.dispatcher

# Set up a global variable to store the watermark text
watermark_text = None

# Set up a global variable to store the watermark text opacity (default to 50)
watermark_opacity = 50

def start(update, context):
    """Handler for the /start command"""
    context.bot.send_message(chat_id=update.effective_chat.id, text="Hi! I'm a bot that can add watermark text to videos. Use the /setwatermark command to set the watermark text, and the /opacity command to set the text opacity. Use the /watermark command to watermark a video.")

def set_watermark(update, context):
    """Handler for the /setwatermark command"""
    # Declare the watermark_text variable as global
    global watermark_text
    # Get the watermark text from the user's message
    watermark_text = " ".join(context.args)
    # Save the watermark text to the global variable
    watermark_text = watermark_text
    context.bot.send_message(chat_id=update.effective_chat.id, text="Watermark text set! You can now use the /watermark command to watermark a video.")

def set_opacity(update, context):
    """Handler for the /opacity command"""
    # Declare the watermark_opacity variable as global
    global watermark_opacity
    # Get the watermark text opacity from the user's message
    watermark_opacity = int(context.args[0])

def watermark(update, context):
    """Handler for the /watermark command"""
    # Check if a watermark text has been set
    if watermark_text is None:
        context.bot.send_message(chat_id=update.effective_chat.id, text="No watermark text has been set. Use the /setwatermark command to set the watermark text.")
        return

# Create a Session object with a larger connection pool size
        session = requests.Session()
        adapter = requests.adapters.HTTPAdapter(max_pool_size=8)
        session.mount('https://', adapter)

    # Get the file ID of the video sent by the user
    file_id = update.message.video.file_id
    # Download the video file using the Session object
    file = session.get(bot.get_file(file_id).file_path)
    open('input.mp4', 'wb').write(file.content)

    # Watermark the video using ffmpeg
    os.system(f'ffmpeg -i input.mp4 -vf "drawtext=fontfile=arial.ttf: text={watermark_text}: fontcolor=white: fontsize=48: x=10: y=10: alpha={watermark_opacity}" output.mp4')

    # Send the watermarked video back to the user
    context.bot.send_video(chat_id=update.effective_chat.id, video=open('output.mp4', 'rb'))

# Set up the command handlers
start_handler = CommandHandler('start', start)
dispatcher.add_handler(start_handler)

set_watermark_handler = CommandHandler('setwatermark', set_watermark)
dispatcher.add_handler(set_watermark_handler)

set_opacity_handler = CommandHandler('opacity', set_opacity)
dispatcher.add_handler(set_opacity_handler)

watermark_handler = CommandHandler('watermark', watermark)
dispatcher.add_handler(watermark_handler)

# Start the bot
updater.start_polling()

解决思路 #

• 修复Session的缩进与作用域问题
  你代码里创建Session的部分缩进错误，位于return语句之后，实际不会被执行。而且每次请求新建Session无法复用连接池，应该把Session改为全局单例，在代码开头初始化：

  # 全局初始化Session，复用连接池
  session = requests.Session()
  adapter = requests.adapters.HTTPAdapter(max_pool_size=8)
  session.mount('https://', adapter)
  

  之后在watermark函数里直接使用这个全局Session即可，无需重复创建。

• 给Bot对象绑定自定义Session
  当前仅在下载文件时用了自定义Session，但bot.get_file()本身的请求用的是telegram库默认Session，这部分也会占用连接资源。初始化Bot时传入自定义Session，让所有Bot API请求复用同一个连接池：

  # 先初始化Session，再传入Bot
  session = requests.Session()
  adapter = requests.adapters.HTTPAdapter(max_pool_size=8)
  session.mount('https://', adapter)
  bot = telegram.Bot(token='YOUR_TOKEN_HERE', session=session)
  

• 调整Updater轮询的连接池大小
  start_polling()默认有自己的连接配置，通过pool_size参数设置为匹配的最优值：

  updater.start_polling(pool_size=8)
  

• 验证连接池配置
  可以打印Session适配器的参数确认配置生效：

  print(session.adapters['https://'].max_pool_size)
  

  输出为8则说明连接池设置正确。

内容的提问来源于stack exchange，提问作者Pastel