python之requests urllib3 连接池

0.目录

1.参考

2. pool_connections 默认值为10，一个站点主机host对应一个pool
　　(4)分析
　　host A>>host B>>host A page2>>host A page3
　　限定只保留一个pool（host），根据TCP源端口可知，第四次get才能复用连接。

3. pool_maxsize 默认值为10，一个站点主机host对应一个pool， 该pool内根据多线程需求可保留到某一相同主机host的多条连接
　　(4)分析
　　多线程启动时到特定主机host的连接数没有收到 pool_maxsize 的限制，但是之后只有min(线程数，pool_maxsize ) 的连接数能够保留。
　　后续线程(应用层)并不关心实际会使用到的具体连接(传输层源端口)

1.参考

【转载-译文】requests库连接池说明

Requests' secret: pool_connections and pool_maxsize

Python - 体验urllib3 -- HTTP连接池的应用

　　通过wireshark抓取包：

　　所有http://ent.qq.com/a/20111216/******.htm对应的src port都是13136，可见端口重用了

2. pool_connections 默认值为10，一个站点主机host对应一个pool

(1)代码

#!/usr/bin/env python
# -*- coding: UTF-8 -*

import time
import requests
from threading import Thread

import logging

logging.basicConfig()
logging.getLogger().setLevel(logging.DEBUG)
requests_log = logging.getLogger("requests.packages.urllib3")
requests_log.setLevel(logging.DEBUG)
requests_log.propagate = True

url_sohu_1 = 'http://www.sohu.com/sohu/1.html'
url_sohu_2 = 'http://www.sohu.com/sohu/2.html'
url_sohu_3 = 'http://www.sohu.com/sohu/3.html'
url_sohu_4 = 'http://www.sohu.com/sohu/4.html'
url_sohu_5 = 'http://www.sohu.com/sohu/5.html'
url_sohu_6 = 'http://www.sohu.com/sohu/6.html'

url_news_1 = 'http://news.163.com/air/'
url_news_2 = 'http://news.163.com/domestic/'
url_news_3 = 'http://news.163.com/photo/'
url_news_4 = 'http://news.163.com/shehui/'
url_news_5 = 'http://news.163.com/uav/5/'
url_news_6 = 'http://news.163.com/world/6/'

s = requests.Session()
s.mount('http://', requests.adapters.HTTPAdapter(pool_connections=1))
s.get(url_sohu_1)
s.get(url_news_1)
s.get(url_sohu_2)
s.get(url_sohu_3)

(2)log输出

DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): www.sohu.com              #host A
DEBUG:urllib3.connectionpool:http://www.sohu.com:80 "GET /sohu/1.html HTTP/1.1" 404 None
DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): news.163.com　　　　　　　　 #host B
DEBUG:urllib3.connectionpool:http://news.163.com:80 "GET /air/ HTTP/1.1" 200 None
DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): www.sohu.com　　　　　　　　 #host A　　
DEBUG:urllib3.connectionpool:http://www.sohu.com:80 "GET /sohu/2.html HTTP/1.1" 404 None  #host A page2
DEBUG:urllib3.connectionpool:http://www.sohu.com:80 "GET /sohu/3.html HTTP/1.1" 404 None  #host A page3

(3)wireshark抓包  https过滤方法？用tcp syn？ ping m.10010.com 然后 tcp.flags == 0x0002 and ip.dst == 157.255.128.111

(4)分析

host A>>host B>>host A page2>>host A page3

限定只保留一个pool（host），根据TCP源端口可知，第四次get才能复用连接。

3. pool_maxsize 默认值为10，一个站点主机host对应一个pool， 该pool内根据多线程需求可保留到某一相同主机host的多条连接

(1)代码

def thread_get(url):
    s.get(url)
def thread_get_wait_3s(url):
    s.get(url)    
    time.sleep(3)
    s.get(url) 
def thread_get_wait_5s(url):
    s.get(url)    
    time.sleep(5)
    s.get(url)
    
s = requests.Session() 
s.mount('http://', requests.adapters.HTTPAdapter(pool_maxsize=2))
t1 = Thread(target=thread_get_wait_5s, args=(url_sohu_1,))
t2 = Thread(target=thread_get, args=(url_news_1,))
t3 = Thread(target=thread_get_wait_3s, args=(url_sohu_2,))
t4 = Thread(target=thread_get_wait_5s, args=(url_sohu_3,))
t1.start()
t2.start()
t3.start()
t4.start()
t1.join()
t2.join()
t3.join()
t4.join()

(2)log输出

DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): www.sohu.com    #pool_sohu_connection_1_port_54805
DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): news.163.com  　 #pool_163_connection_1_port_54806
DEBUG:urllib3.connectionpool:Starting new HTTP connection (2): www.sohu.com    #pool_sohu_connection_2_port_54807
DEBUG:urllib3.connectionpool:Starting new HTTP connection (3): www.sohu.com    #pool_sohu_connection_3_port_54808
DEBUG:urllib3.connectionpool:http://news.163.com:80 "GET /air/ HTTP/1.1" 200 None
DEBUG:urllib3.connectionpool:http://www.sohu.com:80 "GET /sohu/3.html HTTP/1.1" 404 None
DEBUG:urllib3.connectionpool:http://www.sohu.com:80 "GET /sohu/2.html HTTP/1.1" 404 None
DEBUG:urllib3.connectionpool:http://www.sohu.com:80 "GET /sohu/1.html HTTP/1.1" 404 None
WARNING:urllib3.connectionpool:Connection pool is full, discarding connection: www.sohu.com  #pool_sohu_connection_1_port_54805 被丢弃？最初host sohu能够建立3条连接，之后终究只能保存2条？？？
DEBUG:urllib3.connectionpool:http://www.sohu.com:80 "GET /sohu/2.html HTTP/1.1" 404 None     #pool_sohu_connection_2_port_54807 3秒后sohu/2复用了原来sohu/2的端口
DEBUG:urllib3.connectionpool:http://www.sohu.com:80 "GET /sohu/3.html HTTP/1.1" 404 None     #pool_sohu_connection_2_port_54807 5秒后sohu/3复用了原来sohu/2的端口
DEBUG:urllib3.connectionpool:http://www.sohu.com:80 "GET /sohu/1.html HTTP/1.1" 404 None     #pool_sohu_connection_3_port_54807 5秒后sohu/1复用了原来sohu/3的端口

(3)wireshark抓包

(4)分析

多线程启动时到特定主机host的连接数没有收到 pool_maxsize 的限制，但是之后只有min(线程数，pool_maxsize ) 的连接数能够保留。

后续线程(应用层)并不关心实际会使用到的具体连接(传输层源端口)