对memcaced的python客户端测试

在线上发现python-memcached客户端很有问题,我在本地通过pthon manager runserver测试,刷新用户首页就有大量的TIME_WAIT:

 

yunpeng@yunpeng-duitang:/duitang/dist/conf/redis$ netstat -an | grep 11211
tcp        0      0 0.0.0.0:11211           0.0.0.0:*               LISTEN    
tcp        0      0 127.0.0.1:53169         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53315         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53293         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53161         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53306         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53308         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53283         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53262         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53163         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53284         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53139         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53140         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53264         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53157         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53167         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53173         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53177         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53209         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53159         127.0.0.1:11211         TIME_WAIT 
tcp        0      0 127.0.0.1:53203         127.0.0.1:11211         TIME_WAIT

 

经测试发现django在每次请求请求之后都会关闭socket,导致大量TIME_WAIT,试着把memcache.py的disconnect_all注释掉,大量TIME_WAIT消失。

yunpeng@yunpeng-duitang:/duitang/dist/conf/redis$ netstat -an | grep 11211
tcp        0      0 0.0.0.0:11211           0.0.0.0:*               LISTEN    
tcp        0      0 127.0.0.1:11211         127.0.0.1:53352         ESTABLISHED
tcp        0      0 127.0.0.1:53352         127.0.0.1:11211         ESTABLISHED

 

但是我去线上机器测试发现又不太一样。

1. gunicorn开启一个进程

刷新个人主页,memcached连接数

[admin@server1 ~]$ netstat -an | grep "ESTABLISHED" |grep 11211 | wc -l
10

[admin@server1 ~]$ netstat -an | grep "TIME_WAIT" |grep 11211 | wc -l
236

注释掉disconnect_all:

[admin@server1 duitang]$ netstat -an | grep "TIME_WAIT" |grep 11211 | wc -l
322
[admin@server1 duitang]$ netstat -an | grep "ESTABLISHED" |grep 11211 | wc -l
11

 

2.使用uwsgi一个单个进程,2线程模式

[admin@server2 duitang]$ netstat -antp| grep 11211 | grep -v java | grep ESTABLISHED | wc -l
20

[admin@server2 duitang]$ netstat -antp| grep 11211 | grep -v java | grep TIME_WAIT | wc -l
0

 

你可能感兴趣的:(python)