The default Linux kernel parameters and nginx parameter to consider is the most common scenario, is not suitable for high concurrency
so optimizing plan consists of two levels: the system level + nginx level
First, the system level
1, while adjusting the number of open files
the ulimit -n 65535
2, the maximum number of the TCP (SOMAXCONN) connected
echo 10000> / proc / SYS / NET / Core / SOMAXCONN
. 3, the TCP connection immediately recovered, reuse (recycle, reuse )
echo 1> / proc / SYS / NET / ipv4 / tcp_tw_reuse
echo 1> / proc / SYS / NET / ipv4 / tcp_tw_recycle
4, TCP flood not resist
echo 0> / proc / sys / net / ipv4 / tcp_syncookies
//修改配置
ulimit -n 65535
echo 10000 > /proc/sys/net/core/somaxconn
echo 1 > /proc/sys/net/ipv4/tcp_tw_reuse
echo 1 > /proc/sys/net/ipv4/tcp_tw_recycle
echo 0 > /proc/sys/net/ipv4/tcp_syncookies
//测试
cat /proc/sys/net/core/somaxconn
cat /proc/sys/net/ipv4/tcp_tw_reuse
cat /proc/sys/net/ipv4/tcp_tw_recycle
cat /proc/sys/net/ipv4/tcp_syncookies
二、nginx层面
修改nginx配置文件，nginx.conf
增加work_rlimit_nofile和worker_connections数量，并禁用keepalive_timeout。
worker_processes 1; #nginx 进程数，建议按照cpu 数目来指定，一般为它的倍数
worker_rlimit_nofile 20000; #一个nginx 进程打开的最多文件描述符数目，理论值应该是最多打开文件数（ulimit -n）与nginx 进程数相除，但是nginx 分配请求并不是那么均匀，所以最好与ulimit -n 的值保持一致
events {
use epoll;#使用epoll的I/O模型
worker_connections 20000;#每个进程允许的最多连接数， 理论上每台nginx 服务器的最大连接数为worker_processes*worker_connections
multi_accept on;
}
http {
keepalive_timeout 0;
}
