使用nginx作为tcp代理的安装配置例子

   以前一般代理tcp时候，都选用haproxy。nginx 1.9 之后，就多了一个选择。可以使用nginx做代理。这样大家也不用重新去熟悉haproxy。
tcp 代理本质是 端口到端口映射转发，比较简单。没有http层代理那么复杂。我一般用来代理https，大家都知道，以前使用nginx 代理https要配置证书等各种琐事。然而对于tcp层来说，没有证书一说。纯粹转发简单干净的配置，非常的爽; 当然有时候也用tcp替换http层代理。原因是tcp代理配置更加简单，性能也更高。

一、nginx 下载安装

方式一：下载地址：

方式二：通过rpm包安装

1、添加安装源，在/etc/yum.repos.d底下创建文件nginx.repo

[nginx]
name=nginx repo
baseurl=
gpgcheck=0
enabled=1

如果是ubuntu
添加安装源 /etc/apt/sources.list.d/nginx.list， 文件内容为

deb  xenial nginx
deb-src  xenial nginx

2、安装nginx

#yum clean all
#yum install nginx

如果是ubuntu

#sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys $key
#sudo apt-get update
#sudo apt-get install nginx

3、查看安装路径和版本

#whereis nginx
#/usr/sbin/nginx -v
nginx version: nginx/1.10.1

4、查看缺省配置文件路径

#/usr/sbin/nginx -h
-c filename   : set configuration file (default: /etc/nginx/nginx.conf)

二、nginx配置

1、目录规划

mkdir -p /opt/service/nginx/conf
mkdir -p /opt/logs/nginx
cd /opt/service/nginx
ln -s /usr/sbin/nginx
ln -s /opt/logs/nginx log

/opt/service/nginx/
├── conf
├── log -> /opt/logs/nginx
└── nginx -> /usr/sbin/nginx

2、配置nginx.conf

user  nginx;
worker_processes  16;
worker_rlimit_nofile 100000;

error_log  /opt/service/nginx/log/error.log error;
pid        /opt/service/nginx/nginx.pid;
events {
    use epoll;
    worker_connections 10240;
}

include /opt/service/nginx/conf/*.conf;

3、配置tcp 代理 
/opt/service/nginx/conf/nginx_tcp_proxy.conf

stream {
#---------------------------------------------------------------------
#  tcp 代理
#---------------------------------------------------------------------
upstream weixin_proxy {
hash $remote_addr consistent;
server wx.qq.com:443 weight=1 max_fails=3 fail_timeout=60s;
}

server {
listen 443;
proxy_connect_timeout 10s;
proxy_pass weixin_proxy;
proxy_buffer_size 64k;
}
}

三、启动维护nginx

1、编辑 nginx.sh

#!/bin/sh
# description:nginx server
# nginx - this script is used to control nginx service
# processname nginx
# nginx version: nginx/1.10.1

nginx="/usr/sbin/nginx"
prog="nginx"
conf_file="/etc/nginx/nginx.conf"

start() {
    if [ `pgrep $prog | wc -l` -eq 2 ];then
        if [ -x $nginx ] && [ -f $conf_file ];then
            $nginx -c $conf_file
            ret=$?
            if [ $ret -eq 0 ];then
                echo "$prog start       successed"
            else
                echo "$prog start       failed"
            fi
        else
            echo "$prog config file not exist"
        fi
    else
        num=`pgrep $prog `
        echo "$prog is already started ... $num "
    fi
}


stop() {
    if [ `pgrep $prog | wc -l` -ne 2 ];then
        killall -9 $prog
        ret=$?
        if [ $ret -eq 0 ];then
            echo "$prog stop       successed"
        else
            echo "$prog stop    failed"
        fi
    else
        echo "$prog is already stopped ..."
    fi
}

restart() {
    stop
    sleep 2
    start
}

reload() {
    if [ `pgrep $prog | wc -l` -ne 0 ];then
        pid=`ps -ef | grep $prog | grep master | awk '{print $2}'`
        if [ -x $nginx ] && [ -f $conf_file ];then
            kill -HUP $pid
            ret=$?
            if [ $ret -eq 0 ];then
                echo "$prog reload   successed"
            else
                echo "$prog reload    failed"
            fi
        else
            echo "$prog config file is not exist"
        fi
    else
        echo "$prog is stopped, please start $prog first ..."
    fi
}

check() {
    if [ -x $nginx ] && [ -f $conf_file ];then
        $nginx -t -c $conf_file
        ret=$?
        if [ $ret -eq 0 ];then
            echo "$prog check successed"
        else
            echo "$prog check failed"
        fi
    else
        echo "$prog program or config file not exit!"
    fi
}

case $1 in
    start)
        start
        ;;
    stop)
        stop
        ;;
    restart)
        restart
        ;;
    reload)
        reload
        ;;
    check)
        check
        ;;
    *)
        echo "Usage: $0 {start|stop|restart|reload|check}"
esac

2、启动 nginx.sh

chmod a+rwx nginx.sh
./nginx.sh restart

整个目录结构如下

# ll
total 8
drwxr-xr-x 2 root root   33 Dec 21 17:16 conf
lrwxrwxrwx 1 root root   15 Dec 21 17:00 log -> /opt/logs/nginx
lrwxrwxrwx 1 root root   15 Dec 21 16:59 nginx -> /usr/sbin/nginx
-rw-r--r-- 1 root root    6 Dec 21 17:21 nginx.pid
-rwxrwxrwx 1 root root 2172 Dec 21 17:20 nginx.sh

问题：
当代理的后端是域名的时候，域名的dns发生变化。nginx并不知道。看文档nginx提供了定时刷新dns的功能，但是我配置好像没有起到作用。
在 nginx.conf 尾部加上这个配置
    resolver  100.100.2.136 valid=1s;
    include /etc/nginx/nginx_vhost/*.conf;