Nginx series (Thirteen nginx load balancing)

一.负载均衡配置

<code><span># 负载均衡</span><span>upstream</span> webserver {
    <span>server</span><span>192.168.27.134:8080</span>;
    <span>server</span><span>192.168.27.135:80</span>;
    <span>server</span><span>192.168.27.136:80</span>;
}

<span># 反向代理</span><span>server</span> { 
    <span>listen</span><span>80</span>;
    <span>server_name</span> www.test01.com;

    <span>location</span> / {
        <span>proxy_pass</span><span>http://webserver</span>;
        <span>proxy_redirect</span><span>off</span>;
        <span>proxy_set_header</span> Host <span>$host</span>;
        <span>proxy_set_header</span> X-Real-IP <span>$remote_addr</span>;
        <span>proxy_set_header</span> X-Forwarded-For <span>$proxy_add_x_forwarded_for</span>;
    }
}

<span># web服务器</span><span>server</span> {
    <span>listen</span><span>8080</span>;
    <span>server_name</span>  www.test01.com;
    <span>access_log</span>  logs/host.access.log  main;

    <span>location</span> / {
        <span>root</span>   /web/www/thinkphp;
        <span>index</span>  index.php index.html index.htm;

        <span>if</span> (!-e <span>$request_filename</span>) {
            <span>rewrite</span><span> ^/(.*)$</span> /index.php/<span>$1</span><span>last</span>;
        }
    }

    <span>location</span><span>~ [^/]\.php(/|$)</span> {
        <span>root</span> /web/www/thinkphp;
        <span>fastcgi_index</span> index.php;
        <span>fastcgi_pass</span><span>127.0.0.1:9000</span>;
        <span>include</span>      fastcgi_params;

        <span>fastcgi_split_path_info</span><span> ^(.+?\.php)(/.*)$</span>;
        <span>fastcgi_param</span> SCRIPT_FILENAME <span>$document_root</span><span>$fastcgi_script_name</span>;
        <span>fastcgi_param</span> SCRIPT_NAME <span>$fastcgi_script_name</span>;
        <span>fastcgi_param</span> PATH_INFO <span>$fastcgi_path_info</span>;
        <span>fastcgi_param</span> PATH_TRANSLATED <span>$document_root</span><span>$fastcgi_path_info</span>;
    }
}</code>

二. 负载均衡算法

1.轮询 (默认)

<code>upstream webserver {
    <span>server</span><span>192.168</span><span>.27</span><span>.134</span>:<span>8080</span>;
    <span>server</span><span>192.168</span><span>.27</span><span>.135</span>:<span>80</span>;
    <span>server</span><span>192.168</span><span>.27</span><span>.136</span>:<span>80</span>;
}</code>

2.weight

<code>upstream webserver {
    <span>server</span><span>192.168</span><span>.27</span><span>.134</span>:<span>8080</span> weight=<span>1</span>;
    <span>server</span><span>192.168</span><span>.27</span><span>.135</span>:<span>80</span> weight=<span>2</span>;
    <span>server</span><span>192.168</span><span>.27</span><span>.136</span>:<span>80</span> weight=<span>2</span>;
}</code>

3.ip_hash ( 解决session共享问题）

<code>upstream webserver {
    ip_hash;
    <span>server</span><span>192.168</span><span>.27</span><span>.134</span>:<span>8080</span>;
    <span>server</span><span>192.168</span><span>.27</span><span>.135</span>:<span>80</span>;
    <span>server</span><span>192.168</span><span>.27</span><span>.136</span>:<span>80</span>;
}</code>

4.fair 按后端服务器的响应时间来分配请求，响应时间短的优先分配。（第三方）

<code>upstream webserver {
    <span>server</span><span>192.168</span><span>.27</span><span>.134</span>:<span>8080</span>;
    <span>server</span><span>192.168</span><span>.27</span><span>.135</span>:<span>80</span>;
    <span>server</span><span>192.168</span><span>.27</span><span>.136</span>:<span>80</span>;
    fair;
}</code>

参考：http://wiki.nginx.org/HttpUpstreamFairModule

5.url_hash 按访问url的hash结果来分配请求，使每个url定向到同一个后端服务器，后端服务器为缓存时比较有效。（第三方）

<code><span>upstream</span> webserver {
    <span>server</span><span>192.168.27.134:8080</span>;
    <span>server</span><span>192.168.27.135:80</span>;
    <span>server</span><span>192.168.27.136:80</span>;
    <span>hash</span><span>$request_uri</span>;
}</code>

参考：http://wiki.nginx.org/HttpUpstreamRequestHashModule

6.一致性hash （第三方）
参考：http://wiki.nginx.org/HttpUpstreamConsistentHash

三.相关配置

Weight:指定轮询权值，Weight值越大，分配到的访问机率截越高，主要用于后端每个服务器性能不均的情况下。

down:表示当前的server暂时不参与负载均衡，通常和ip_hash一起使用。

backup:预留的备份机器。当其他所有的非backup机器出现故障或者忙的时候，才会请求backup机器，因此这台机器的压力最轻。

max_fails:在一定时间内（这个时间在fail_timeout参数中设置）检查这个服务器是否可用时产生的最多失败请求数，默认为1。将其设置为0可以关闭检查，这些错误在proxy_next_upstream和fastcgi_next_upstream（404错误不会使max_fails增加）中定义。

fail_timeout:在这个时间内产生了max_fails所设置大小的失败尝试连接请求后这个服务器可能不可用，同样它指定了服务器不可用的时间（在下一次尝试连接请求发起之前），默认为10秒，fail_timeout与前端响应时间没有直接关系，不过可以使用proxy_connect_timeout和proxy_read_timeout来控制。

四.参考资料

http://baidutech.blog.51cto.com/4114344/1033718

http://lobert.iteye.com/blog/1929623

http://blog.csdn.net/poechant/article/details/7256184

以上就介绍了Nginx系列（十三 nginx负载均衡），包括了方面的内容，希望对PHP教程有兴趣的朋友有所帮助。