haproxy配置实现动静分离

  上一篇文章介绍了haproxy的简单安装以及最小化启动.实际生产环境中haproxy是没有这么简单的.下面贴上我们公司根据request_url来实现动静分离.

global
   maxconn 10240
   user haproxy
   group haproxy
   daemon
   nbproc 4
   pidfile /var/run/haproxy.pid
defaults
   log global
   option dontlognull   #不记录空连接
   option redispatch
   timeout connect 10000
   timeout client 300000
   timeout server 300000
   maxconn 60000
   retries 3
frontend main *:80
   mode http
   option httpclose
   option httplog
   option forwardfor
   acl host_dpool hdr_beg(host) -i cache static attach1
   acl host_dpool hdr_bet(host) -i yiqisoo.com
   use_backend dpool if host_dpool
   acl host_imgcdn hdr_end(host) -i  imgcdn.house.sina.com.cn
   use_backend imgcdn if host_imgcdn
   acl host_bbs hdr_end(host) -i bbs.house.sina.com.cn bbs0.house.sina.com.cn
   use_backend bbs if host_bbs
   acl host_jiajustatic hdr_end(host) -i static.jiaju.com
   use_backend jiajustatic if host_jiajustatic
   default_backend spool
frontend status
    bind *:81
    default_backend  status
backend status
    mode http
    stats enable
    stats hide-version
    stats uri /haproxy-stats
    stats realm Haproxy\ statistics
    stats auth evil:redhat
backend dpool
    mode http
    balance uri len 128 depth 8 根据uri 进行hash 因为后端的real_server为squid.
    server p-01 1.1.1.1:80 maxconn 2000 check inter 2000 rise 3 fall 3
    server p-02 2.2.2.2:80 maxconn 2000 check inter 2000 rise 3 fall 3
    server p-03 3.3.3.3:80 maxconn 2000 check inter 2000 rise 3
fall 3
##check 接受服务器健康检测
##inter 两次健康检测所相隔的时间 默认为毫秒
##rise  指多少次检测后认为服务为健康状态
##fall  指多少次检测后认为服务为不可用
backend imgcdn
    mode http
    ba
    server imgcdn-01 11.12.13.14:80 weight 1 check inter 10000
backend bbs
    mode http
    balance roundrobin
    server b-01 172.16.1.1:80 weight 1 check inter 10000
    server b-02 172.16.1.2:80 weight 1 check inter 10000
    server b-03 172.16.1.3:80 weight 1 check inter 10000
backend jiajustatic
    mode http
    balance roundrobin
    server jiajustatic-01 192.168.1.1:80 weight 1 check inter 10000