当网站业务量急剧增大的时候，机器的扩容缩容就是家常便饭，Haproxy作为负载均衡器的变更频率就急剧加大，特别是Docker的出现，这个时候再也不能通过vim修改配置文件的形式去对业务进行变更。

etcd:k/v形式的键值存储，用于存放配置

confd:根据从etcd中获取的配置，对haproxy进行修改

#confd的安装
wget https://github.com/kelseyhightower/confd/releases/download/v0.11.0/confd-0.11.0-linux-amd64
mv confd* /usr/local/bin/confd
chmod +x /usr/local/bin/confd
/usr/local/bin/confd -version#etcd的安装
wget https://github.com/coreos/etcd/releases/download/v3.1.1/etcd-v3.1.1-linux-amd64.tar.gz
tar zxf etcd-v3.1.1-linux-amd64.tar.gz
cd etcd-v3.1.1-linux-amd64
cp etcd* /bin/
/bin/etcd -version#etcd的启动
mkdir /data/etcd
etcd --name etcdserver --data-dir /data/dir --listen-peer-urls 'http://192.168.2.202:2380' --listen-client-urls 'http://192.168.2.202:2379,http://127.0.0.1:2379' --initial-advertise-peer-urls 'http://192.168.2.202:2380' --advertise-client-urls 'http://192.168.2.202:2379,http://127.0.0.1:2379'  &#confg的配置
mkdir -p /etc/confd/{conf.d,templates}vim /etc/confd/conf.d/haproxy.toml
[template]
src = "haproxy.cfg.tmpl"
dest = "/etc/haproxy/haproxy.cfg"
keys = [  "/web",
]
reload_cmd = "/etc/init.d/haproxy reload"/etc/confd/templates/haproxy.cfg.tmpl
globallog 127.0.0.1 local3maxconn 5000uid 99gid 99daemondefaults#log 127.0.0.1 local3mode httpoption dontlognullretries 3option redispatchmaxconn 2000timeout connect  5000timeout client  50000timeout server  50000
frontend webin
bind :80
mode http
{{ $domains := lsdir "/web"}}
{{range $domain := $domains}}
acl {{$domain}} hdr(host) -i {{$domain}}
`end`
{{range $domain := $domains}}
use_backend {{$domain}}_cluster if {{$domain}}
`end`
{{range $domain := $domains}}
backend {{$domain}}_clusteroption httpchk HEAD /health.html HTTP/1.1\r\nHost:{{$domain}}cookie SERVERID insert indirect nocache
{{$servers := printf "/web/%s/*" $domain}}{{range gets $servers}}server {{base .Key}} ``.`Value` cookie {{base .Key}} check inter 5000 rise 2 fall 3 weight 1
`end`
`end`#启动confd
/usr/local/bin/confd -interval 5 -node '192.168.2.202:2379' -confdir /etc/confd &

注册服务器

curl -XPUT http://192.168.2.202:2379/v2/keys/web/www.test.com/prickly_blackwell -d value="192.168.2.201:49162"
curl -XPUT http://192.168.2.202:2379/v2/keys/web/bbs.test.com/prickly_blackwell -d value="192.168.2.201:49162"
curl -XPUT http://192.168.2.202:2379/v2/keys/web/admin.test.com/prickly_blackwell -d value="192.168.2.201:49162"

etcd注册格式

http://192.168.2.202:2379/v2/keys/web/DOMAIN_NAME/SERVER_NAME -d value="IP:PORT"

http://192.168.2.202:2379/v2/keys/web/域名/服务器名   -d value="IP:端口"

查看/etc/haproxy/haproxy.cfg

global  log 127.0.0.1 local3  maxconn 5000  uid 99  gid 99  daemon  defaults  log 127.0.0.1 local3  mode http  option dontlognull  retries 3  option redispatch  maxconn 2000  contimeout  5000  clitimeout  50000  srvtimeout  50000
frontend webin
bind :80
mode httpacl is_admin.test.com hdr(host) -i admin.test.com
acl is_bbs.test.com hdr(host) -i bbs.test.com
acl is_www.test.com hdr(host) -i www.test.comuse_backend admin.test.com_cluster if is_admin.test.com
use_backend bbs.test.com_cluster if is_bbs.test.com
use_backend www.test.com_cluster if is_www.test.combackend admin.test.com_clusteroption httpchk HEAD /health.html HTTP/1.1\r\nHost:admin.test.comcookie SERVERID insert indirect nocacheserver prickly_blackwell 192.168.2.201:49162 cookie prickly_blackwell check inter 5000 rise 2 fall 3 weight 1backend bbs.test.com_clusteroption httpchk HEAD /health.html HTTP/1.1\r\nHost:bbs.test.comcookie SERVERID insert indirect nocacheserver prickly_blackwell 192.168.2.201:49162 cookie prickly_blackwell check inter 5000 rise 2 fall 3 weight 1backend option httpchk HEAD /health.html HTTP/1.1\r\nHost:www.test.com cookie SERVERID insert indirect nocacheserver prickly_blackwell 192.168.2.201:49162 cookie prickly_blackwell check inter 5000 rise 2 fall 3 weight 1

本文参考:

http://blog.liuts.com/post/242/

https://yq.aliyun.com/articles/8708

在此感谢！