基于keepalived+nginx部署强健的高可用7层负载均衡方案

高可用是个老生常谈的问题了，开源的高可用软件已经做的相当成熟了，之前也在debian下做过lvs+heartbeat的4层LB，一直很稳定（可惜流量不大啊），现在由于业务的需要，做一个基于keepalived+nginx的高可用7层负载均衡。

拓扑结构也比较简单，就不画拓扑图了：2个节点上分别安装配置keepalived和nginx，配置nginx反向代理后端的real server

比较关键的几个点：

1、为避免同一个局域网中有多个keepalived组中的多播相互响应，采用单播通信

2、状态切换的过程触发邮件通知、短信通知、web通知、log记录，便于通过各种途径了解主备工作状态

3、nginx的检测脚本采用了轻量级的方式：”killall -0 nginx”，还可以使用 pidof nginx的方式或者调用其他自定义检测脚本的方式

4、特别要注意优先级的大小及检测到异常时权重的变化

5、了解免费ARP的工作机制

6、了解VRRP协议的适用范围：局域网，第一跳网关冗余

7、单个vrrp实例工作在主备模式，为最大程度的利用2个节点的资源，可以做多个vrrp实例，实现高可用和负载均衡

为便于软件包的管理，采用centos自带的keepalived，nginx1.8.0采用nginx官方源，整体安装也比较简单。

#yum install keepalived nginx -y

设置关键服务的开机启动

#chkconfig nginx on
#chkconfig keepalived on

查看keepalived包安装了那些文件（文档一定好好看）：

#rpm -ql keepalived

/etc/keepalived
/etc/keepalived/keepalived.conf
/etc/rc.d/init.d/keepalived
/etc/sysconfig/keepalived
/usr/bin/genhash
/usr/libexec/keepalived
/usr/sbin/keepalived
/usr/share/doc/keepalived-1.2.13
/usr/share/doc/keepalived-1.2.13/AUTHOR

………

keepalived的主配置文件，2台机器的配置文件略有区别，具体请看配置文件中注释

[root@proxy101 ~]# more /etc/keepalived/keepalived.conf
####Configuration File for keepalived
####内部API网关 keepalived HA配置
#### laijingli 20151213

global_defs {
   notification_email {
     xxxx@qq.com
   }
   notification_email_from yn_alert@xxx.me
   smtp_server 127.0.0.1
   smtp_connect_timeout 30
   router_id proxy101  ## proxy101 on master101 , proxy102 on backup102
}

###simple check with killall -0 which is less expensive than pidof to verify that nginx is running
vrrp_script chk_nginx {
        script “killall -0 nginx”
        interval 1
        weight 2
fall 2
rise 1
}

vrrp_instance YN_API_GATEWAY {
    state MASTER   ## MASTER on master101 ,  BACKUP on backup102
    interface em1
    virtual_router_id 101
    priority 200   ## 200 on master101 ,  199 on backup102
    advert_int 1
    ###采用单播通信，避免同一个局域网中多个keepalived组之间的相互影响
    unicast_src_ip 192.168.0.101  ##本机ip
    unicast_peer {
192.168.0.102  ##对端ip
    }
    authentication {
        auth_type PASS
        auth_pass testpass
    }
    virtual_ipaddress {
        192.168.0.105  ## VIP
    }
    ###如果只有一块网卡的话监控网络接口就没有必要了
    #track_interface {
    # em1
    #}
    track_script {
        chk_nginx
    }
    ###状态切换是发送邮件通知，本机记录log，后期会触发短信通知
    notify_master /usr/local/bin/keepalived_notify.sh notify_master
    notify_backup /usr/local/bin/keepalived_notify.sh notify_backup
    notify_fault  /usr/local/bin/keepalived_notify.sh notify_fault
    notify        /usr/local/bin/keepalived_notify.sh notify
    smtp_alert
}


VRRP 实例组节点状态切换触发邮件通知、短信通知、log记录的脚本
# more /usr/local/bin/keepalived_notify.sh
#!/bin/bash
###keepalived notify script for record ha state transtion to log files

###将将状态转换过程记录到log，便于排错
logfile=/var/log/keepalived.notify.log
echo ————— >> $logfile<br>echo `date` [`hostname`] keepalived HA role state transition: &nbsp;$1 $2$3 $4$5 $6 &gt;&gt;$logfile

###将状态转换记录到nginx的文件，便于通过web查看ha状态（一定注意不要开放到公网）
echo `date` `hostname` $1$2 $3$4 $5$6 > /usr/share/nginx/html/index.html

nginx的部分配置文件，仅供参考

# more /etc/nginx/nginx.conf

###运维管理用途: 用于区别vip跑在那台服务器上
server {
    listen       80;
    server_name  localhost;

    location / {
        root   /usr/share/nginx/html;
        index  index.html index.htm;
    }

## nginx monitor use only
###add by lai monitor nginx status
                location /server-status {
                        stub_status on;
allow 127.0.0.1;
allow 192.168.0.0/24;
                }
        }

通过启停keepalived和nginx来模拟故障，测试vrrp 实例的状态切换过程（即VIP的漂移）：

/etc/init.d/keepalived start/stop

/etc/init.d/nginx stop start/stop

通过启停em1内网网卡、em2公网网卡来模拟故障，测试vrrp 实例的状态切换过程（即VIP的漂移）

ifdown em1

ifdown em2

查看VIP跑在那台服务器上：

# ip addr show|grep 192.168
    inet 192.168.0.101/24 brd 192.168.0.255 scope global em1
    inet 192.168.0.105/32 scope global em1

测试：

[root@test28 ~]$curl 192.168.0.101<br>Mon Dec 14 16:27:10 CST 2015 proxy101 INSTANCE YN_API_GATEWAY MASTER 202<br>[root@test28 ~]$ curl 192.168.0.102
Mon Dec 14 16:34:40 CST 2015 proxy102 INSTANCE YN_API_GATEWAY BACKUP 199
[root@test28 ~]$ curl 192.168.0.105
Mon Dec 14 16:27:10 CST 2015 proxy101 INSTANCE YN_API_GATEWAY MASTER 202

# tail /var/log/keepalived.notify.log
Mon Dec 14 16:25:13 CST 2015 [proxy101] keepalived HA role state transition:
Mon Dec 14 16:25:13 CST 2015 [proxy101] keepalived HA role state transition: INSTANCE YN_API_GATEWAY MASTER 202
—————
—————
Mon Dec 14 16:26:34 CST 2015 [proxy101] keepalived HA role state transition:
Mon Dec 14 16:26:34 CST 2015 [proxy101] keepalived HA role state transition: INSTANCE YN_API_GATEWAY BACKUP 200
—————
—————
Mon Dec 14 16:27:10 CST 2015 [proxy101] keepalived HA role state transition:
Mon Dec 14 16:27:10 CST 2015 [proxy101] keepalived HA role state transition: INSTANCE YN_API_GATEWAY MASTER 202

# tail /var/log/messages
Dec 14 16:27:08 localhost Keepalived_vrrp[74308]: VRRP_Instance(YN_API_GATEWAY) forcing a new MASTER election
Dec 14 16:27:08 localhost Keepalived_vrrp[74308]: VRRP_Instance(YN_API_GATEWAY) forcing a new MASTER election
Dec 14 16:27:09 localhost Keepalived_vrrp[74308]: VRRP_Instance(YN_API_GATEWAY) Transition to MASTER STATE
Dec 14 16:27:10 localhost Keepalived_vrrp[74308]: VRRP_Instance(YN_API_GATEWAY) Entering MASTER STATE
Dec 14 16:27:10 localhost Keepalived_vrrp[74308]: VRRP_Instance(YN_API_GATEWAY) setting protocol VIPs.
Dec 14 16:27:10 localhost Keepalived_vrrp[74308]: VRRP_Instance(YN_API_GATEWAY) Sending gratuitous ARPs on em1 for 192.168.0.105
Dec 14 16:27:10 localhost Keepalived_healthcheckers[74307]: Netlink reflector reports IP 192.168.0.105 added
Dec 14 16:27:10 localhost Keepalived_vrrp[74308]: Remote SMTP server [127.0.0.1]:25 connected.
Dec 14 16:27:10 localhost Keepalived_vrrp[74308]: SMTP alert successfully sent.
Dec 14 16:27:15 localhost Keepalived_vrrp[74308]: VRRP_Instance(YN_API_GATEWAY) Sending gratuitous ARPs on em1 for 192.168.0.105

主备状态转换通知邮件：

通过抓包查看来了解详细的工作过程：

# tcpdump -ni em1 vrrp
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on em1, link-type EN10MB (Ethernet), capture size 65535 bytes
17:36:47.098225 IP 192.168.0.101 > 192.168.0.102: VRRPv2, Advertisement, vrid 101, prio 202, authtype simple, intvl 1s, length 20
17:36:47.388540 IP 192.168.0.22 > 224.0.0.18: VRRPv2, Advertisement, vrid 51, prio 120, authtype simple, intvl 1s, length 20
17:36:48.099409 IP 192.168.0.101 > 192.168.0.102: VRRPv2, Advertisement, vrid 101, prio 202, authtype simple, intvl 1s, length 20
17:36:48.389504 IP 192.168.0.22 > 224.0.0.18: VRRPv2, Advertisement, vrid 51, prio 120, authtype simple, intvl 1s, length 20
17:36:49.100544 IP 192.168.0.101 > 192.168.0.102: VRRPv2, Advertisement, vrid 101, prio 202, authtype simple, intvl 1s, length 20
17:36:49.390487 IP 192.168.0.22 > 224.0.0.18: VRRPv2, Advertisement, vrid 51, prio 120, authtype simple, intvl 1s, length 20
17:36:50.101713 IP 192.168.0.101 > 192.168.0.102: VRRPv2, Advertisement, vrid 101, prio 202, authtype simple, intvl 1s, length 20
17:36:50.391453 IP 192.168.0.22 > 224.0.0.18: VRRPv2, Advertisement, vrid 51, prio 120, authtype simple, intvl 1s, length 20

参考：https://www.nginx.com/resources/admin-guide/nginx-ha-keepalived/