一、集群高可用概述
单纯的lvs/nginx反向代理模型做负载集群应用时，DR（director）存在单点故障隐患，故需要有机制来保证DR的高可用性。常用的高可用性方案有Keepalived、corosync，Keepalived主要是由VRRP协议实现了VIPfloating，比较适用于前端DR的高可用性，Corosync一般用于更专业的集群模型实现Service的高可用。Keepalived起初就是为了实现LVS集群director高可用而开发的，本处仅做Keepalived+LVS-DR模型实验。
二、Keepalived原理简介
Keepalived中优先级高的节点为MASTER。MASTER其中一个职责就是响应VIP的arp包，将VIP和mac地址映射关系告诉局域网内其 他主机，同时，它还会以多播的形式向局域网中发送VRRP通告，告知BACKUP组自己的优先级。网络中的所有BACKUP节点只负责 处理MASTER发出的多播包，当发现MASTER的优先级没自己高（脚本检测故障触发自我降级），或者没收到MASTER的VRRP通告（网络故障/MASTER宕机）时，BACKUP将自己切换到MASTER状 态，然后做MASTER该做的事：1.响应arp包，2.发送VRRP通告。
三、实验环境
1.网络拓补图
2.软件环境
- CentOS7.4
- keepalived.x86_64 1.3.5-6.el7
- nginx.x86_64 1:1.12.2-2.el7
四、配置流程
（1）两台DR配置keepalived.conf
-
配置DR1：
global_defs {
notification_email {
root@localhost #此处仅发给本机，更定制化的邮件通知功能一般由zabbix来做。
}
notification_email_from keepalived@localhost
smtp_server 127.0.0.1
smtp_connect_timeout 30
router_id keepalivedR1 #定义路由器标识,每台服务器局域网内唯一就行。
vrrp_mcast_group4 224.0.0.33 #定义master向backup组播vrrp报文的地址。
}
vrrp_script chk_down {
script "/etc/keepalived/chk_down.sh"
interval 1 #脚本检测间隔
weight -15 #即原有优先级+weight，负值即降低。注意当weight=0时, 用于变更vrrp_instance的状态例如脚本检测失败, 则vrrp的状态直接变为FAULT. (不管有没有其他节点存在来接管MASTER)
fall 2 #执行脚本两次exit非0则降低优先级 rise 1 #执行脚本一次exit 0 则还原优先级 user keepalived_script #默认用户yum安装keepalived需自建，不存在则调用root（不推荐） } vrrp_instance VI_1 { state MASTER #自定义的state，但如果你的优先级小于backup也不会成为master。 interface ens39 #绑定为当前虚拟路由器使用的物理接口。 virtual_router_id 3 #当前虚拟路由器的惟一标识，范围是0-255。同实例一致！ priority 99 #初始优先级，范围1-254。 advert_int 1 #vrrp通告的时间间隔。 authentication { auth_type PASS #简单密码验证，不超过8位。 auth_pass 736w4ib2 #最好使用随机字符串,同vip实例保持一致！ } virtual_ipaddress { 192.168.7.120/24 dev ens39 } notify_master "/etc/keepalived/notify.sh master" #调用通知脚本 notify_backup "/etc/keepalived/notify.sh backup" notify_fault "/etc/keepalived/notify.sh fault" } virtual_server 192.168.7.120 80 { #此处可用IP port/fwmark id 标识VIP对应服务 delay_loop 2 #服务轮询的时间间隔2s lb_algo rr lb_kind DR protocol TCP real_server 192.168.7.125 80 { weight 1 HTTP_GET { url { path / status_code 200 } connect_timeout 2 nb_get_retry 3 delay_before_retry 3 } } real_server 192.168.7.126 80 { weight 1 HTTP_GET { url { path / status_code 200 } connect_timeout 2 nb_get_retry 3 delay_before_retry 3 } }
将上述配置录入/etc/keepalived/keepalived.conf中，过程如下：
[root@DR1 ~]# cd /etc/keepalived/
[root@DR1 keepalived]# cp keepalived.conf{,.bak} #备份下配置文件
[root@DR1 keepalived]# ls
keepalived.conf keepalived.conf.bak
[root@DR1 keepalived]# vim keepalived.conf
-
配置中调用的notify脚本内容如下：
#!/bin/bash # contact='root@localhost' notify() { local mailsubject="$(hostname) to be $1, vip floating" local mailbody="$(date +'%F %T'): vrrp transition, $(hostname) changed to be $1" echo "$mailbody" | mail -s "$mailsubject" $contact } case $1 in master) notify master ;; backup) notify backup ;; fault) notify fault ;; *) echo "Usage: $(basename $0) {master|backup|fault}" exit 1 ;; esac -
chk_down.sh只有条件判断：[[ -f /etc/keepalived/down ]] && exit 1 || exit 0
- 配置DR2
DR2，仅需修改如下配置：
- router_id keepalivedR2
- state BACKUP
- priority 90
（2）两台RS配置
两台RS均yum安装nginx，启动服务监听80端口即可。
需要注意的有如下几点：
-
检查系统是否自带httpd服务并停止，否则会与nginx发生冲突。
-
添加vip到网卡辅助接口，并修改arp参数抑制apr报文响应，此处可用脚本如下：
#!/bin/bash
vip=192.168.7.120
/usr/sbin/ip addr add $vip/32 dev lo label lo:0
echo "1" >/proc/sys/net/ipv4/conf/lo/arp_ignore
echo "2" >/proc/sys/net/ipv4/conf/lo/arp_announce
echo "1" >/proc/sys/net/ipv4/conf/all/arp_ignore
echo "2" >/proc/sys/net/ipv4/conf/all/arp_announce注意：因客户端在同网段，添加完RS抑制arp响应后，需要将客户端的arp缓存清空防止干扰效果。
-
修改默认主页，主要是做标记区分RS1/RS2。可采用如下方法：
编辑默认index.html
在vim末行模式输入:%s/nginx/server1/g回车即可
五、测试实验效果
-
测试VIP漂移
-
先在DR2初始BACKUP开启keepalived观察是否有状态变更：
[root@DR2 ~]# systemctl start keepalived.service
[root@DR2 ~]# systemctl status keepalived.service
?.keepalived.service - LVS and VRRP High Availability Monitor
Loaded: loaded (/usr/lib/systemd/system/keepalived.service; disabled; vendor preset: disabled)
Active: active (running) since Sun 2018-07-15 13:28:10 CST; 19s ago
Process: 2620 ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
Main PID: 2621 (keepalived)
CGroup: /system.slice/keepalived.service
?..2621 /usr/sbin/keepalived -D
?..2622 /usr/sbin/keepalived -D
?..2623 /usr/sbin/keepalived -DJul 15 13:28:14 DR2 Keepalived_vrrp[2623]: Sending gratuitous ARP on ens39 for 192.168.7.120 Jul 15 13:28:14 DR2 Keepalived_vrrp[2623]: Sending gratuitous ARP on ens39 for 192.168.7.120 Jul 15 13:28:14 DR2 Keepalived_vrrp[2623]: Sending gratuitous ARP on ens39 for 192.168.7.120 Jul 15 13:28:14 DR2 Keepalived_vrrp[2623]: Opening script file /etc/keepalived/notify.sh Jul 15 13:28:19 DR2 Keepalived_vrrp[2623]: Sending gratuitous ARP on ens39 for 192.168.7.120 Jul 15 13:28:19 DR2 Keepalived_vrrp[2623]: VRRP_Instance(VI_1) Sending/queueing gratuitous ARPs on ens39 for 192.168.7.120 [root@DR2 ~]# mail Heirloom Mail version 12.5 7/5/10. Type ? for help. "/var/spool/mail/root": 6 messages 2 new 2 unread >N 5 root Sun Jul 15 13:28 18/667 "DR2 to be backup, vip floating" N 6 root Sun Jul 15 13:28 18/667 "DR2 to be master, vip floating" & 6 Message 6: From root@DR2.localdomain Sun Jul 15 13:28:15 2018 Return-Path: <root@DR2.localdomain> X-Original-To: root@localhost Delivered-To: root@localhost.localdomain Date: Sun, 15 Jul 2018 13:28:14 +0800 To: root@localhost.localdomain Subject: DR2 to be master, vip floating User-Agent: Heirloom mailx 12.5 7/5/10 Content-Type: text/plain; charset=us-ascii From: root@DR2.localdomain (root) Status: R 2018-07-15 13:28:14: vrrp transition, DR2 changed to be master可以看到，因为master没上线，DR2由backup变为master。我们的notify.sh脚本成功发送状态变更通知到本地邮箱。
-
测试DR1上线效果：
[root@DR2 ~]# systemctl status keepalived.service
Jul 15 14:52:48 DR2 Keepalived_vrrp[1812]: VRRP_Instance(VI_1) Received advert with higher priority 99, ours 90
Jul 15 14:52:48 DR2 Keepalived_vrrp[1812]: VRRP_Instance(VI_1) Entering BACKUP STATE
Jul 15 14:52:48 DR2 Keepalived_vrrp[1812]: VRRP_Instance(VI_1) removing protocol VIPs.
Jul 15 14:52:48 DR2 Keepalived_vrrp[1812]: Opening script file /etc/keepalived/notify.sh
You have new mail in /var/spool/mail/root
[root@DR2 ~]# ip addr show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: ens39: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
link/ether 00:50:56:3b:a3:7c brd ff:ff:ff:ff:ff:ff
inet 192.168.7.122/24 brd 192.168.7.255 scope global ens39
valid_lft forever preferred_lft forever
inet6 fe80::250:56ff:fe3b:a37c/64 scope link
valid_lft forever preferred_lft forever
[root@DR1 ~]# systemctl status keepalived.service
Jul 15 14:52:49 DR1 Keepalived_vrrp[3471]: Sending gratuitous ARP on ens39 for 192.168.7.120
Jul 15 14:52:49 DR1 Keepalived_vrrp[3471]: Opening script file /etc/keepalived/notify.sh
Jul 15 14:52:54 DR1 Keepalived_vrrp[3471]: Sending gratuitous ARP on ens39 for 192.168.7.120
Jul 15 14:52:54 DR1 Keepalived_vrrp[3471]: VRRP_Instance(VI_1) Sending/queueing gratuitous ARPs on ens39 for 192.168.7.120
DR2因为收到了更高优先级的VRRP组播报文而卸载VIP，DR1获取到VIP。 -
手工touch生成/etc/keepalived/down文件模拟单机故障也会发生上述VIP漂移现象。
Jul 15 15:21:51 DR1 Keepalived_vrrp[3907]: WARNING - default user 'keepalived_script' for script execution does not exist - please create.
# 此处踩坑了，提示conf文件中配置的脚本需要keepalived_script用户执行
# 还有个巨坑是参考资料中直接在keepalived.conf中定义script "[[ -f /etc/keepalived/down ]] && # exit 1 || exit 0" 但是我多次测试不成功，将条件判断放入脚本，conf文件仅引用路径才成功......
Jul 15 15:54:24 DR1 Keepalived_vrrp[4438]: /etc/keepalived/chk_down.sh exited with status 1
Jul 15 15:54:24 DR1 Keepalived_vrrp[4438]: VRRP_Script(chk_down) failed
Jul 15 15:54:25 DR1 Keepalived_vrrp[4438]: VRRP_Instance(VI_1) Changing effective priority from 99 to 84
Jul 15 15:54:25 DR1 Keepalived_vrrp[4438]: /etc/keepalived/chk_down.sh exited with status 1
Jul 15 15:54:25 DR1 Keepalived_vrrp[4438]: VRRP_Instance(VI_1) Received advert with higher priority 90, ours 84
Jul 15 15:54:25 DR1 Keepalived_vrrp[4438]: VRRP_Instance(VI_1) Entering BACKUP STATE
Jul 15 15:54:25 DR1 Keepalived_vrrp[4438]: VRRP_Instance(VI_1) removing protocol VIPs.
Jul 15 15:54:25 DR1 Keepalived_vrrp[4438]: Opening script file /etc/keepalived/notify.sh
-
-
测试RS故障检测
- 先测试RS是否正常被轮询 for i in {1,5};do curl http://192.168.7.120 ;done
- 手动关停一台rs的nginx
[root@RS2 ~]# systemctl stop nginx
[root@DR1 ~]# journalctl -xe
Jul 15 16:59:04 DR1 Keepalived_healthcheckers[4437]: Error connecting server [192.168.7.126]:80.
[root@DR1 ~]# systemctl status keepalived
Jul 15 16:59:05 DR1 Keepalived_healthcheckers[4437]: Error connecting server [192.168.7.126]:80.
Jul 15 16:59:06 DR1 Keepalived_healthcheckers[4437]: Error connecting server [192.168.7.126]:80.
Jul 15 16:59:07 DR1 Keepalived_healthcheckers[4437]: Error connecting server [192.168.7.126]:80.
Jul 15 16:59:07 DR1 Keepalived_healthcheckers[4437]: Check on service [192.168.7.126]:80 failed after 3 retry.
Jul 15 16:59:07 DR1 Keepalived_healthcheckers[4437]: Removing service [192.168.7.126]:80 from VS [192.168.7.120]:80
Jul 15 16:59:07 DR1 Keepalived_healthcheckers[4437]: Remote SMTP server [127.0.0.1]:25 connected.
Jul 15 16:59:07 DR1 Keepalived_healthcheckers[4437]: SMTP alert successfully sent.
[root@DR1 ~]# mail
N 16 keepalived@localhost Sun Jul 15 16:59 17/646 "[keepalivedR1] Realserver [192.168.7.126]:80 - DOWN"- 重新启动nginx
Jul 15 17:02:55 DR1 Keepalived_healthcheckers[4437]: HTTP status code success to [192.168.7.126]:80 url(1).
Jul 15 17:02:55 DR1 Keepalived_healthcheckers[4437]: Remote Web server [192.168.7.126]:80 succeed on service.
Jul 15 17:02:55 DR1 Keepalived_healthcheckers[4437]: Adding service [192.168.7.126]:80 to VS [192.168.7.120]:80
Jul 15 17:02:55 DR1 Keepalived_healthcheckers[4437]: Remote SMTP server [127.0.0.1]:25 connected.
Jul 15 17:02:55 DR1 Keepalived_healthcheckers[4437]: SMTP alert successfully sent.
You have new mail in /var/spool/mail/root
重新上线的RS可以顺利被检测到并上线服务，至此，本次实验结束。
- 重新启动nginx