一、集群高可用概述
單純的lvs/nginx反向代理模型做負載集群應用時，DR（director）存在單點故障隱患，故需要有機制來保證DR的高可用性。常用的高可用性方案有Keepalived、corosync，Keepalived主要是由VRRP協議實現了VIPfloating，比較適用於前端DR的高可用性，Corosync一般用於更專業的集群模型實現Service的高可用。Keepalived起初就是為了實現LVS集群director高可用而開發的，本處僅做Keepalived+LVS-DR模型實驗。
二、Keepalived原理簡介
Keepalived中優先級高的節點為MASTER。MASTER其中一個職責就是響應VIP的arp包，將VIP和mac地址映射關系告訴局域網內其 他主機，同時，它還會以多播的形式向局域網中發送VRRP通告，告知BACKUP組自己的優先級。網絡中的所有BACKUP節點只負責 處理MASTER發出的多播包，當發現MASTER的優先級沒自己高（腳本檢測故障觸發自我降級），或者沒收到MASTER的VRRP通告（網絡故障/MASTER宕機）時，BACKUP將自己切換到MASTER狀 態，然后做MASTER該做的事：1.響應arp包，2.發送VRRP通告。
三、實驗環境
1.網絡拓補圖
2.軟件環境
- 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）兩台DR配置keepalived.conf
-
配置DR1：
global_defs {
notification_email {
root@localhost #此處僅發給本機，更定制化的郵件通知功能一般由zabbix來做。
}
notification_email_from keepalived@localhost
smtp_server 127.0.0.1
smtp_connect_timeout 30
router_id keepalivedR1 #定義路由器標識,每台服務器局域網內唯一就行。
vrrp_mcast_group4 224.0.0.33 #定義master向backup組播vrrp報文的地址。
}
vrrp_script chk_down {
script "/etc/keepalived/chk_down.sh"
interval 1 #腳本檢測間隔
weight -15 #即原有優先級+weight，負值即降低。注意當weight=0時, 用於變更vrrp_instance的狀態例如腳本檢測失敗, 則vrrp的狀態直接變為FAULT. (不管有沒有其他節點存在來接管MASTER)
fall 2 #執行腳本兩次exit非0則降低優先級 rise 1 #執行腳本一次exit 0 則還原優先級 user keepalived_script #默認用戶yum安裝keepalived需自建，不存在則調用root（不推薦） } vrrp_instance VI_1 { state MASTER #自定義的state，但如果你的優先級小於backup也不會成為master。 interface ens39 #綁定為當前虛擬路由器使用的物理接口。 virtual_router_id 3 #當前虛擬路由器的惟一標識，范圍是0-255。同實例一致！ priority 99 #初始優先級，范圍1-254。 advert_int 1 #vrrp通告的時間間隔。 authentication { auth_type PASS #簡單密碼驗證，不超過8位。 auth_pass 736w4ib2 #最好使用隨機字符串,同vip實例保持一致！ } virtual_ipaddress { 192.168.7.120/24 dev ens39 } notify_master "/etc/keepalived/notify.sh master" #調用通知腳本 notify_backup "/etc/keepalived/notify.sh backup" notify_fault "/etc/keepalived/notify.sh fault" } virtual_server 192.168.7.120 80 { #此處可用IP port/fwmark id 標識VIP對應服務 delay_loop 2 #服務輪詢的時間間隔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 } }
將上述配置錄入/etc/keepalived/keepalived.conf中，過程如下：
[root@DR1 ~]# cd /etc/keepalived/
[root@DR1 keepalived]# cp keepalived.conf{,.bak} #備份下配置文件
[root@DR1 keepalived]# ls
keepalived.conf keepalived.conf.bak
[root@DR1 keepalived]# vim keepalived.conf
-
配置中調用的notify腳本內容如下：
#!/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只有條件判斷：[[ -f /etc/keepalived/down ]] && exit 1 || exit 0
- 配置DR2
DR2，僅需修改如下配置：
- router_id keepalivedR2
- state BACKUP
- priority 90
（2）兩台RS配置
兩台RS均yum安裝nginx，啟動服務監聽80端口即可。
需要注意的有如下幾點：
-
檢查系統是否自帶httpd服務並停止，否則會與nginx發生沖突。
-
添加vip到網卡輔助接口，並修改arp參數抑制apr報文響應，此處可用腳本如下：
#!/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：因客戶端在同網段，添加完RS抑制arp響應后，需要將客戶端的arp緩存清空防止干擾效果。
-
修改默認主頁，主要是做標記區分RS1/RS2。可采用如下方法：
編輯默認index.html
在vim末行模式輸入:%s/nginx/server1/g回車即可
五、測試實驗效果
-
測試VIP漂移
-
先在DR2初始BACKUP開啟keepalived觀察是否有狀態變更：
[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，因為master沒上線，DR2由backup變為master。我們的notify.sh腳本成功發送狀態變更通知到本地郵箱。
-
測試DR1上線效果：
[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因為收到了更高優先級的VRRP組播報文而卸載VIP，DR1獲取到VIP。 -
手工touch生成/etc/keepalived/down文件模擬單機故障也會發生上述VIP漂移現象。
Jul 15 15:21:51 DR1 Keepalived_vrrp[3907]: WARNING - default user 'keepalived_script' for script execution does not exist - please create.
# 此處踩坑了，提示conf文件中配置的腳本需要keepalived_script用戶執行
# 還有個巨坑是參考資料中直接在keepalived.conf中定義script "[[ -f /etc/keepalived/down ]] && # exit 1 || exit 0" 但是我多次測試不成功，將條件判斷放入腳本，conf文件僅引用路徑才成功......
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
-
-
測試RS故障檢測
- 先測試RS是否正常被輪詢 for i in {1,5};do curl http://192.168.7.120 ;done
- 手動關停一台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"- 重新啟動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
重新上線的RS可以順利被檢測到並上線服務，至此，本次實驗結束。
- 重新啟動nginx