Keepalived是Linux下一个轻量级别的高可用解决方案。高可用(High Avalilability,HA),其实两种不同的含义:广义来讲,是指整个系统的高可用行,狭义的来讲就是之主机的冗余和接管,它与HeartBeat RoseHA 实现相同类似的功能,都可以实现服务或者网络的高可用,但是又有差别,HeartBeat是一个专业的、功能完善的高可用软件,它提供了HA 软件所需的基本功能,比如:心跳检测、资源接管,检测集群中的服务,在集群节点转移共享IP地址的所有者等等。HeartBeat功能强大,但是部署和使用相对比较麻烦,与HeartBeat相比,Keepalived主要是通过虚拟路由冗余来实现高可用功能,虽然它没有HeartBeat功能强大,但是Keepalived部署和使用非常的简单,所有配置只需要一个配置文件即可以完成.
Keepalived 软件起初是专为LVS负载均衡软件设计的,用来管理并监控LVS集群系统中各个服务节点的状态,后来又加入了可以实现高可用的VRRP功能。因此,Keepalived除了能够管理LVS软件外,还可以作为其他服务(例如:Nginx、Haproxy、MySQL等)的高可用解决方案软件。
Keepalived软件主要是通过VRRP协议实现高可用功能的。VRRP是Virtual Router RedundancyProtocol(虚拟路由器冗余协议)的缩写,VRRP出现的目的就是为了解决静态路由单点故障问题的,它能够保证当个别节点宕机时,整个网络可以不间断地运行。
所以,Keepalived 一方面具有配置管理LVS的功能,同时还具有对LVS下面节点进行健康检查的功能,另一方面也可实现系统网络服务的高可用功能。
keepalived 有三个重要的功能,分别是:
★管理LVS负载均衡软件
★实现LVS集群节点的健康检查
★作为系统网络服务的高可用性(failover)
Keepalived 高可用服务之间的故障切换转移,是通过 VRRP (Virtual Router Redundancy Protocol ,虚拟路由器冗余协议)来实现的。
在 Keepalived 服务正常工作时,主 Master 节点会不断地向备节点发送(多播的方式)心跳消息,用以告诉备 Backup 节点自己还活看,当主 Master 节点发生故障时,就无法发送心跳消息,备节点也就因此无法继续检测到来自主 Master 节点的心跳了,于是调用自身的接管程序,接管主 Master 节点的 IP 资源及服务。而当主 Master 节点恢复时,备 Backup 节点又会释放主节点故障时自身接管的IP资源及服务,恢复到原来的备用角色。
那么,什么是VRRP呢?
VRRP ,全 称 Virtual Router Redundancy Protocol ,中文名为虚拟路由冗余协议 ,VRRP的出现就是为了解决静态踣甶的单点故障问题,VRRP是通过一种竞选机制来将路由的任务交给某台VRRP路由器的。
在现实的网络环境中。主机之间的通信都是通过配置静态路由或者(默认网关)来完成的,而主机之间的路由器一旦发生故障,通信就会失效,因此这种通信模式当中,路由器就成了一个单点瓶颈,为了解决这个问题,就引入了VRRP协议。
熟悉网络的学员对VRRP协议应该不陌生,它是一种主备模式的协议,通过VRRP可以在网络发生故障时透明的进行设备切换而不影响主机之间的数据通信,这其中涉及到两个概念:物理路由器和虚拟路由器。
VRRP可以将两台或者多台物理路由器设备虚拟成一个虚拟路由,这个虚拟路由器通过虚拟IP(一个或者多个)对外提供服务,而在虚拟路由器内部十多个物理路由器协同工作,同一时间只有一台物理路由器对外提供服务,这台物理路由设备被成为:主路由器(Master角色),一般情况下Master是由选举算法产生,它拥有对外服务的虚拟IP,提供各种网络功能,如:ARP请求,ICMP 数据转发等,而且其它的物理路由器不拥有对外的虚拟IP,也不提供对外网络功能,仅仅接收MASTER的VRRP状态通告信息,这些路由器被统称为“BACKUP的角色”,当主路由器失败时,处于BACKUP角色的备份路由器将重新进行选举,产生一个新的主路由器进入MASTER角色,继续提供对外服务,整个切换对用户来说是完全透明的。
每个虚拟路由器都有一个唯一的标识号,称为VRID,一个VRID与一组IP地址构成一个虚拟路由器,在VRRP协议中,所有的报文都是通过IP多播方式发送的,而在一个虚拟路由器中,只有处于Master角色的路由器会一直发送VRRP数据包,处于BACKUP角色的路由器只会接受Master角色发送过来的报文信息,用来监控Master运行状态,一一般不会发生BACKUP抢占的情况,除非它的优先级更高,而当MASTER不可用时,BACKUP也就无法收到Master发过来的信息,于是就认定Master出现故障,接着多台BAKCUP就会进行选举,优先级最高的BACKUP将称为新的MASTER,这种选举角色切换非常之快,因而保证了服务的持续可用性。
keepalived 的主配置文件是 /etc/keepalived/keepalived.conf其内容如下:
! Configuration File for keepalived
global_defs { //全局配置
notification_email { //定义报警收件人邮件地址
[email protected]
[email protected]
[email protected]
}
notification_email_from [email protected] //定义报警发件人邮箱
smtp_server 192.168.200.1 //邮箱服务器地址
smtp_connect_timeout 30 //定义邮箱超时时间
router_id LVS_DEVEL //定义路由标识信息,同局域网内唯一
vrrp_skip_check_adv_addr
vrrp_strict
vrrp_garp_interval 0
vrrp_gna_interval 0
}
vrrp_instance VI_1 { //定义实例
state MASTER //指定keepalived节点的初始状态,可选值为MASTER|BACKUP
interface eth0 //VRRP实例绑定的网卡接口,用户发送VRRP包
virtual_router_id 51 //虚拟路由的ID,同一集群要一致
priority 100 //定义优先级,按优先级来决定主备角色,优先级越大越优先
nopreempt //设置不抢占
advert_int 1 //主备通讯时间间隔
authentication { //配置认证
auth_type PASS //认证方式,此处为密码
auth_pass 1111 //同一集群中的keepalived配置里的此处必须一致,推荐使用8位随机数
}
virtual_ipaddress { //配置要使用的VIP地址
192.168.200.16
}
}
virtual_server 192.168.200.16 1358 { //配置虚拟服务器
delay_loop 6 //健康检查的时间间隔
lb_algo rr //lvs调度算法
lb_kind NAT //lvs模式
persistence_timeout 50 //持久化超时时间,单位是秒
protocol TCP //4层协议
sorry_server 192.168.200.200 1358 //定义备用服务器,当所有RS都故障时用sorry_server来响应客户端
real_server 192.168.200.2 1358 { //定义真实处理请求的服务器
weight 1 //给服务器指定权重,默认为1
HTTP_GET {
url {
path /testurl/test.jsp //指定要检查的URL路径
digest 640205b7b0fc66c1ea91c463fac6334d //摘要信息
}
url {
path /testurl2/test.jsp
digest 640205b7b0fc66c1ea91c463fac6334d
}
url {
path /testurl3/test.jsp
digest 640205b7b0fc66c1ea91c463fac6334d
}
connect_timeout 3 //连接超时时间
nb_get_retry 3 //get尝试次数
delay_before_retry 3 //在尝试之前延迟多长时间
}
}
real_server 192.168.200.3 1358 {
weight 1
HTTP_GET {
url {
path /testurl/test.jsp
digest 640205b7b0fc66c1ea91c463fac6334c
}
url {
path /testurl2/test.jsp
digest 640205b7b0fc66c1ea91c463fac6334c
}
connect_timeout 3
nb_get_retry 3
delay_before_retry 3
}
}
}
nopreempt //设置为不抢占。默认是抢占的,当高优先级的机器恢复后,会抢占低优先 \
级的机器成为MASTER,而不抢占,则允许低优先级的机器继续成为MASTER,即使高优先级 \
的机器已经上线。如果要使用这个功能,则初始化状态必须为BACKUP。
preempt_delay //设置抢占延迟。单位是秒,范围是0---1000,默认是0.发现低优先 \
级的MASTER后多少秒开始抢占。
//作用:添加一个周期性执行的脚本。脚本的退出状态码会被调用它的所有的VRRP Instance记录。
//注意:至少有一个VRRP实例调用它并且优先级不能为0.优先级范围是1-254.
vrrp_script {
...
}
//选项说明:
script "/path/to/somewhere" //指定要执行的脚本的路径。
interval //指定脚本执行的间隔。单位是秒。默认为1s。
timeout //指定在多少秒后,脚本被认为执行失败。
weight <-254 --- 254> //调整优先级。默认为2.
rise //执行成功多少次才认为是成功。
fall //执行失败多少次才认为失败。
user [GROUPNAME] //运行脚本的用户和组。
init_fail //假设脚本初始状态是失败状态。
//weight说明:
1. 如果脚本执行成功(退出状态码为0),weight大于0,则priority增加。
2. 如果脚本执行失败(退出状态码为非0),weight小于0,则priority减少。
3. 其他情况下,priority不变。
weight //给服务器指定权重。默认是1
inhibit_on_failure //当服务器健康检查失败时,将其weight设置为0, \
//而不是从Virtual Server中移除
notify_up //当服务器健康检查成功时,执行的脚本
notify_down //当服务器健康检查失败时,执行的脚本
uthreshold //到这台服务器的最大连接数
lthreshold //到这台服务器的最小连接数
connect_ip //连接的IP地址。默认是real server的ip地址
connect_port //连接的端口。默认是real server的端口
bindto //发起连接的接口的地址。
bind_port //发起连接的源端口。
connect_timeout //连接超时时间。默认是5s。
fwmark //使用fwmark对所有出去的检查数据包进行标记。
warmup //指定一个随机延迟,最大为N秒。可防止网络阻塞。如果为0,则关闭该功能。
retry //重试次数。默认是1次。
delay_before_retry //默认是1秒。在重试之前延迟多少秒。
环境
主机名 | 系统 | IP |
---|---|---|
WangJiayue(master) | CentOS7 | 192.168.222.139 |
WangJiayue-2(slave) | CentOS7 | 192.168.222.133 |
5.1.1 关闭防火墙与selinux
[root@WangJiayue ~]# systemctl stop firewalld
[root@WangJiayue ~]# systemctl disable firewalld
//这里因为我已经提前关闭所以少显示两行
[root@WangJiayue ~]# setenforce 0
setenforce: SELinux is disabled
[root@WangJiayue ~]# sed -ri 's/^(SELINUX=).*/\1disabled/g' /etc/selinux/config
5.1.2 配置网络源
[root@WangJiayue ~]# curl -o /etc/yum.repos.d/CentOS-Base.repo https://mirrors.aliyun.com/repo/Centos-7.repo
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 2523 100 2523 0 0 18939 0 --:--:-- --:--:-- --:--:-- 18969
[root@WangJiayue ~]# sed -i 's/\$releasever/7/g' /etc/yum.repos.d/CentOS-Base.repo
[root@WangJiayue ~]# sed -i -e '/mirrors.cloud.aliyuncs.com/d' -e '/mirrors.aliyuncs.com/d' /etc/yum.repos.d/CentOS-Base.repo
[root@WangJiayue ~]# wget -O /etc/yum.repos.d/epel.repo http://mirrors.aliyun.com/repo/epel-7.repo
--2020-08-06 10:28:00-- http://mirrors.aliyun.com/repo/epel-7.repo
正在解析主机 mirrors.aliyun.com (mirrors.aliyun.com)... 58.49.194.248, 116.211.183.217, 116.211.183.218, ...
正在连接 mirrors.aliyun.com (mirrors.aliyun.com)|58.49.194.248|:80... 已连接。
已发出 HTTP 请求,正在等待回应... 200 OK
长度:664 [application/octet-stream]
正在保存至: “/etc/yum.repos.d/epel.repo”
100%[======================================================================================================================================>] 664 --.-K/s 用时 0s
2020-08-06 10:28:00 (210 MB/s) - 已保存 “/etc/yum.repos.d/epel.repo” [664/664])
5.1.3 安装keepalived并查看生成的文件
[root@WangJiayue ~]# yum -y install keepalived gcc gcc-c++
//省略安装过程
[root@WangJiayue ~]# rpm -ql keepalived
/etc/keepalived //配置目录
/etc/keepalived/keepalived.conf //此为主配置文件
/etc/sysconfig/keepalived
/usr/bin/genhash
/usr/lib/systemd/system/keepalived.service //此为服务控制文件
/usr/libexec/keepalived
/usr/sbin/keepalived
/usr/share/doc/keepalived-1.3.5
/usr/share/doc/keepalived-1.3.5/AUTHOR
/usr/share/doc/keepalived-1.3.5/CONTRIBUTORS
/usr/share/doc/keepalived-1.3.5/COPYING
/usr/share/doc/keepalived-1.3.5/ChangeLog
/usr/share/doc/keepalived-1.3.5/NOTE_vrrp_vmac.txt
/usr/share/doc/keepalived-1.3.5/README
/usr/share/doc/keepalived-1.3.5/TODO
/usr/share/doc/keepalived-1.3.5/keepalived.conf.SYNOPSIS
/usr/share/doc/keepalived-1.3.5/samples
......
5.2.1 关闭防火墙与selinux
[root@WangJiayue-2 ~]# systemctl stop firewalld
[root@WangJiayue-2 ~]# systemctl disable firewalld
//这里因为我已经提前关闭所以少显示两行
[root@WangJiayue-2 ~]# setenforce 0
setenforce: SELinux is disabled
[root@WangJiayue-2 ~]# sed -ri 's/^(SELINUX=).*/\1disabled/g' /etc/selinux/config
5.2.2 配置网络源
[root@WangJiayue-2 ~]# curl -o /etc/yum.repos.d/CentOS-Base.repo https://mirrors.aliyun.com/repo/Centos-7.repo
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 2523 100 2523 0 0 18939 0 --:--:-- --:--:-- --:--:-- 18969
[root@WangJiayue-2 ~]# sed -i 's/\$releasever/7/g' /etc/yum.repos.d/CentOS-Base.repo
[root@WangJiayue-2 ~]# sed -i -e '/mirrors.cloud.aliyuncs.com/d' -e '/mirrors.aliyuncs.com/d' /etc/yum.repos.d/CentOS-Base.repo
[root@WangJiayue-2 ~]# wget -O /etc/yum.repos.d/epel.repo http://mirrors.aliyun.com/repo/epel-7.repo
--2020-08-06 10:28:00-- http://mirrors.aliyun.com/repo/epel-7.repo
正在解析主机 mirrors.aliyun.com (mirrors.aliyun.com)... 58.49.194.248, 116.211.183.217, 116.211.183.218, ...
正在连接 mirrors.aliyun.com (mirrors.aliyun.com)|58.49.194.248|:80... 已连接。
已发出 HTTP 请求,正在等待回应... 200 OK
长度:664 [application/octet-stream]
正在保存至: “/etc/yum.repos.d/epel.repo”
100%[======================================================================================================================================>] 664 --.-K/s 用时 0s
2020-08-06 10:29:00 (210 MB/s) - 已保存 “/etc/yum.repos.d/epel.repo” [664/664])
[root@WangJiayue ~]# yum -y install httpd
//省略安装过程
[root@WangJiayue ~]# cd /var/www/html/
[root@WangJiayue html]# echo "master" > /var/www/html/index.html
[root@WangJiayue html]# cat index.html
master
[root@WangJiayue html]# systemctl enable --now httpd
Created symlink from /etc/systemd/system/multi-user.target.wants/httpd.service to /usr/lib/systemd/system/httpd.service.
[root@WangJiayue html]# ss -antl
State Recv-Q Send-Q Local Address:Port Peer Address:Port
LISTEN 0 128 *:22 *:*
LISTEN 0 100 127.0.0.1:25 *:*
LISTEN 0 128 :::80 :::*
LISTEN 0 128 :::22 :::*
LISTEN 0 100 ::1:25 :::*
[root@WangJiayue html]#
[root@WangJiayue-2 ~]# yum -y install httpd
//省略安装过程
[root@WangJiayue-2 ~]# cd /var/www/html/
[root@WangJiayue-2 html]# echo "slave" > /var/www/html/index.html
[root@WangJiayue-2 html]# cat index.html
slave
[root@WangJiayue-2 html]# systemctl enable --now httpd
Created symlink from /etc/systemd/system/multi-user.target.wants/httpd.service to /usr/lib/systemd/system/httpd.service.
[root@WangJiayue-2 html]# ss -antl
State Recv-Q Send-Q Local Address:Port Peer Address:Port
LISTEN 0 128 *:22 *:*
LISTEN 0 100 127.0.0.1:25 *:*
LISTEN 0 128 :::80 :::*
LISTEN 0 128 :::22 :::*
LISTEN 0 100 ::1:25 :::*
[root@WangJiayue-2 html]#
[root@WangJiayue ~]# vim /etc/keepalived/keepalived.conf
[root@WangJiayue ~]# cat /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
router_id lb01
}
vrrp_instance VI_1 {
state MASTER
interface ens33
virtual_router_id 51
priority 100
advert_int 1
authentication {
auth_type PASS
auth_pass wjy123
}
virtual_ipaddress {
192.168.222.250
}
}
virtual_server 192.168.222.250 80 {
delay_loop 6
lb_algo rr
lb_kind DR
persistence_timeout 50
protocol TCP
real_server 192.168.222.139 80 {
weight 1
TCP_CHECK {
connect_port 80
connect_timeout 3
nb_get_retry 3
delay_before_retry 3
}
}
real_server 192.168.222.133 80 {
weight 1
TCP_CHECK {
connect_port 80
connect_timeout 3
nb_get_retry 3
delay_before_retry 3
}
}
}
[root@WangJiayue ~]# systemctl start keepalived
[root@WangJiayue ~]# systemctl enable keepalived
Created symlink from /etc/systemd/system/multi-user.target.wants/keepalived.service to /usr/lib/systemd/system/keepalived.service.
[root@WangJiayue ~]# systemctl status keepalived.service
● keepalived.service - LVS and VRRP High Availability Monitor
Loaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)
Active: active (running) since 四 2020-08-06 10:43:24 CST; 49s ago
Main PID: 4237 (keepalived)
CGroup: /system.slice/keepalived.service
├─4237 /usr/sbin/keepalived -D
├─4238 /usr/sbin/keepalived -D
└─4239 /usr/sbin/keepalived -D
8月 06 10:43:26 WangJiayue Keepalived_vrrp[4239]: Sending gratuitous ARP on ens33 for 192.168.222.250
8月 06 10:43:26 WangJiayue Keepalived_vrrp[4239]: Sending gratuitous ARP on ens33 for 192.168.222.250
8月 06 10:43:26 WangJiayue Keepalived_vrrp[4239]: Sending gratuitous ARP on ens33 for 192.168.222.250
8月 06 10:43:26 WangJiayue Keepalived_vrrp[4239]: Sending gratuitous ARP on ens33 for 192.168.222.250
8月 06 10:43:31 WangJiayue Keepalived_vrrp[4239]: Sending gratuitous ARP on ens33 for 192.168.222.250
8月 06 10:43:31 WangJiayue Keepalived_vrrp[4239]: VRRP_Instance(VI_1) Sending/queueing gratuitous ARPs on ens33 for 192.168.222.250
8月 06 10:43:31 WangJiayue Keepalived_vrrp[4239]: Sending gratuitous ARP on ens33 for 192.168.222.250
8月 06 10:43:31 WangJiayue Keepalived_vrrp[4239]: Sending gratuitous ARP on ens33 for 192.168.222.250
8月 06 10:43:31 WangJiayue Keepalived_vrrp[4239]: Sending gratuitous ARP on ens33 for 192.168.222.250
8月 06 10:43:31 WangJiayue Keepalived_vrrp[4239]: Sending gratuitous ARP on ens33 for 192.168.222.250
[root@WangJiayue-2 ~]# vim /etc/keepalived/keepalived.conf
[root@WangJiayue-2 ~]# cat /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
router_id lb01
}
vrrp_instance VI_1 {
state MASTER
interface ens33
virtual_router_id 51
priority 100
advert_int 1
authentication {
auth_type PASS
auth_pass wjy123
}
virtual_ipaddress {
192.168.222.250
}
}
virtual_server 192.168.222.250 80 {
delay_loop 6
lb_algo rr
lb_kind DR
persistence_timeout 50
protocol TCP
real_server 192.168.222.139 80 {
weight 1
TCP_CHECK {
connect_port 80
connect_timeout 3
nb_get_retry 3
delay_before_retry 3
}
}
real_server 192.168.222.133 80 {
weight 1
TCP_CHECK {
connect_port 80
connect_timeout 3
nb_get_retry 3
delay_before_retry 3
}
}
}
[root@WangJiayue-2 ~]# systemctl start keepalived
[root@WangJiayue-2 ~]# systemctl enable keepalived.service
Created symlink from /etc/systemd/system/multi-user.target.wants/keepalived.service to /usr/lib/systemd/system/keepalived.service.
[root@WangJiayue-2 ~]# systemctl status keepalived.service
● keepalived.service - LVS and VRRP High Availability Monitor
Loaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)
Active: active (running) since Thu 2020-08-06 11:02:50 CST; 11s ago
Main PID: 1442 (keepalived)
CGroup: /system.slice/keepalived.service
├─1442 /usr/sbin/keepalived -D
├─1443 /usr/sbin/keepalived -D
└─1444 /usr/sbin/keepalived -D
Aug 06 11:02:57 WangJiayue-2 Keepalived_vrrp[1444]: Sending gratuitous ARP on ens33 for 192.168.222.250
Aug 06 11:02:57 WangJiayue-2 Keepalived_vrrp[1444]: Sending gratuitous ARP on ens33 for 192.168.222.250
Aug 06 11:02:57 WangJiayue-2 Keepalived_vrrp[1444]: Sending gratuitous ARP on ens33 for 192.168.222.250
Aug 06 11:02:57 WangJiayue-2 Keepalived_vrrp[1444]: Sending gratuitous ARP on ens33 for 192.168.222.250
Aug 06 11:02:58 WangJiayue-2 Keepalived_vrrp[1444]: (VI_1): received an invalid passwd!
Aug 06 11:02:58 WangJiayue-2 Keepalived_vrrp[1444]: bogus VRRP packet received on ens33 !!!
Aug 06 11:02:58 WangJiayue-2 Keepalived_vrrp[1444]: VRRP_Instance(VI_1) Dropping received VRRP packet...
Aug 06 11:02:59 WangJiayue-2 Keepalived_vrrp[1444]: (VI_1): received an invalid passwd!
Aug 06 11:02:59 WangJiayue-2 Keepalived_vrrp[1444]: bogus VRRP packet received on ens33 !!!
Aug 06 11:02:59 WangJiayue-2 Keepalived_vrrp[1444]: VRRP_Instance(VI_1) Dropping received VRRP packet...
[root@WangJiayue ~]# ip a
1: lo: mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
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: ens33: mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 00:0c:29:9e:71:dd brd ff:ff:ff:ff:ff:ff
inet 192.168.222.139/24 brd 192.168.222.255 scope global noprefixroute dynamic ens33
valid_lft 1208874sec preferred_lft 1208874sec
inet 192.168.222.250/32 scope global ens33
valid_lft forever preferred_lft forever
inet6 fe80::6397:ea5d:6e1d:bc45/64 scope link noprefixroute
valid_lft forever preferred_lft forever
[root@WangJiayue-2 ~]# ip a
1: lo: mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
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: ens33: mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 00:0c:29:2d:7e:16 brd ff:ff:ff:ff:ff:ff
inet 192.168.222.133/24 brd 192.168.222.255 scope global noprefixroute dynamic ens33
valid_lft 1207259sec preferred_lft 1207259sec
inet6 fe80::eb84:2580:b52e:79ad/64 scope link noprefixroute
valid_lft forever preferred_lft forever
[root@WangJiayue-2 ~]# systemctl stop httpd.service
keepalived通过脚本来监控apache负载均衡机的状态
[root@WangJiayue ~]# mkdir /scripts
[root@WangJiayue ~]# cd /scripts/
[root@WangJiayue scripts]# vim check_apache.sh
[root@WangJiayue scripts]# cat check_apache.sh
#!/bin/bash
apache_status=$(ps -ef|grep -Ev "grep|$0"|grep '\bhttpd\b'|wc -l)
if [ $apache_status -eq 0 ];then
systemctl stop keepalived
fi
[root@WangJiayue scripts]# chmod +x check_apache.sh
[root@WangJiayue scripts]# ll
总用量 4
-rwxr-xr-x 1 root root 144 8月 6 11:18 check_apache.sh
[root@WangJiayue scripts]# vim notify.sh
[root@WangJiayue scripts]# cat notify.sh
#!/bin/bash
VIP=$2
sendmail (){
subject="${VIP}'s server keepalived state is translate"
content="`date +'%F %T'`: `hostname`'s state change to master"
echo $content | mail -s "$subject" [email protected]
}
case "$1" in
master)
apache_status=$(ps -ef|grep -Ev "grep|$0"|grep '\bhttpd\b'|wc -l)
if [ $apache_status -eq 0 ];then
systemctl start httpd
fi
sendmail
;;
backup)
apache_status=$(ps -ef|grep -Ev "grep|$0"|grep '\bhttpd\b'|wc -l)
if [ $apache_status -gt 0 ];then
systemctl stop httpd
fi
;;
*)
echo "Usage:$0 master|backup VIP"
;;
esac
[root@WangJiayue scripts]# chmod +x notify.sh
[root@WangJiayue scripts]# ll
总用量 8
-rwxr-xr-x 1 root root 144 8月 6 11:18 check_apache.sh
-rwxr-xr-x 1 root root 665 8月 6 11:19 notify.sh
[root@WangJiayue-2 ~]# mkdir /scripts
[root@WangJiayue-2 ~]# cd /scripts/
[root@WangJiayue-2 scripts]# vim notify.sh
[root@WangJiayue-2 scripts]# cat notify.sh
#!/bin/bash
VIP=$2
sendmail (){
subject="${VIP}'s server keepalived state is translate"
content="`date +'%F %T'`: `hostname`'s state change to master"
echo $content | mail -s "$subject" [email protected]
}
case "$1" in
master)
apache_status=$(ps -ef|grep -Ev "grep|$0"|grep '\bhttpd\b'|wc -l)
if [ $apache_status -eq 0 ];then
systemctl start httpd
fi
sendmail
;;
backup)
apache_status=$(ps -ef|grep -Ev "grep|$0"|grep '\bhttpd\b'|wc -l)
if [ $apache_status -gt 0 ];then
systemctl stop httpd
fi
;;
*)
echo "Usage:$0 master|backup VIP"
;;
esac
[root@WangJiayue-2 scripts]# chmod +x notify.sh
[root@WangJiayue-2 scripts]# ll
total 4
-rwxr-xr-x 1 root root 665 Aug 6 11:22 notify.sh
[root@WangJiayue ~]# vim /etc/keepalived/keepalived.conf
[root@WangJiayue ~]# cat /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
router_id lb01
}
vrrp_script httpd_check {
script "/scripts/check_apache.sh"
interval 1
weight -20
}
vrrp_instance VI_1 {
state MASTER
interface ens33
virtual_router_id 51
priority 100
advert_int 1
authentication {
auth_type PASS
auth_pass wjy123
}
virtual_ipaddress {
192.168.222.250
}
track_script {
nginx_check
}
notify_master "/scripts/notify.sh master 192.168.222.250"
notify_backup "/scripts/notify.sh backup 192.168.222.250"
}
virtual_server 192.168.222.250 80 {
delay_loop 6
lb_algo rr
lb_kind DR
persistence_timeout 50
protocol TCP
real_server 192.168.222.139 80 {
weight 1
TCP_CHECK {
connect_port 80
connect_timeout 3
nb_get_retry 3
delay_before_retry 3
}
}
real_server 192.168.222.133 80 {
weight 1
TCP_CHECK {
connect_port 80
connect_timeout 3
nb_get_retry 3
delay_before_retry 3
}
}
}
[root@WangJiayue ~]# systemctl restart keepalived.service
[root@WangJiayue-2 ~]# vim /etc/keepalived/keepalived.conf
[root@WangJiayue-2 ~]# cat /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
router_id lb01
}
vrrp_instance VI_1 {
state MASTER
interface ens33
virtual_router_id 51
priority 100
advert_int 1
authentication {
auth_type PASS
auth_pass wjy123
}
virtual_ipaddress {
192.168.222.250
}
notify_master "/scripts/notify.sh master 192.168.222.250"
notify_backup "/scripts/notify.sh backup 192.168.222.250"
}
virtual_server 192.168.222.250 80 {
delay_loop 6
lb_algo rr
lb_kind DR
persistence_timeout 50
protocol TCP
real_server 192.168.222.139 80 {
weight 1
TCP_CHECK {
connect_port 80
connect_timeout 3
nb_get_retry 3
delay_before_retry 3
}
}
real_server 192.168.222.133 80 {
weight 1
TCP_CHECK {
connect_port 80
connect_timeout 3
nb_get_retry 3
delay_before_retry 3
}
}
}
[root@WangJiayue-2 ~]# systemctl restart keepalived.service
[root@WangJiayue ~]# ss -antl
State Recv-Q Send-Q Local Address:Port Peer Address:Port
LISTEN 0 128 *:22 *:*
LISTEN 0 100 127.0.0.1:25 *:*
LISTEN 0 128 :::80 :::*
LISTEN 0 128 :::22 :::*
LISTEN 0 100 ::1:25 :::*
[root@WangJiayue ~]# systemctl stop httpd.service
[root@WangJiayue ~]# ss -antl
State Recv-Q Send-Q Local Address:Port Peer Address:Port
LISTEN 0 128 *:22 *:*
LISTEN 0 100 127.0.0.1:25 *:*
LISTEN 0 128 :::22 :::*
LISTEN 0 100 ::1:25 :::*
[root@WangJiayue ~]# ip a
1: lo: mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
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: ens33: mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 00:0c:29:2d:7e:16 brd ff:ff:ff:ff:ff:ff
inet 192.168.222.139/24 brd 192.168.222.255 scope global noprefixroute dynamic ens33
valid_lft 1207166sec preferred_lft 1207166sec
inet6 fe80::eb84:2580:b52e:79ad/64 scope link noprefixroute
valid_lft forever preferred_lft forever
[root@WangJiayue-2 ~]# ss -antl
State Recv-Q Send-Q Local Address:Port Peer Address:Port
LISTEN 0 128 *:22 *:*
LISTEN 0 100 127.0.0.1:25 *:*
LISTEN 0 128 :::80 :::*
LISTEN 0 128 :::22 :::*
LISTEN 0 100 ::1:25 :::*
[root@WangJiayue-2 ~]# ip a
1: lo: mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
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: ens33: mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 00:0c:29:9e:71:dd brd ff:ff:ff:ff:ff:ff
inet 192.168.222.133/24 brd 192.168.222.255 scope global noprefixroute dynamic ens33
valid_lft 1207088sec preferred_lft 1207088sec
inet 192.168.222.250/32 scope global ens33
valid_lft forever preferred_lft forever
inet6 fe80::6397:ea5d:6e1d:bc45/64 scope link noprefixroute
valid_lft forever preferred_lft forever
[root@WangJiayue-2 ~]# systemctl stop httpd.service
[root@WangJiayue-2 ~]# ss -antl
State Recv-Q Send-Q Local Address:Port Peer Address:Port
LISTEN 0 128 *:22 *:*
LISTEN 0 100 127.0.0.1:25 *:*
LISTEN 0 128 :::22 :::*
LISTEN 0 100 ::1:25 :::*
[root@WangJiayue-2 ~]# ip a
1: lo: mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
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: ens33: mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 00:0c:29:2d:7e:16 brd ff:ff:ff:ff:ff:ff
inet 192.168.222.133/24 brd 192.168.222.255 scope global noprefixroute dynamic ens33
valid_lft 1206966sec preferred_lft 1206966sec
inet6 fe80::eb84:2580:b52e:79ad/64 scope link noprefixroute
valid_lft forever preferred_lft forever
[root@WangJiayue ~]# ip a
1: lo: mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
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: ens33: mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 00:0c:29:9e:71:dd brd ff:ff:ff:ff:ff:ff
inet 192.168.222.139/24 brd 192.168.222.255 scope global noprefixroute dynamic ens33
valid_lft 1206877sec preferred_lft 1206877sec
inet 192.168.222.250/32 scope global ens33
valid_lft forever preferred_lft forever
inet6 fe80::6397:ea5d:6e1d:bc45/64 scope link noprefixroute
valid_lft forever preferred_lft forever
[root@WangJiayue ~]# ss -antl
State Recv-Q Send-Q Local Address:Port Peer Address:Port
LISTEN 0 128 *:22 *:*
LISTEN 0 100 127.0.0.1:25 *:*
LISTEN 0 128 :::80 :::*
LISTEN 0 128 :::22 :::*
LISTEN 0 100 ::1:25 :::*
在高可用(HA)系统中,当联系2个节点的“心跳线”断开时,本来为一整体、动作协调的HA系统,就分裂成为2个独立的个体。由于相互失去了联系,都以为是对方出了故障。两个节点上的HA软件像“裂脑人”一样,争抢“共享资源”、争起“应用服务”,就会发生严重后果——或者共享资源被瓜分、2边“服务”都起不来了;或者2边“服务”都起来了,但同时读写“共享存储”,导致数据损坏(常见如数据库轮询着的联机日志出错)。
对付HA系统“裂脑”的对策,目前达成共识的的大概有以下几条:
★添加冗余的心跳线,例如:双线条线(心跳线也HA),尽量减少“裂脑”发生几率;
★启用磁盘锁。正在服务一方锁住共享磁盘,“裂脑”发生时,让对方完全“抢不走”共享磁盘资源。但使用锁磁盘也会有一个不小的问题,如果占用共享盘的一方不主动“解锁”,另一方就永远得不到共享磁盘。现实中假如服务节点突然死机或崩溃,就不可能执行解锁命令。后备节点也就接管不了共享资源和应用服务。于是有人在HA中设计了“智能”锁。即:正在服务的一方只在发现心跳线全部断开(察觉不到对端)时才启用磁盘锁。平时就不上锁了。
★设置仲裁机制。例如设置参考IP(如网关IP),当心跳线完全断开时,2个节点都各自ping一下参考IP,不通则表明断点就出在本端。不仅“心跳”、还兼对外“服务”的本端网络链路断了,即使启动(或继续)应用服务也没有用了,那就主动放弃竞争,让能够ping通参考IP的一端去起服务。更保险一些,ping不通参考IP的一方干脆就自我重启,以彻底释放有可能还占用着的那些共享资源。
一般来说,脑裂的发生,有以下几种原因:
★高可用服务器对之间心跳线链路发生故障,导致无法正常通信
★高可用服务器上开启了 iptables防火墙阻挡了心跳消息传输
★高可用服务器上心跳网卡地址等信息配置不正确,导致发送心跳失败
★其他服务配置不当等原因,如心跳方式不同,心跳广插冲突、软件Bug等
可能发生的故障:
因心跳线坏了(包括断了,老化)
因网卡及相关驱动坏了,ip配置及冲突问题(网卡直连)
因心跳线间连接的设备故障(网卡及交换机)
因仲裁的机器出问题(采用仲裁的方案)
注意:Keepalived配置里同一 VRRP实例如果 virtual_router_id两端参数配置不一致也会导致裂脑问题发生。
在实际生产环境中,我们可以从以下几个方面来防止裂脑问题的发生:
★同时使用串行电缆和以太网电缆连接,同时用两条心跳线路,这样一条线路坏了,另一个还是好的,依然能传送心跳消息
★当检测到裂脑时强行关闭一个心跳节点(这个功能需特殊设备支持,如Stonith、feyce)。相当于备节点接收不到心跳消患,通过单独的线路发送关机命令关闭主节点的电源
★做好对裂脑的监控报警(如邮件及手机短信等或值班).在问题发生时人为第一时间介入仲裁,降低损失。例如,百度的监控报警短信就有上行和下行的区别。报警消息发送到管理员手机上,管理员可以通过手机回复对应数字或简单的字符串操作返回给服务器.让服务器根据指令自动处理相应故障,这样解决故障的时间更短.
当然,在实施高可用方案时,要根据业务实际需求确定是否能容忍这样的损失。对于一般的网站常规业务.这个损失是可容忍的
对脑裂的监控应在备用服务器上进行,通过添加zabbix自定义监控进行。
监控什么信息呢?监控备上有无VIP地址
备机上出现VIP有两种情况:
★发生了脑裂
★正常的主备切换
监控只是监控发生脑裂的可能性,不能保证一定是发生了脑裂,因为正常的主备切换VIP也是会到备上的。
监控脚本如下:
[root@WangJiayue ~]# mkdir -p /scripts && cd /scripts
[root@WangJiayue scripts]# vim check_keepalived.sh
#!/bin/bash
if [ `ip a show ens33 |grep 192.168.222..250|wc -l` -ne 0 ]
then
echo "keepalived is error!"
else
echo "keepalived is OK !"
fi
编写脚本时要注意,网卡要改成你自己的网卡名称,VIP也要改成你自己的VIP,最后不要忘了给脚本赋予执行权限,且要修改/scripts目录的属主属组为zabbix
环境
系统 | IP | 所需服务 |
---|---|---|
WangJiayue(CentOS7) | 192.168.222.139 | lamp、zabbix-server、zabbix-agentd |
WangJiayue-2(CentOS7 | 192.168.222.133 | keepalived |
WangJiayue-3(CentOS7 | 192.168.222.134 | keepalived、zabbix-agentd |
需要提前搭建好:LAMP、zabbix
//在备keepalived上写监控脚本
[root@WangJiayue-3 ~]# cd /scripts/
[root@WangJiayue-3 scripts]# vim check_keepalived.sh
#!/bin/bash
if [ `ip a show ens33 |grep 192.168.222.250|wc -l` -ne 0 ]
then
echo 1
else
echo 0
fi
//给脚本赋予执行权限
[root@WangJiayue-3 scripts]# chmod +x check_keepalived.sh
[root@WangJiayue-3 scripts]# ll
total 8
-rwxr-xr-x 1 root root 126 Aug 6 10:50 check_keepalived.sh
-rwxr-xr-x 1 root root 666 Aug 5 11:42 notify.sh
//修改/scripts目录的属主属组为zabbix
[root@WangJiayue-3 scripts]# chown zabbix.zabbix check_keepalived.sh
[root@WangJiayue-3 scripts]# ll
total 8
-rwxr-xr-x 1 zabbix zabbix 126 Aug 6 10:50 check_keepalived.sh
-rwxr-xr-x 1 root root 666 Aug 5 11:42 notify.sh
//把脚本写入配置文件
[root@WangJiayue-3 ~]# vim /usr/local/etc/zabbix_agentd.conf
UnsafeUserParameters=1
UserParameter=check_keepalived,/bin/bash /scripts/check_keepalived.sh
//重启服务
[root@WangJiayue-3 ~]# pkill zabbix
[root@WangJiayue-3 ~]# zabbix_agentd
修改备上的配置文件里的virtual_router_id,使其不一致,触发脑裂
[root@WangJiayue-3 ~]# vim /etc/keepalived/keepalived.conf
virtual_router_id 62
[root@WangJiayue-3 ~]# systemctl restart keepalived