Keepalived+LVS(dr)高可用负载均衡集群的实现

本文涉及的产品
应用型负载均衡 ALB,每月750个小时 15LCU
网络型负载均衡 NLB,每月750个小时 15LCU
传统型负载均衡 CLB,每月750个小时 15LCU
简介: 一 环境介绍1.操作系统CentOS Linux release 7.2.1511 (Core)2.服务keepalived+lvs双主高可用负载均衡集群及LAMP应用keepalived-1.2.

一 环境介绍

1.操作系统
CentOS Linux release 7.2.1511 (Core)

2.服务
keepalived+lvs双主高可用负载均衡集群及LAMP应用
keepalived-1.2.13-7.el7.x86_64
ipvsadm-1.27-7.el7.x86_64
httpd-2.4.6-45.el7.centos.x86_64
mariadb-5.5.52-1.el7.x86_64
php-5.4.16-42.el7.x86_64

二 原理及拓扑图

1.vrrp协议
vrrp(Virtual Redundant Routing Protocol)协议:
在现实的网络环境中,两台需要通信的主机大多数情况下并没有直接的物理连接。对于这样的情况,它们之间路由怎样选择?主机如何选定到达目的主机的下一跳路由,这个问题通常的解决方法有两种:
 在主机上使用动态路由协议(RIP、OSPF等)
 在主机上配置静态路由
很明显,在主机上配置动态路由是非常不切实际的,因为管理、维护成本以及是否支持等诸多问题。配置静态路由就变得十分流行,但路由器(或者说默认网关default gateway)却经常成为单点故障。VRRP的目的就是为了解决静态路由单点故障问题,VRRP通过一竞选(election)协议来动态的将路由任务交给LAN中虚拟路由器中的某台VRRP路由器。

2.keepalived简介
Keepalived 是一个基于VRRP协议来实现的LVS服务高可用方案,可以利用其来避免单点故障。一个LVS服务会有2台服务器运行Keepalived,一台为主服务器(MASTER),一台为备份服务器(BACKUP),但是对外表现为一个虚拟IP,主服务器会发送特定的消息给备份服务器,当备份服务器收不到这个消息的时候,即主服务器宕机的时候,备份服务器就会接管虚拟IP,继续提供服务,从而保证了高可用性。Keepalived是VRRP的完美实现。

3.lvs-dr
Direct Routing,直接路由,通过为请求报文重新封装一个MAC首部进行转发,源MAC是DIP所在的接口的MAC,目标MAC是某挑选出的RS的RIP所在接口的MAC地址;源IP/PORT,以及目标IP/PORT均保持不变。  
Director和各RS都得配置使用VIP; 
(1) 确保前端路由器将目标IP为VIP的请求报文发往Director;
 (a) 在前端网关做静态绑定;
 (b) 在RS上使用arptables;
 (c) 在RS上修改内核参数以限制arp通告及应答级别;
  arp_announce
  arp_ignore
(2) RS的RIP可以使用私网地址,也可以是公网地址;RIP与DIP在同一IP网络;RIP的网关不能指向DIP,以确保响应报文不会经由Director; 
(3) RS跟Director要在同一个物理网络;
(4) 请求报文要经由Director,但响应不能经由Director,而是由RS直接发往Client;
(5) 不支持端口映射。

4.IP分配
VIP1:172.18.67.66
VIP2:172.18.67.88
DIP1:172.18.67.13
DIP2:172.18.67.14
RIP1:172.18.67.11
RIP2:172.18.67.12
CIP:172.18.67.3

5.拓扑图

 

三 keepalived配置

1.安装keepalived

[root@inode2 ~]# yum install -y keepalived
[root@inode3 ~]# yum install -y keepalived

 

2.高可用的ipvs双主集群配置  
第一个节点:

[root@inode2 ~]# cd /etc/keepalived/
[root@inode2 keepalived]# vim keepalived.conf
! Configuration File for keepalived
global_defs {
notification_email {
root@localhost                                #接受通知的邮件地址
}
notification_email_from kaadmin@localhost
smtp_server 127.0.0.1
smtp_connect_timeout 30
router_id node2                                #路由节点
vrrp_mcast_group4 224.0.67.67                      #多播地址,范围224.0.0.0~239.255.255.255
}
vrrp_instance myr1 {
state MASTER
interface eno16777736                            #网卡接口
virtual_router_id 167                            #虚拟路由ID号,0~255
priority 100                                  #优先级,MASTER比BACKUP优先级高
advert_int 1
authentication {
 auth_type PASS
 auth_pass f1bf7fda
}
virtual_ipaddress {
 172.18.67.66/16 dev eno16777736 label eno16777736:0
}
track_interface {
 eno16777736
}
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"
}
vrrp_instance myr2 {
state BACKUP
interface eno16777736
virtual_router_id 168
priority 98
advert_int 1
authentication {
 auth_type PASS
 auth_pass f2bf7ade
}
virtual_ipaddress {
 172.18.67.88/16 dev eno16777736 label eno16777736:1
}
track_interface {
 eno16777736
}
notify_master "/etc/keepalived/notify.sh master"
notify_backup "/etc/keepalived/notify.sh backup"
notify_fault "/etc/keepalived/notify.sh fault"
}
virtual_server 172.18.67.66 80 {              #VIP
delay_loop 2
lb_algo wrr                            #lvs负载均衡调度算法
lb_kind DR                             #负载均衡类型
protocol TCP                            #传输协议
sorry_server 127.0.0.1 80                   #localhost
real_server 172.18.67.11 80 {                 #后端RIP
 weight 1                              #调度权重 
 HTTP_GET {                            #http请求方式
 url {
  path /
  status_code 200                       #状态码
 }
 connect_timeout 2                        #连接超时
 nb_get_retry 3
 delay_before_retry 3
 }
}
real_server 172.18.67.12 80 {
 weight 1
 HTTP_GET {
  url {
   path /
   status_code 200
  }
  connect_timeout 2
  nb_get_retry 3
  delay_before_retry 3
  }
 }
}

 

第二个节点:

[root@inode3 ~]# cd /etc/keepalived/
[root@inode3 keepalived]# vim keepalived.conf
! Configuration File for keepalived
global_defs {
notification_email {
root@localhost
}
notification_email_from kaadmin@localhost
smtp_server 127.0.0.1
smtp_connect_timeout 30
router_id node3
vrrp_mcast_group4 224.0.67.67
}
vrrp_instance myr1 {
state BACKUP
interface eno16777736
virtual_router_id 167
priority 98
advert_int 1
authentication {
 auth_type PASS
 auth_pass f1bf7fda
}
virtual_ipaddress {
 172.18.67.66/16 dev eno16777736 label eno16777736:0
}
track_interface {
 eno16777736
}
notify_master "/etc/keepalived/notify.sh master"
notify_backup "/etc/keepalived/notify.sh backup"
notify_fault "/etc/keepalived/notify.sh fault"
}
vrrp_instance myr2 {
state MASTER
interface eno16777736
virtual_router_id 168
priority 100
advert_int 1
authentication {
 auth_type PASS
 auth_pass f2bf7ade
}
virtual_ipaddress {
 172.18.67.88/16 dev eno16777736 label eno16777736:1
}
track_interface {
 eno16777736
}
notify_master "/etc/keepalived/notify.sh master"
notify_backup "/etc/keepalived/notify.sh backup"
notify_fault "/etc/keepalived/notify.sh fault"
}
virtual_server 172.18.67.88 80 {
delay_loop 2
lb_algo wrr
lb_kind DR
protocol TCP
sorry_server 127.0.0.1 80
real_server 172.18.67.11 80 {
 weight 1
 HTTP_GET {
 url {
  path /
  status_code 200
 }
 connect_timeout 2
 nb_get_retry 3
 delay_before_retry 3
 }
}
real_server 172.18.67.12 80 {
 weight 1
 HTTP_GET {
  url {
   path /
   status_code 200
  }
  connect_timeout 2
  nb_get_retry 3
  delay_before_retry 3
  }
 }
}


3.邮件通知脚本
当双主高可用集群主备切换时可通过邮件通知管理员,此时在配置文件中可自动调用实现编辑好的脚本

[root@inode2 ~]# vim notify.sh
#!/bin/bash
#
contact='root@localhost'
notify() {
 mailsubject="$(hostname) to be $1, vip floating"
 mailbody="$(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


节点二同样配置。

四 后端RS服务器的配置

1.配置LAMP环境

[root@inode4 ~]# yum install httpd mariadb-server php -y
[root@inode5 ~]# yum install httpd mariadb-server php -y

 

2.简单编辑测试网页

[root@inode4 ~]# echo "RS1:172.18.67.11" > /var/www/html/index.html
[root@inode5 ~]# echo "RS2:172.18.67.12" > /var/www/html/index.html

 

3.修改RS内核参数
dr模型中,各主机上均需要配置VIP,解决地址冲突的方式有三种:
(1) 在前端网关做静态绑定;
(2) 在各RS使用arptables;
(3) 在各RS修改内核参数,来限制arp响应和通告的级别;
限制响应级别:arp_ignore
 0:默认值,表示可使用本地任意接口上配置的任意地址进行响应;
 1: 仅在请求的目标IP配置在本地主机的接收到请求报文接口上时,才给予响应;
限制通告级别:arp_announce
 0:默认值,把本机上的所有接口的所有信息向每个接口上的网络进行通告;
 1:尽量避免向非直接连接网络进行通告;
 2:必须避免向非本网络通告。
可通过编辑脚本实现:

[root@inode4 ~]# vim dr.sh
#!/bin/bash
#
case $1 in
start)
 echo 1 > /proc/sys/net/ipv4/conf/all/arp_ignore
 echo 1 > /proc/sys/net/ipv4/conf/eth0/arp_ignore
 echo 2 > /proc/sys/net/ipv4/conf/all/arp_announce
 echo 2 > /proc/sys/net/ipv4/conf/eth0/arp_announce
 ;;
stop)
 echo 0 > /proc/sys/net/ipv4/conf/all/arp_ignore
 echo 0 > /proc/sys/net/ipv4/conf/eth0/arp_ignore
 echo 0 > /proc/sys/net/ipv4/conf/all/arp_announce
 echo 0 > /proc/sys/net/ipv4/conf/eth0/arp_announce
 ;;
*) 
 echo "Usage $(basename $0) start|stop"
 exit 1
 ;;
esac


同理第二个RS需同样配置

4.添加路由
节点一:

[root@inode4 ~]# ifconfig lo:0 172.18.67.66 netmask 255.255.255.255 broadcast 172.18.67.66 up
[root@inode4 ~]# ifconfig lo:1 172.18.67.88 netmask 255.255.255.255 broadcast 172.18.67.88 up
[root@inode4 ~]# route add -host 172.18.67.66 dev lo:0
[root@inode4 ~]# route add -host 172.18.67.88 dev lo:1


节点二:

[root@inode5 ~]# ifconfig lo:0 172.18.67.88 netmask 255.255.255.255 broadcast 172.18.67.88 up
[root@inode5 ~]# ifconfig lo:1 172.18.67.66 netmask 255.255.255.255 broadcast 172.18.67.66 up
[root@inode5 ~]# route add -host 172.18.67.88 dev lo:0
[root@inode5 ~]# route add -host 172.18.67.66 dev lo:1

 

五 测试

1.启动服务

RS:

[root@inode4 ~]# systemctl start httpd
[root@inode5 ~]# systemctl start httpd

 

节点一:

[root@inode2 ~]# systemctl start keepalived.service
[root@inode2 ~]# systemctl status -l  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 2017-05-14 01:19:27 CST; 17s ago
  Process: 2120 ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
 Main PID: 2121 (keepalived)
   CGroup: /system.slice/keepalived.service
           ├─2121 /usr/sbin/keepalived -D
           ├─2122 /usr/sbin/keepalived -D
           └─2123 /usr/sbin/keepalived -D
May 14 01:19:29 inode2 Keepalived_vrrp[2123]: Opening script file /etc/keepalived/notify.sh
May 14 01:19:29 inode2 Keepalived_healthcheckers[2122]: Netlink reflector reports IP 172.18.67.66 added
May 14 01:19:31 inode2 Keepalived_vrrp[2123]: VRRP_Instance(myr2) Transition to MASTER STATE
May 14 01:19:32 inode2 Keepalived_vrrp[2123]: VRRP_Instance(myr2) Entering MASTER STATE
May 14 01:19:32 inode2 Keepalived_vrrp[2123]: VRRP_Instance(myr2) setting protocol VIPs.
May 14 01:19:32 inode2 Keepalived_vrrp[2123]: VRRP_Instance(myr2) Sending gratuitous ARPs on eno16777736 for 172.18.67.88
May 14 01:19:32 inode2 Keepalived_vrrp[2123]: Opening script file /etc/keepalived/notify.sh
May 14 01:19:32 inode2 Keepalived_healthcheckers[2122]: Netlink reflector reports IP 172.18.67.88 added
May 14 01:19:34 inode2 Keepalived_vrrp[2123]: VRRP_Instance(myr1) Sending gratuitous ARPs on eno16777736 for 172.18.67.66
May 14 01:19:37 inode2 Keepalived_vrrp[2123]: VRRP_Instance(myr2) Sending gratuitous ARPs on eno16777736 for 172.18.67.88
[root@inode2 ~]# ip a l
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN 
    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: eno16777736: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:0c:29:8b:08:6f brd ff:ff:ff:ff:ff:ff
    inet 172.18.67.13/16 brd 172.18.255.255 scope global eno16777736
       valid_lft forever preferred_lft forever
    inet 172.18.67.66/16 scope global secondary eno16777736:0
       valid_lft forever preferred_lft forever
    inet 172.18.67.88/16 scope global secondary eno16777736:1
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe8b:86f/64 scope link tentative dadfailed 
       valid_lft forever preferred_lft forever

 

节点二:

[root@inode3 ~]# systemctl start keepalived.service
[root@inode3 ~]# systemctl status -l  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 2017-05-14 01:20:25 CST; 6s ago
  Process: 2110 ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
 Main PID: 2111 (keepalived)
   CGroup: /system.slice/keepalived.service
           ├─2111 /usr/sbin/keepalived -D
           ├─2112 /usr/sbin/keepalived -D
           └─2113 /usr/sbin/keepalived -D
May 14 01:20:25 inode3 Keepalived_vrrp[2113]: VRRP sockpool: [ifindex(2), proto(112), unicast(0), fd(10,11)]
May 14 01:20:26 inode3 Keepalived_vrrp[2113]: VRRP_Instance(myr2) Transition to MASTER STATE
May 14 01:20:26 inode3 Keepalived_vrrp[2113]: VRRP_Instance(myr2) Received lower prio advert, forcing new election
May 14 01:20:26 inode3 Keepalived_vrrp[2113]: VRRP_Instance(myr2) Received lower prio advert, forcing new election
May 14 01:20:26 inode3 Keepalived_vrrp[2113]: VRRP_Instance(myr2) Received lower prio advert, forcing new election
May 14 01:20:27 inode3 Keepalived_vrrp[2113]: VRRP_Instance(myr2) Entering MASTER STATE
May 14 01:20:27 inode3 Keepalived_vrrp[2113]: VRRP_Instance(myr2) setting protocol VIPs.
May 14 01:20:27 inode3 Keepalived_vrrp[2113]: VRRP_Instance(myr2) Sending gratuitous ARPs on eno16777736 for 172.18.67.88
May 14 01:20:27 inode3 Keepalived_vrrp[2113]: Opening script file /etc/keepalived/notify.sh
May 14 01:20:27 inode3 Keepalived_healthcheckers[2112]: Netlink reflector reports IP 172.18.67.88 added
[root@inode3 ~]# ip a l
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN 
    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: eno16777736: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:0c:29:78:24:c3 brd ff:ff:ff:ff:ff:ff
    inet 172.18.67.14/16 brd 172.18.255.255 scope global eno16777736
       valid_lft forever preferred_lft forever
    inet 172.18.67.88/16 scope global secondary eno16777736:1
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe78:24c3/64 scope link tentative dadfailed 
       valid_lft forever preferred_lft forever

 

2.访问测试

[root@inode1 ~]# curl http://172.18.67.66
RS2:172.18.67.12
[root@inode1 ~]# curl http://172.18.67.66
RS1:172.18.67.11
[root@inode1 ~]# curl http://172.18.67.66
RS2:172.18.67.12
[root@inode1 ~]# curl http://172.18.67.66
RS1:172.18.67.11
[root@inode1 ~]# curl http://172.18.67.88
RS2:172.18.67.12
[root@inode1 ~]# curl http://172.18.67.88
RS1:172.18.67.11
[root@inode1 ~]# curl http://172.18.67.88
RS2:172.18.67.12
[root@inode1 ~]# curl http://172.18.67.88
RS1:172.18.67.11


高可用负载集群成功搭建完成。

3.模拟一台web服务器宕机
将RS1的httpd服务停掉,并再此访问:

[root@inode4:~]# systemctl stop httpd
[root@inode1 ~]# curl http://172.18.67.66
RS2:172.18.67.12
[root@inode1 ~]# curl http://172.18.67.66
RS2:172.18.67.12
[root@inode1 ~]# curl http://172.18.67.88
RS2:172.18.67.12
[root@inode1 ~]# curl http://172.18.67.88
RS2:172.18.67.12


发现照样可以访问服务器,实际环境中两台RS的内容应该是一模一样的,在这里我为了以示区别将两台内容编辑成不同。

4.模拟一台高可用负载集群宕机
将VS1的keepalived服务关闭并测试:

[root@inode2 ~]# systemctl stop keepalived.service
[root@inode1 ~]# curl http://172.18.67.88
RS1:172.18.67.11
[root@inode1 ~]# curl http://172.18.67.88
RS2:172.18.67.12
[root@inode1 ~]# curl http://172.18.67.88
RS1:172.18.67.11
[root@inode1 ~]# curl http://172.18.67.88
RS2:172.18.67.12

 

我们发现即使高可用负载均衡集群中的某一个主机宕机了,我们任然可以通过其中的一个IP访问web服务器,体现出了高可用的实用性,并且在访问中lvs调度器将客户端请求按设置的权重分别向后端的服务器实现调度。

相关实践学习
SLB负载均衡实践
本场景通过使用阿里云负载均衡 SLB 以及对负载均衡 SLB 后端服务器 ECS 的权重进行修改,快速解决服务器响应速度慢的问题
负载均衡入门与产品使用指南
负载均衡(Server Load Balancer)是对多台云服务器进行流量分发的负载均衡服务,可以通过流量分发扩展应用系统对外的服务能力,通过消除单点故障提升应用系统的可用性。 本课程主要介绍负载均衡的相关技术以及阿里云负载均衡产品的使用方法。
目录
相关文章
|
1天前
|
负载均衡 算法 Linux
LVS+Keepalived:实现高效软负载均衡的利器
本文介绍了如何使用LVS(Linux Virtual Server)和Keepalived搭建高可用负载均衡集群。LVS通过不同调度算法将请求转发给后端服务器,而Keepalived基于VRRP协议实现服务高可用,避免IP单点故障。具体步骤包括环境准备、安装配置ipvsadm和Keepalived、启动服务及测试。文中还详细解释了配置文件中的关键参数,并提供了故障转移测试方法。最后,文章简要对比了软件、硬件和云负载均衡方案的特点,帮助读者选择合适的负载均衡策略。
21 4
|
3月前
|
运维 负载均衡 网络协议
LVS+Keepalived 负载均衡
LVS+Keepalived 负载均衡
94 8
LVS+Keepalived 负载均衡
|
3月前
|
域名解析 运维 负载均衡
LVS+Keepalived 负载均衡(二)28-1
【8月更文挑战第28天】LVS+Keepalived 负载均衡 配置 LVS VIP
74 5
|
4月前
|
负载均衡 算法 应用服务中间件
负载均衡技术在Web服务器集群中的应用
【8月更文第28天】随着互联网的发展和用户对Web服务需求的增长,单台服务器很难满足大规模访问的需求。为了提高系统的稳定性和扩展性,通常会采用Web服务器集群的方式。在这种架构中,负载均衡器扮演着至关重要的角色,它能够合理地分配客户端请求到不同的后端服务器上,从而实现资源的最优利用。
148 2
|
4月前
|
负载均衡 算法 关系型数据库
MySQL集群如何实现负载均衡?
【8月更文挑战第16天】MySQL集群如何实现负载均衡?
250 6
|
4月前
|
负载均衡 网络协议
使用LVS搭建集群实现负载均衡(二)安装使用
【8月更文挑战第8天】使用LVS搭建集群实现负载均衡(二)安装使用
82 5
|
4月前
|
缓存 负载均衡 算法
在Linux中, LVS负载均衡有哪些策略?
在Linux中, LVS负载均衡有哪些策略?
|
负载均衡 应用服务中间件 Linux
企业实战(13)LVS负载均衡NAT(网络地址转换)模式实战详解(一)
企业实战(13)LVS负载均衡NAT(网络地址转换)模式实战详解(一)
181 0
|
存储 负载均衡 网络协议
企业实战(13)LVS负载均衡DR(直接路由)模式实战详解(二)
企业实战(13)LVS负载均衡DR(直接路由)模式实战详解(二)
226 0
|
负载均衡 应用服务中间件 Linux
Nginx系列教程(14) - LVS+KeepAlived+Nginx实现高性能负载均衡集群
Nginx系列教程(14) - LVS+KeepAlived+Nginx实现高性能负载均衡集群
1324 0