一 环境介绍

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

2.服务
keepalived+nginx双主高可用负载均衡集群及LAMP应用
keepalived-1.2.13-7.el7.x86_64
nginx-1.10.2-1.el7.x86_64
httpd-2.4.6-45.el7.centos.x86_64

二 原理及拓扑图

1.vrrp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二种：
 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.nginx反代
nginx是以反向代理的方式进行负载均衡的。反向代理（Reverse Proxy）方式是指以代理服务器来接受Internet上的连接请求，然后将请求转发给内部网络上的服务器，并将从服务器上得到的结果返回给Internet上请求连接的客户端，此时代理服务器对外就表现为一个服务器。(为了理解反向代理，这里插播一条什么是正向代理：正向代理指的是，一个位于客户端和原始服务器之间的服务器，为了从原始服务器取得内容，客户端向代理发送一个请求并指定目标(原始服务器)，然后代理向原始服务器转交请求并将获得的内容返回给客户端。)
3.拓扑图

三 配置

1.后端RS配置

1
2
[root@inode4 ~]# yum install httpd -y
[root@inode5 ~]# yum install httpd -y

2.Nginx反代配置
MASTER:

1
2
3
4
5
6
7
8
9
10
upstream websrvs {
 server 172.18.67.11:80;
 server 172.18.67.12:80;
 server 127.0.0.1:80 backup;
}
server {
    listen       80 ;
    location / {
    proxy_pass http://websrvs;
    }

BACKUP:

1
2
3
4
5
6
7
8
9
10
upstream websrvs {
 server 172.18.67.11:80;
 server 172.18.67.12:80;
 server 127.0.0.1:80 backup;
}
server {
    listen       80 ;
    location / {
    proxy_pass http://websrvs;
    }

3.keepalived高可用配置
MASTER:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
! Configuration File for keepalived
global_defs {
    notification_email {
 root@localhost
    }
    notification_email_from keepalived@localhost
    smtp_server 127.0.0.1
    smtp_connect_timeout 30
    router_id node1
    vrrp_mcast_group4 224.0.67.67
}
vrrp_script chk_down {
    script "[[ -f /etc/keepalived/down ]] && exit 1 || exit 0"
    interval 1
    weight -5
}
vrrp_script chk_nginx {
    script "killall -0 nginx && exit 0 || exit 1"
    interval 1
    weight -5
    fall 2
    rise 1
}
vrrp_instance myr {
    state MASTER
    interface eno16777736
    virtual_router_id 167
    priority 100
    advert_int 1
    authentication {
 auth_type PASS
 auth_pass 571f97b2
    }
    virtual_ipaddress {
 172.18.67.33/16 dev eno16777736
    }
    track_script {
 chk_down
 chk_nginx
    }
    notify_master "/etc/keepalived/notify.sh master"
    notify_backup "/etc/keepalived/notify.sh backup"
    notify_fault "/etc/keepalived/notify.sh fault"
}

BACKUP:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
! Configuration File for keepalived
global_defs {
    notification_email {
 root@localhost
    }
    notification_email_from keepalived@localhost
    smtp_server 127.0.0.1
    smtp_connect_timeout 30
    router_id node1
    vrrp_mcast_group4 224.0.67.67
}
vrrp_script chk_down {
    script "[[ -f /etc/keepalived/down ]] && exit 1 || exit 0"
    interval 1
    weight -5
}
vrrp_script chk_nginx {
    script "killall -0 nginx && exit 0 || exit 1"
    interval 1
    weight -5
    fall 2
    rise 1
}
vrrp_instance myr {
    state BACKUP
    interface eno16777736
    virtual_router_id 167
    priority 95
    advert_int 1
    authentication {
 auth_type PASS
 auth_pass 571f97b2
    }
    virtual_ipaddress {
 172.18.67.33/16 dev eno16777736
    }
    track_script {
 chk_down
 chk_nginx
    }
    notify_master "/etc/keepalived/notify.sh master"
    notify_backup "/etc/keepalived/notify.sh backup"
    notify_fault "/etc/keepalived/notify.sh fault"
}

4.通知脚本示例

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
[root@inode2 nginx]# 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 启动服务并测试

1.启动后端web服务器

1
2
[root@inode4 ~]# systemctl start httpd
[root@inode5 ~]# systemctl start httpd

为了测试显示效果明显一点，自定义一个访问页面

1
2
[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

2.测试
MASTER:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
[root@inode2 ~]# systemctl start  keepalived
[root@inode2 ~]# systemctl status -l  keepalived
● keepalived.service - LVS and VRRP High Availability Monitor
   Loaded: loaded (/usr/lib/systemd/system/keepalived.service; disabled; vendor preset: disabled)
   Active: active (running) since Mon 2017-05-15 15:45:20 CST; 3s ago
  Process: 20971 ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
 Main PID: 20972 (keepalived)
   CGroup: /system.slice/keepalived.service
           ├─20972 /usr/sbin/keepalived -D
           ├─20973 /usr/sbin/keepalived -D
           └─20974 /usr/sbin/keepalived -D
May 15 15:45:20 inode2 Keepalived_healthcheckers[20973]: Opening file '/etc/keepalived/keepalived.conf'.
May 15 15:45:20 inode2 Keepalived_healthcheckers[20973]: Configuration is using : 7521 Bytes
May 15 15:45:20 inode2 Keepalived_healthcheckers[20973]: Using LinkWatch kernel netlink reflector...
May 15 15:45:20 inode2 Keepalived_vrrp[20974]: VRRP_Script(chk_nginx) succeeded
May 15 15:45:21 inode2 Keepalived_vrrp[20974]: VRRP_Instance(myr) Transition to MASTER STATE
May 15 15:45:22 inode2 Keepalived_vrrp[20974]: VRRP_Instance(myr) Entering MASTER STATE
May 15 15:45:22 inode2 Keepalived_vrrp[20974]: VRRP_Instance(myr) setting protocol VIPs.
May 15 15:45:22 inode2 Keepalived_vrrp[20974]: VRRP_Instance(myr) Sending gratuitous ARPs on eno16777736 for 172.18.67.33
May 15 15:45:22 inode2 Keepalived_vrrp[20974]: Opening script file /etc/keepalived/notify.sh
May 15 15:45:22 inode2 Keepalived_healthcheckers[20973]: Netlink reflector reports IP 172.18.67.33 added
[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 172.18.67.33/16 scope global secondary eno16777736
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe8b:86f/64 scope link tentative dadfailed
       valid_lft forever preferred_lft forever

主节点启动，Entering MASTER STATE，此时我们在客户端进行测试访问

1
2
3
4
5
[root@inode1 ~]# for i in {1..4};do curl http://172.18.67.33;done
RS1:172.18.67.11
RS2:172.18.67.12
RS1:172.18.67.11
RS2:172.18.67.12

访问正常，接下来我们启动备用节点的服务器

BACKUP:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
[root@inode3 keepalived]# systemctl start keepalived
[root@inode3 keepalived]# systemctl status -l keepalived
● keepalived.service - LVS and VRRP High Availability Monitor
   Loaded: loaded (/usr/lib/systemd/system/keepalived.service; disabled; vendor preset: disabled)
   Active: active (running) since Mon 2017-05-15 15:46:51 CST; 3s ago
  Process: 24329 ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
 Main PID: 24330 (keepalived)
   CGroup: /system.slice/keepalived.service
           ├─24330 /usr/sbin/keepalived -D
           ├─24331 /usr/sbin/keepalived -D
           └─24332 /usr/sbin/keepalived -D
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: Registering Kernel netlink command channel
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: Registering gratuitous ARP shared channel
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: Opening file '/etc/keepalived/keepalived.conf'.
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: Configuration is using : 66427 Bytes
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: Using LinkWatch kernel netlink reflector...
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: VRRP_Instance(myr) Entering BACKUP STATE
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: Opening script file /etc/keepalived/notify.sh
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: VRRP sockpool: [ifindex(2), proto(112), unicast(0), fd(10,11)]
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: VRRP_Script(chk_down) succeeded
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: VRRP_Script(chk_nginx) succeeded
[root@inode3 keepalived]# 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
    inet6 fe80::20c:29ff:fe78:24c3/64 scope link tentative dadfailed
       valid_lft forever preferred_lft forever

此时，我们可以看到备用节点服务器启动后进入了BACKUP状态，Entering BACKUP STATE。接下来我们测试主节点宕机的情形下，我们的服务是否还可用

1
[root@inode2 ~]# systemctl stop keepalived

主节点宕机后我们查看备用节点的状态

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
[root@inode3 keepalived]# systemctl status -l keepalived
● keepalived.service - LVS and VRRP High Availability Monitor
   Loaded: loaded (/usr/lib/systemd/system/keepalived.service; disabled; vendor preset: disabled)
   Active: active (running) since Mon 2017-05-15 15:46:51 CST; 2min 19s ago
  Process: 24329 ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
 Main PID: 24330 (keepalived)
   CGroup: /system.slice/keepalived.service
           ├─24330 /usr/sbin/keepalived -D
           ├─24331 /usr/sbin/keepalived -D
           └─24332 /usr/sbin/keepalived -D
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: VRRP sockpool: [ifindex(2), proto(112), unicast(0), fd(10,11)]
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: VRRP_Script(chk_down) succeeded
May 15 15:46:51 inode3 Keepalived_vrrp[24332]: VRRP_Script(chk_nginx) succeeded
May 15 15:48:35 inode3 Keepalived_vrrp[24332]: VRRP_Instance(myr) Transition to MASTER STATE
May 15 15:48:36 inode3 Keepalived_vrrp[24332]: VRRP_Instance(myr) Entering MASTER STATE
May 15 15:48:36 inode3 Keepalived_vrrp[24332]: VRRP_Instance(myr) setting protocol VIPs.
May 15 15:48:36 inode3 Keepalived_vrrp[24332]: VRRP_Instance(myr) Sending gratuitous ARPs on eno16777736 for 172.18.67.33
May 15 15:48:36 inode3 Keepalived_vrrp[24332]: Opening script file /etc/keepalived/notify.sh
May 15 15:48:36 inode3 Keepalived_healthcheckers[24331]: Netlink reflector reports IP 172.18.67.33 added
May 15 15:48:41 inode3 Keepalived_vrrp[24332]: VRRP_Instance(myr) Sending gratuitous ARPs on eno16777736 for 172.18.67.33
[root@inode3 keepalived]# 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 172.18.67.33/16 scope global secondary eno16777736
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe78:24c3/64 scope link tentative dadfailed
       valid_lft forever preferred_lft forever

我们发现备用节点由备用状态进入了主状态，并且IP地址也成功绑定至备用节点下。再次进行测试访问

1
2
3
4
5
[root@inode1 ~]# for i in {1..4};do curl http://172.18.67.33;done
RS1:172.18.67.11
RS2:172.18.67.12
RS1:172.18.67.11
RS2:172.18.67.12

测试一台web服务器宕机

1
2
3
4
5
6
[root@inode4 ~]# systemctl stop httpd
[root@inode1 ~]# for i in {1..4};do curl http://172.18.67.33;done
RS2:172.18.67.12
RS2:172.18.67.12
RS2:172.18.67.12
RS2:172.18.67.12

在实际生产环境中后端两台web服务器的内容应该一样的，在这里我们可认为客户端已成功访问到服务器，因此我们可认为这样的架构体现了高可用负载均衡。

Keepalived+Nginx实现高可用负载均衡集群的更多相关文章

  1. LVS+Keepalived+Nginx+Tomcat高可用负载均衡集群配置（DR模式，一个VIP，多个端口）

    一．概述 LVS作用:实现负载均衡 Keepalived作用:监控集群系统中各个服务节点的状态,HA cluster. 配置LVS有两种方式: 1. 通过ipvsadm命令行方式配置 2. 通过Red ...

  2. LVS+Keepalived搭建MyCAT高可用负载均衡集群

    LVS+Keepalived 介绍 LVS LVS是Linux Virtual Server的简写,意即Linux虚拟服务器,是一个虚拟的服务器集群系统.本项目在1998年5月由章文嵩博士成立,是中国 ...

  3. Haproxy+Keepalived搭建Weblogic高可用负载均衡集群

    配置环境说明: KVM虚拟机配置 用途 数量 IP地址 机器名 虚拟IP地址 硬件 内存3G  系统盘20G cpu 4核 Haproxy keepalived 2台 192.168.1.10 192 ...

  4. Keepalived+LVS(dr)高可用负载均衡集群的实现

    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 ...

  5. docker下用keepalived+Haproxy实现高可用负载均衡集群

    启动keepalived后宿主机无法ping通用keepalived,报错: [root@localhost ~]# ping 172.18.0.15 PING () bytes of data. F ...

  6. Linux 笔记 - 第十八章 Linux 集群之（三）Keepalived+LVS 高可用负载均衡集群

    一.前言 前两节分别介绍了 Linux 的高可用集群和负载均衡集群,也可以将这两者相结合,即 Keepalived+LVS 组成的高可用负载均衡集群,Keepalived 加入到 LVS 中的原因有以 ...

  7. 测试LVS+Keepalived高可用负载均衡集群

    测试LVS+Keepalived高可用负载均衡集群 1. 启动LVS高可用集群服务 此时查看Keepalived服务的系统日志信息如下: [root@localhost ~]# tail -f /va ...

  8. Dubbo入门到精通学习笔记（二十）：MyCat在MySQL主从复制的基础上实现读写分离、MyCat 集群部署(HAProxy + MyCat)、MyCat 高可用负载均衡集群Keepalived

    文章目录 MyCat在MySQL主从复制的基础上实现读写分离 一.环境 二.依赖课程 三.MyCat 介绍 ( MyCat 官网:http://mycat.org.cn/ ) 四.MyCat 的安装 ...

  9. LVS+Keepalived 实现高可用负载均衡集群

    LVS+Keepalived  实现高可用负载均衡集群     随着网站业务量的增长,网站的服务器压力越来越大?需要负载均衡方案!商业的硬件如 F5 ,Array又太贵,你们又是创业型互联公司如何有效 ...

随机推荐

  1. FreeImage库如何转换图片格式？

    FreeImage下载地址:http://freeimage.sourceforge.net/ //freeimagemain.h #ifndef FREEIMAGEMAIN_H #define FR ...

  2. SpringBoot+Mybatis+PageHelper简化分页实现

    前言 经过一段时间的测试和修改PageHelper插件逐渐走到了让我觉得靠谱的时候,它功能的就是简化分页的实现,让分页不需要麻烦的多写很多重复的代码. 已经加入我的github模版中:https:// ...

  3. MySQL Crash Errcode: 28 - No space left on device

    一台MySQL服务器突然Crash了,检查进程 ps -ef | grep -i mysql 发现mysqld进程已经没有了, 检查错误日志时发现MySQL确实Crash了.具体如下所示: 注意日志中 ...

  4. 3.3.4 PCI设备进行DMA写时发生Cache命中

    如果PCI设备访问的地址在某个CPU的Cache行中命中时,可能会出现三种情况. 第一种情况是命中的Cache行其状态为E,即Cache行中的数据与存储器中的数据一致:而第二种情况是命中的Cache行 ...

  5. RFM模型+SOM聚类︱离群值筛选问题

        笔者寄语:一般情况下离群值不应该直接删除,应该进行筛选,然后进行专门的离群值分析.笔者在这进行一下思考,在聚类基础之上的一种离群点检验. 基于聚类的离群点检测的步骤如下:数据标准化--聚类-- ...

  6. Java Web项目（Extjs）报错五

    1. Java Web项目(Extjs)报错五 具体报错如下: usage: java org.apache.catalina.startup.Catalina [ -config {pathname ...

  7. 使用VLC和live555MediaServer搭建RTSP服务器

    一.使用VLC搭建 服务端 1.打开"媒体--流"选项,如图所示:  2.点击"添加"按钮选择流媒体文件,然后点击"串流"按钮  3.点击& ...

  8. Flex动态获取方法报错

    1.错误描述 2.错误原因 由于Flex文件修改后,需要将其编译成swf文件,刚修改的方法没有编译,再加上历史缓存的原因,导致报错 3.解决办法 将Flex项目重新clean一下,并将MyEclips ...

  9. STM32f4 ARM Bootloader

    参考资料: 基于ARM 的嵌入式系统Bootloader 启动流程分析 Bootloader 详解 ( 代码环境 | ARM 启动流程 | uboot 工作流程 | 架构设计) Android系统启动 ...

  10. DirectSound---音效使用

    DirectSound以DMO(DirectX Nedua Objects)的方式提供了对原始音频数据的处理,开发者能够通过VC6.0里的Audio Effect DMO Wizard来开发自定义的D ...