企业业务量比较小的时候,单台服务器就可以满足业务需要了。但是随着业务发展,单服务器的问题就凸显出来了:
- 当服务器挂掉时,业务就会中断
- 当业务量增加,单台服务器性能变差,如何透明的扩展服务器和带宽,增加服务器吞吐量
负载均衡器可以解决以上问题
本文会根据拓扑图,用haproxy和keepalived搭建一个负载均衡器
1 负载均衡器拓扑图
2 准备
2.1 准备环境
准备5台CentOS7.3主机和一个VIP地址:
- 准备一个可用IP用作虚拟IP(VIP):
VIP: 192.168.1.100
- 负载均衡器会用到2台主机,一主一备的架构
lb1(默认为主): 192.168.1.101
lb2(默认为备): 192.168.1.102
- 后端服务器集群中主机的IP地址
s1: 192.168.1.2
s2: 192.168.1.3
s3: 192.168.1.4
2.2 主机配置
2.2.1 所有主机上关闭防火墙
- systemctl stop firewalld
- systemctl disable firewalld
2.2.2 所有主机关闭selinux
- setenforce 0
- vi /etc/selinux/config
- SELINUX=disabled
2.3 安装haproxy和keepalived
lb1和lb2上安装haproxy和keepalived
- yum install haproxy keepalived -y
2.4 安装nginx(有其他后端测程序,可省略此步)
s1 s2 s3上安装nginx,目的是把nginx作为后端,如果有其他后端程序,这一步可以省略
- yum install epel-release -y
- yum install nginx -y
2.3 配置keepalived
KeepAlived是基于VRRP(Virtual Router Redundancy Protocol,虚拟路由冗余协议)实现的一个高可用方案,通过VIP(虚拟IP)和心跳检测来实现高可用
Keepalived有两个角色,Master和Backup。一般会是1个Master,多个Backup。
Master会绑定VIP到自己网卡上,对外提供服务。Master和Backup会定时确定对方状态,当Master不可用的时候,Backup会通知网关,并把VIP绑定到自己的网卡上,实现服务不中断,高可用
2.3.1 配置Master
- ! Configuration File for keepalived
- global_defs {
- # 通知邮件服务器的配置
- notification_email {
- # 当master失去VIP或则VIP的时候,会发一封通知邮件到your-email@qq.com
- your-email@qq.com
- }
- # 发件人信息
- notification_email_from keepalived@qq.com
- # 邮件服务器地址
- smtp_server 127.0.0.1
- # 邮件服务器超时时间
- smtp_connect_timeout 30
- # 邮件TITLE
- router_id LVS_DEVEL
- }
- vrrp_instance VI_1 {
- # 主机: MASTER
- # 备机: BACKUP
- state MASTER
- # 实例绑定的网卡, 用ip a命令查看网卡编号
- interface eno16777984
- # 虚拟路由标识,这个标识是一个数字(1-255),在一个VRRP实例中主备服务器ID必须一样
- virtual_router_id 88
- # 优先级,数字越大优先级越高,在一个实例中主服务器优先级要高于备服务器
- priority 100
- # 主备之间同步检查的时间间隔单位秒
- advert_int 1
- # 验证类型和密码
- authentication {
- # 验证类型有两种 PASS和HA
- auth_type PASS
- # 验证密码,在一个实例中主备密码保持一样
- auth_pass 11111111
- }
- # 虚拟IP地址,可以有多个,每行一个
- virtual_ipaddress {
- 192.168.1.100
- }
- }
- virtual_server 192.168.1.100 443 {
- # 健康检查时间间隔
- delay_loop 6
- # 调度算法
- # Doc: http://www.keepalived.org/doc/scheduling_algorithms.html
- # Round Robin (rr)
- # Weighted Round Robin (wrr)
- # Least Connection (lc)
- # Weighted Least Connection (wlc)
- # Locality-Based Least Connection (lblc)
- # Locality-Based Least Connection with Replication (lblcr)
- # Destination Hashing (dh)
- # Source Hashing (sh)
- # Shortest Expected Delay (seq)
- # Never Queue (nq)
- # Overflow-Connection (ovf)
- lb_algo rr
- lb_kind NAT
- persistence_timeout 50
- protocol TCP
- # 通过调度算法把Master切换到真实的负载均衡服务器上
- # 真实的主机会定期确定进行健康检查,如果MASTER不可用,则切换到备机上
- real_server 192.168.1.101 443 {
- weight 1
- TCP_CHECK {
- # 连接超端口
- connect_port 443
- # 连接超时时间
- connect_timeout 3
- }
- }
- real_server 192.168.1.102 443 {
- weight 1
- TCP_CHECK {
- connect_port 443
- connect_timeout 3
- }
- }
- }
2.3.2 配置BACKUP
编辑lb2(192.168.1.102)上的/etc/keepalived/keepalived.conf
- ! Configuration File for keepalived
- global_defs {
- # 通知邮件服务器的配置
- notification_email {
- # 当master失去VIP或则VIP的时候,会发一封通知邮件到your-email@qq.com
- your-email@qq.com
- }
- # 发件人信息
- notification_email_from keepalived@qq.com
- # 邮件服务器地址
- smtp_server 127.0.0.1
- # 邮件服务器超时时间
- smtp_connect_timeout 30
- # 邮件TITLE
- router_id LVS_DEVEL
- }
- vrrp_instance VI_1 {
- # 主机: MASTER
- # 备机: BACKUP
- state BACKUP
- # 实例绑定的网卡, 用ip a命令查看网卡编号
- interface eno16777984
- # 虚拟路由标识,这个标识是一个数字(1-255),在一个VRRP实例中主备服务器ID必须一样
- virtual_router_id 88
- # 优先级,数字越大优先级越高,在一个实例中主服务器优先级要高于备服务器
- priority 99
- # 主备之间同步检查的时间间隔单位秒
- advert_int 1
- # 验证类型和密码
- authentication {
- # 验证类型有两种 PASS和HA
- auth_type PASS
- # 验证密码,在一个实例中主备密码保持一样
- auth_pass 11111111
- }
- # 虚拟IP地址,可以有多个,每行一个
- virtual_ipaddress {
- 192.168.1.100
- }
- }
- virtual_server 192.168.1.100 443 {
- # 健康检查时间间隔
- delay_loop 6
- # 调度算法
- # Doc: http://www.keepalived.org/doc/scheduling_algorithms.html
- # Round Robin (rr)
- # Weighted Round Robin (wrr)
- # Least Connection (lc)
- # Weighted Least Connection (wlc)
- # Locality-Based Least Connection (lblc)
- # Locality-Based Least Connection with Replication (lblcr)
- # Destination Hashing (dh)
- # Source Hashing (sh)
- # Shortest Expected Delay (seq)
- # Never Queue (nq)
- # Overflow-Connection (ovf)
- lb_algo rr
- lb_kind NAT
- persistence_timeout 50
- protocol TCP
- # 通过调度算法把Master切换到真实的负载均衡服务器上
- # 真实的主机会定期确定进行健康检查,如果MASTER不可用,则切换到备机上
- real_server 192.168.1.101 443 {
- weight 1
- TCP_CHECK {
- # 连接超端口
- connect_port 443
- # 连接超时时间
- connect_timeout 3
- }
- }
- real_server 192.168.1.102 443 {
- weight 1
- TCP_CHECK {
- connect_port 443
- connect_timeout 3
- }
- }
- }
2.4 配置haproxy
编辑lb1(192.168.1.101)和lb2(192.168.1.102)上的/etc/haproxy/haproxy.cfg
把后端服务器IP(192.168.1.2, 192.168.1.3, 192.168.1.4)加到backend里
- #---------------------------------------------------------------------
- # Global settings
- #---------------------------------------------------------------------
- global
- log 127.0.0.1 local2
- chroot /var/lib/haproxy
- pidfile /var/run/haproxy.pid
- maxconn 4096
- user haproxy
- group haproxy
- daemon
- # turn on stats unix socket
- stats socket /var/lib/haproxy/stats
- listen stats
- bind *:9000
- mode http
- stats enable
- stats hide-version
- stats uri /stats
- stats refresh 30s
- stats realm Haproxy\ Statistics
- stats auth admin:admin
- frontend k8s-api
- bind *:443
- mode tcp
- option tcplog
- tcp-request inspect-delay 5s
- tcp-request content accept if { req_ssl_hello_type 1 }
- default_backend k8s-api-backend
- backend k8s-api-backend
- mode tcp
- option tcplog
- option tcp-check
- balance roundrobin
- server master1 192.167.1.2:80 maxconn 1024 weight 5 check
- server master2 192.167.1.3:80 maxconn 1024 weight 5 check
- server master3 192.167.1.4:80 maxconn 1024 weight 5 check
2.5 配置nginx
给nginx添加SSL证书,配置过程略
- vi /usr/share/nginx/html/index.html
把index.html里面字符串Welcome to nginx改成Welcome to nginx HA
3 启动服务
3.1 启动nginx
- sudo systemctl start nginx
- sudo systemctl enable nginx
3.2 启动haproxy
- sudo systemctl start haproxy
- sudo systemctl enable haproxy
3.3 启动keepalived
- sudo systemctl start keepalived
- sudo systemctl enable keepalived
在MASTER上运行ip a
- eno16777984: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000
- link/ether 00:xx:xx:xx:3d:0c brd ff:ff:ff:ff:ff:ff
- inet 192.168.1.101/24 brd 192.168.1.255 scope global eno16777984
- valid_lft forever preferred_lft forever
- inet 192.168.1.100/32 scope global eno16777984
- valid_lft forever preferred_lft forever
- inet6 eeee:eeee:1c9d:2009:250:56ff:fe9c:3d0c/64 scope global noprefixroute dynamic
- valid_lft 7171sec preferred_lft 7171sec
- inet6 eeee::250:56ff:eeee:3d0c/64 scope link
- valid_lft forever preferred_lft forever
会发现VIP(192.168.1.100)已经绑定好了
- inet 192.168.1.100/32 scope global eno16777984
- valid_lft forever preferred_lft forever
如果发现VIP无法绑定
- vi /etc/sysctl.conf
添加两行
- net.ipv4.ip_forward = 1
- net.ipv4.ip_nonlocal_bind = 1
让新配置生效
- sysctl -p
4 验证
4.1 查看状态
1. 在浏览器输入 http://192.168.1.100:9000/stats 查看haproxy状态
2. 在浏览器输入 https://192.168.1.100 查看服务状态
是否成功显示为nginx欢迎页面
4.2 主备切换
1. 在浏览器输入 https://192.168.1.100 查看是否成功显示nginx欢迎页面
2. lb1(192.168.1.101)关机,查看是否还可以访问https://192.168.1.100, 如果成功,则说明VIP成功切换到备机
3. 在lb2(192.168.1.102)上执行ip a,查看网卡是否绑定VIP(192.168.1.100)
3. 启动lb1(192.168.1.101)
目的是为了验证VIP是否切回MASTER主机(因为MASTER端的配置文件中priority为100,而BACKUP为99,health check会自动把VIP绑定到priority高的主机上)。