linux集群技术（二）--keepalived（高可用集群）（二）

1. 案例1--keepalived
2. 案例2--keepalived  +  Lvs集群

1.案例1--keepalived

        1.1   环境

初识keepalived，实现web服务器的高可用集群。
    Server1:       192.168.26.144
    Server2:  192.168.26.169
           VIP:        192.168.26.190

        1.2   server1

创建etc下的keepalived目录,编辑配置文件：

# yum -y install keepalived

# vim /etc/keepalived/keepalived.conf

        ! Configuration File for keepalived
        global_defs {
         router_id 1                            #设备在组中的标识，设置不一样即可
         }

        #vrrp_script chk_nginx {                        #健康检查
        # script "/etc/keepalived/ck_ng.sh"     #检查脚本
        # interval 2                            #检查频率.秒
        # weight -5                             #priority减5
        # fall 3                                        #失败三次        
        # }

        vrrp_instance VI_1 {               #VI_1。实例名两台路由器相同。同学们要注意区分。
            state MASTER                        #主或者从状态
            interface ens32                     #监控网卡
            mcast_src_ip 192.168.0.118          #心跳源IP
            virtual_router_id 55                #虚拟路由编号，主备要一致。同学们注意区分
            priority 100                        #优先级
            advert_int 1                        #心跳间隔

            authentication {                    #秘钥认证(1-8位)
                auth_type PASS
                auth_pass 123456
            }

            virtual_ipaddress {                 #VIP
            192.168.26.190/24
                }

        #  track_script {                       #引用脚本
        #       chk_nginx
        #    }

        }

# scp -r /etc/keepalived/keepalived.conf  192.168.26.169:/etc/keepalived/

        用scp复制给server2

# systemctl enable keepalived.service
    开机启动keepalived

安装Nginx：

        rpm -ivh http://nginx.org/packages/centos/7/noarch/RPMS/nginx-release-centos-7-0.el7.ngx.noarch.rpm
        yum -y install nginx
        systemctl enable nginx.service
        systemctl start nginx.service
        vim /usr/share/nginx/html/index.html

systemctl start keepalived.service

        1.3   server2

BACKUP服务器的配置需要几处修改:

# yum -y install keepalived

# vi /etc/keepalived/keepalived.conf

        router_id 1  改为  router_id 2

        state MASTER改为  state BACKUP
        mcast_src_ip 192.168.26.144改为backup服务器实际的IP mcast_src_ip 192.168.26.169
        priority 100改为priority 99

# systemctl enable keepalived.service

安装Nginx：

        rpm -ivh http://nginx.org/packages/centos/7/noarch/RPMS/nginx-release-centos-7-0.el7.ngx.noarch.rpm
        yum -y install nginx
        systemctl enable nginx.service
        systemctl start nginx.service
        vim /usr/share/nginx/html/index.html

systemctl start keepalived.service

        1.4  客户机测试访问

访问VIP  http://192.168.26.190
拔掉master的网线。
访问VIP  http://192.168.26.190，观察网页已经切换

        1.5   关于keepalived对web状态未知的问题

对于上述实验;若maser服务器上的nginx服务出现了问题，此时用户再访问访问服务器是有问题的；

原因是keepalived监控的是接口IP状态。无法监控nginx服务状态

编辑监控脚本:

        --server1  添加Nginx监控脚本:

        # vim /etc/keepalived/ck_ng.sh

                #!/bin/bash
                #检查nginx进程是否存在
                counter=$(ps -C nginx --no-heading|wc -l)
                if [ "${counter}" = "0" ]; then
                #尝试启动一次nginx，停止5秒后再次检测
                    service nginx start
                    sleep 5
                    counter=$(ps -C nginx --no-heading|wc -l)
                    if [ "${counter}" = "0" ]; then
                #如果启动没成功，就杀掉keepalive触发主备切换
                        service keepalived stop
                    fi
                fi

        # chmod +x /etc/keepalived/ck_ng.sh

         --server2  添加Nginx监控脚本:

        # vim /etc/keepalived/ck_ng.sh

                #!/bin/bash
                #检查nginx进程是否存在
                counter=$(ps -C nginx --no-heading|wc -l)
                if [ "${counter}" = "0" ]; then
                #尝试启动一次nginx，停止5秒后再次检测
                    service nginx start
                    sleep 5
                    counter=$(ps -C nginx --no-heading|wc -l)
                    if [ "${counter}" = "0" ]; then
                #如果启动没成功，就杀掉keepalive触发主备切换
                        service keepalived stop
                    fi
                fi

        # chmod +x /etc/keepalived/ck_ng.sh

启动监控脚本    清除掉配置文件中的注释。    重启keepalived即可

2.案例2--keepalived  +  Lvs集群

        2.1  环境

keepalived+lvs集群

192.168.0.116    26.144  dr1    负载均衡器   master
192.168.0.117    26.169  dr2    负载均衡器
192.168.0.118    26.165    rs1     web1
192.168.0.119    26.166    rs2     web2

        2.2  在master上安装配置Keepalived:

# yum install keepalived  ipvsadm -y       ipvsadm安装并不启动

# vim /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {                        
    router_id Director1    #两边不一样。
    }
    
vrrp_instance VI_1 {                
    state MASTER                #另外一台机器是BACKUP    
    interface ens33                #心跳网卡    
    virtual_router_id 51            #虚拟路由编号，主备要一致
    priority 150                #优先级    
    advert_int 1                #检查间隔，单位秒    
    authentication {
        auth_type PASS
        auth_pass 1111
        }
    virtual_ipaddress {
        192.168.26.190/24       dev      ens33       #VIP和工作接口
        }
    }
    
virtual_server 192.168.26.190 80 {        #LVS 配置，VIP
    delay_loop 3                #服务论询的时间间隔，#每隔3秒检查一次real_server状态
    lb_algo rr                #LVS 调度算法
    lb_kind DR                 #LVS 集群模式
    protocol TCP
    real_server 192.168.26.165 80 {
        weight 1
        TCP_CHECK {
            connect_timeout 3       #健康检查方式,连接超时时间
            }
        }
    real_server 192.168.26.166 80 {
        weight 1
        TCP_CHECK {
            connect_timeout 3
            }
        }
}

        2.3   在backup上安装keepalived:

# yum install keepalived   ipvsadm -y      ipvsadm安装并不启动

拷贝master上的keepalived.conf到backup上：

        # scp  /etc/keepalived/keepalived.conf 192.168.26.169:/etc/keepalived/

拷贝后,修改配置文件：

        router_id Director2
        state    BACKUP
        priority 100

        2.4   master和backup上启动服务:

#systemctl enable keepalived
# systemctl start keepalived
#reboot

        2.5   web服务器配置

web1和web2同配置

安装web测试站点：

        yum install -y httpd && systemctl start httpd && systemctl enable httpd
        netstat -antp | grep httpd
        # elinks 127.0.0.1

        自定义web主页，以便观察负载均衡结果

配置虚拟地址：

        #cp /etc/sysconfig/network-scripts/{ifcfg-lo,ifcfg-lo:0}
        #vim /etc/sysconfig/network-scripts/ifcfg-lo:0
        DEVICE=lo:0
        IPADDR=192.168.0.20
        NETMASK=255.255.255.255
        ONBOOT=yes
        其他行注释掉

配置路由：

        #vim /etc/rc.local
        /sbin/route add -host 192.168.0.20  dev lo:0
        在两台机器（RS）上，添加一个路由：route add -host 192.168.0.20 dev lo:0    
        确保如果请求的目标IP是$VIP，那么让出去的数据包的源地址也显示为$VIP

配置ARP：

        # vim /etc/sysctl.conf

        net.ipv4.conf.all.arp_ignore = 1
        net.ipv4.conf.all.arp_announce = 2
        net.ipv4.conf.default.arp_ignore = 1
        net.ipv4.conf.default.arp_announce = 2
        net.ipv4.conf.lo.arp_ignore = 1
        net.ipv4.conf.lo.arp_announce = 2

        忽略arp请求  可以回复

        # reboot

        2.6   测试：

1）观察lvs路由条目                      master上 查询 # ipvsadm -L

2）观察vip地址在哪台机器上        master上 查询 # ip a

3）客户端浏览器访问vip

4）关闭master上的keepalived服务，再次访问vip
        master上 关闭 # systemctl stop keepalived.service 

5）关闭web1站点服务，再次访问VIP      

        web1 # systemctl stop httpd