Keepalived 笔记

两个 haproxy 不抢占

• 环境
  • haproxy 服务器
    • haproxy101: 10.1.1.101
    • haproxy102: 10.1.1.102
  • 虚拟地址
    • ip: 10.1.1.100
• 在全部 haproxy 服务器上安装 keepalived

  yum install keepalived
  

• 在全部 haproxy 服务器上配置 haproxy 和 keepalived 自启动

  systemctl enable haproxy
  systemctl enable keepalived
  

• MASTER/BACKUP 完整配置

  global_defs {
      router_id haproxy101 #BACKUP 这里是 haproxy102
      script_user root
      enable_script_security
  }
  vrrp_script chk_haproxy {
      script "/usr/bin/systemctl status haproxy"
      interval 2
      weight 0
      fall 2
      rise 2
  }
  vrrp_instance VI_1 {
      state BACKUP #MASTER 和 BACKUP 这里都是 BACKUP
      virtual_router_id 1
      priority 150 #BACKUP 这里是 100
      advert_int 2
      nopreempt #BACKUP 优先级低，需注释此行
      interface eth0
      track_script {
          chk_haproxy
      }
      authentication {
          auth_type PASS
          auth_pass 1011100
      }
      virtual_ipaddress {
          10.1.1.100/24 dev eth0
      }
  }
  

两个 LVS-DR 调度器不抢占均衡后端 MySQL 和 Ceph 负载

• 环境
  • keepalived 服务器
    • ka101: 10.1.1.101
    • ka102: 10.1.1.102
  • 虚拟地址
    • ip: 10.1.1.100
  • mysqld 服务器
    • mysql103: 10.1.1.103
    • mysql104: 10.1.1.104
    • mysql105: 10.1.1.105
  • ceph-radosgw 服务器
    • ceph106: 10.1.1.106
    • ceph107: 10.1.1.107
    • ceph108: 10.1.1.108
• 在全部 keepalived 服务器上安装 keepalived

  yum install keepalived
  

• 在全部 keepalived 服务器上配置 keepalived 自启动

  systemctl enable keepalived
  

• MASTER/BACKUP 完整配置
  • 全局和实例配置

    global_defs {
        router_id keepalive101 #BACKUP 配置 keepalive102
    }
    vrrp_instance V1_1 {
        state BACKUP #BACKUP 也配置 BACKUP
        interface eth0
        virtual_router_id 1
        priority 150 #BACKUP 配置 100
        advert_int 1
        nopreempt #BACKUP 优先级低，需注释此行
        authentication {
            auth_type PASS
            auth_pass 1011100
        }
        virtual_ipaddress {
            10.1.1.100/24 dev eth0
        }
    }
    

  • LVS 均衡 mysql galera cluser 负载

    virtual_server 10.1.1.100 3306 {
        delay_loop 16
        lb_algo sh
        lb_kind DR
        protocol TCP
        real_server 10.1.1.103 3306 {
            weight 1
            TCP_CHECK {
                connect_timeout 4
                delay_before_retry 2
                connect_port 3306
            }
        }
        real_server 10.1.1.104 3306 {
            weight 1
            TCP_CHECK {
                connect_timeout 4
                delay_before_retry 2
                connect_port 3306
            }
        }
        real_server 10.1.1.105 3306 {
            weight 1
            TCP_CHECK {
                connect_timeout 4
                delay_before_retry 2
                connect_port 3306
            }
        }
    }
    

  • LVS 均衡 ceph radosgw 负载

    virtual_server 10.1.1.100 7480 {
        delay_loop 16
        lb_algo sh
        lb_kind DR
        protocol TCP
        real_server 10.1.1.106 7480 {
            weight 1
            HTTP_GET {
                url {
                    path /
                    status_code 200
                }
                connect_timeout 8
                nb_get_retry 2
                delay_before_retry 2
                connect_port 7480
            }
        }
        real_server 10.1.1.107 7480 {
            weight 1
            HTTP_GET {
                url {
                    path /
                    status_code 200
                }
                connect_timeout 8
                nb_get_retry 2
                delay_before_retry 2
                connect_port 7480
            }
        }
        real_server 10.1.1.108 7480 {
            weight 1
            HTTP_GET {
                url {
                    path /
                    status_code 200
                }
                connect_timeout 8
                nb_get_retry 2
                delay_before_retry 2
                connect_port 7480
            }
        }
    }
    

• 在全部 mysql 和 ceph-radosgw 服务器上配置虚拟 ip

  echo "1" >/proc/sys/net/ipv4/conf/lo/arp_ignore
  echo "2" >/proc/sys/net/ipv4/conf/lo/arp_announce
  echo "1" >/proc/sys/net/ipv4/conf/all/arp_ignore
  echo "2" >/proc/sys/net/ipv4/conf/all/arp_announce
  ip addr add 10.1.1.100/32 brd 10.1.1.100 dev lo
  ip route add 10.1.1.100 dev lo
  

不抢占自动切换两台 Redis 主从状态

• 环境
  • Redis 服务器
    • redis101: 10.1.1.101
    • redis102: 10.1.1.102
  • 虚拟地址
    • ip: 10.1.1.100
• 在全部 redis 服务器上安装 keepalived

  yum install keepalived
  

• 在全部 redis 服务器上配置 redis 和 keepalived 自启动

  systemctl enable redis
  systemctl enable keepalived
  

• MASTER/BACKUP 完整配置

  global_defs {
      router_id redis101 #BACKUP 这里是 redis102
      script_user root
      enable_script_security
  }
  vrrp_script chk_redis {
      script "/usr/bin/systemctl status redis"
      interval 2
      weight 0
      fall 2
      rise 2
  }
  vrrp_instance VI_1 {
      state BACKUP #MASTER 和 BACKUP 这里都是 BACKUP
      virtual_router_id 51
      priority 150 #BACKUP 这里是 100
      advert_int 1
      nopreempt #BACKUP 优先级低，需注释此行
      interface eth0
      notify_master /etc/keepalived/scripts/master.sh
      notify_backup /etc/keepalived/scripts/backup.sh
      track_script {
          chk_redis
      }
      authentication {
          auth_type PASS
          auth_pass 123456
      }
      virtual_ipaddress {
          10.1.1.100/24 dev eth0
      }
  }
  

• 在全部 redis 服务器上创建 /etc/keepalived/scripts/master.sh 脚本，内容如下

  #!/bin/bash
  #
  /usr/bin/sed -i '/^slaveof/d' /etc/redis.conf
  /usr/bin/systemctl restart redis
  

• 在 redis101 上创建 /etc/keepalived/scripts/backup.sh 脚本，内容如下

  #!/bin/bash
  #
  /usr/bin/sed -i '/^slaveof/d' /etc/redis.conf
  echo 'slaveof 10.1.1.102 6379' >> /etc/redis.conf
  /usr/bin/systemctl restart redis
  

• 在 redis102 上创建 /etc/keepalived/scripts/backup.sh 脚本，内容如下

  #!/bin/bash
  #
  /usr/bin/sed -i '/^slaveof/d' /etc/redis.conf
  echo 'slaveof 10.1.1.101 6379' >> /etc/redis.conf
  /usr/bin/systemctl restart redis
  

• 在全部 redis 服务器上赋予脚本可执行权限

  chmod 0755 /etc/keepalived/scripts/*.sh
  

• 在全部 redis 服务器上启动 redis 和 keepalived 服务

  systemctl start redis
  systemctl start keepalived