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
1

yum install keepalived
在全部 haproxy 服务器上配置 haproxy 和 keepalived 自启动
1 2

systemctl enable haproxy systemctl enable keepalived

MASTER/BACKUP 完整配置

 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


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
1

yum install keepalived
在全部 keepalived 服务器上配置 keepalived 自启动
1

systemctl enable keepalived

MASTER/BACKUP 完整配置

全局和实例配置

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18


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

 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


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

 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
45


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

1
2
3
4
5
6


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
1

yum install keepalived
在全部 redis 服务器上配置 redis 和 keepalived 自启动
1 2

systemctl enable redis systemctl enable keepalived

MASTER/BACKUP 完整配置

 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


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

1
2
3
4


#!/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

1
2
3
4
5


#!/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

1
2
3
4
5


#!/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
1

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

修改 redis 配置

1
2
3
4


requirepass redis_password
maxclients 1000
maxmemory 4294967296
maxmemory-policy volatile-lru

在全部 redis 服务器上启动 redis 和 keepalived 服务
1 2

systemctl start redis systemctl start keepalived

文章目录

两个 haproxy 不抢占

两个 LVS-DR 调度器不抢占均衡后端 MySQL 和 Ceph 负载

不抢占自动切换两台 Redis 主从状态