Keepalived详解（三）
Keepalived基础功能应用实例：1.Keepalived基础HA功能演示：
在默认情况下，Keepalived可以实现对系统死机、网络异常及Keepalived本身进行监控，也就是说当系统出现死机、网络出现故障或Keepalived进程异常时，Keepalived会进行主备节点的切换。但这些还是不够的，因为集群中运行的服务也随时可能出现问题，因此，还需要对集群中运行服务的状态进行监控，当服务出现问题时也进行主备切换。Keepalived作为一个优秀的高可用集群软件，也考虑到了这一点，它提供了一个vrrp_script模块专门用来对集群中服务资源进行监控。
1. 配置Keepalived：
操作系统：CentOS release 6.7 版本：keepalived-1.2.12
主机名主机IP地址集群角色集群服务虚拟IP地址
keepalived-master10.0.0.35 MASTER(主节点) HTTPD 10.0.0.40
keepalived-backup10.0.0.36 BACKUP(备用节点) HTTPD
keepalived-master节点的keepalived.conf文件的内容：
! Configuration File for keepalived
global_defs {
notification_email {
*********@qq.com
}
notification_email_from *********@163.com
smtp_server smtp.163.com
smtp_connect_timeout 30
router_id LVS_DEVEL
}
vrrp_script check_httpd {
script "killall -0 httpd"
interval 2
}
vrrp_instance HA_1 {
state MASTER
interface eth0
virtual_router_id 80
priority 100
advert_int 2
authentication {
auth_type PASS
auth_pass qwaszx
}
notify_master "/etc/keepalived/master.sh"
notify_backup "/etc/keepalived/backup.sh"
notify_fault "/etc/keepalived/fault.sh"
track_script {
check_httpd
}
virtual_ipaddress {
10.0.0.40/24 dev eth0
}
}
master.sh文件内容为：
#!/bin/bash
LOGFILE=/var/log/keepalived-mysql-state.log
echo "" >> $LOGFILE
date >> $LOGFILE
backup.sh文件内容为：
#!/bin/bash
LOGFILE=/var/log/keepalived-mysql-state.log
echo "" >> $LOGFILE
date >> $LOGFILE
fault.sh文件内容为：
#!/bin/bash
LOGFILE=/var/log/keepalived-mysql-state.log
echo "" >> $LOGFILE
date >> $LOGFILE
keepalived-backup节点上的keepalived.conf配置文件内容与keepalived-master节点上的基本相同，需要修改的地方有两个：
□ 将“state MASTER” 更改为 “state BACKUP”。
■ 将priority 100 更改为一个较小的值，这里改为“priority 80”
2.Keepalived启动过程分析：
keepalived-master节点启动keepalived服务，操作如下：
# /etc/init.d/httpd start
# /etc/init.d/keepalived start
keepalived-backup节点启动keepalived服务，操作如下：
# /etc/init.d/httpd start
# /etc/init.d/keepalived start
3.keepalived的故障切换过程分析：
在keepalived-master节点出现故障后，备用节点keepalived-backup立刻检测到，此时备用机变成MASTER角色，并且接管了keepalived-master主机的虚拟IP资源，最后将虚拟IP绑定在eth0设备上。
4.故障恢复切换分析：
keepalived-master节点通过vrrp_script模块检测到httpd服务已经恢复正常，然后自动切换到MASTER状态，同时也夺回了集群资源，将虚拟IP地址再次绑定在eth0设备上。
纵观Keepalived的整个运行过程和切换过程，看似合理，事实上并非如此：在一个高负载、高并发、追求稳定的业务系统中，执行一次主、备切换对业务系统影响很大，因此，不到万不得已，尽量不要进行主、备角色的切换，也就是说，在主节点发生故障后，必须切换到备用节点，而在主节点故障恢复后，不希望再次切回主节点，直到备用节点发生故障时才进行切换，这就是前面介绍的不抢占功能，可以通过Keepalived的nopreempt选项来实现。
配置：
master:
! Configuration File for keepalived
global_defs {
notification_email {
**********@qq.com
}
notification_email_from *******@163.com
smtp_server smtp.163.com
smtp_connect_timeout 30
router_id LVS_DEVEL
}
vrrp_script check_httpd {
script "killall -0 httpd"
interval 2
}
vrrp_instance HA_1 {
state BACKUP
interface eth0
virtual_router_id 80
priority 100
nopreempt
advert_int 2
authentication {
auth_type PASS
auth_pass qwaszx
}
notify_master "/etc/keepalived/master.sh"
notify_backup "/etc/keepalived/backup.sh"
notify_fault "/etc/keepalived/fault.sh"
track_script {
check_httpd
}
virtual_ipaddress {
10.0.0.40/24 dev eth0
}
}
backup:
! Configuration File for keepalived
global_defs {
notification_email {
********@qq.com
}
notification_email_from *******@163.com
smtp_server smtp.163.com
smtp_connect_timeout 30
router_id LVS_DEVEL
}
vrrp_script check_httpd {
script "killall -0 httpd"
interval 2
}
vrrp_instance HA_1 {
state BACKUP
interface eth0
virtual_router_id 80
priority 80
#nopreempt
advert_int 2
authentication {
auth_type PASS
auth_pass qwaszx
}
notify_master "/etc/keepalived/master.sh"
notify_backup "/etc/keepalived/backup.sh"
notify_fault "/etc/keepalived/fault.sh"
track_script {
check_httpd
}
virtual_ipaddress {
10.0.0.40/24 dev eth0
}
}
页:
[1]