使用keepalived 实现redis主从高可用
keepalived官方有中文文档：LVS + Keepalived Chinese application doc - March 16, 2010。
keepalived 实现VRRP协议，从路由级别实现VIP切换，可以完全避免类似heartbeat 脑裂问题。可以很nice的实现主从、主备、互备方案，尤其是无状态业务，有状态业务就需要额外花些功夫了。
既然mysql 可以使用keepalived很好的做到主从切换，redis自然也就有样学样就成了。
redis主从实现完全没有mysql成熟，仅仅是可用而已，按照作者规划相应改进需要等3.0以后了。经过测试主从也不是那么完全不靠谱。主要问题在于同步连接断开之后需要重新全量同步，如果频繁进行会对主服务带来很大性能影响。 但现实中主从机器往往要求放在一个机柜同一台交换设备下，网络闪断情况极低；再者主从同步在同步数量量大情况下，需要将缓存区调得足够大，不然也容易造成连接断开。
实现切换逻辑如下：A B两台机器
1. A 、B 依次启动，A作为主、B为从
2.主A 挂掉，B接管业务，作为主
3.A 起来，作为从SLAVEOF B
4.B 挂掉，A 切回主
将一台全部作为主，即可实现主从，可做读写分离；也可以一台机器上多个实例一半主、一半从，实现互备份，两机同时接管业务，一台宕机后业务都集中在一台上。
在keepalived 有两个角色：Master(一个)、Backup（多个），如果设置一个为Master，但Master挂了后再起来，必然再次业务又一次切换，这对于有状态服务是不可接受的。解决方案就是两台机器都设置为Backup，而且优先级高的Backup设置为nopreemt 不抢占。
安装比较简单：
需要依赖包：openssl-devel（ubuntu 中为 libssl-dev),popt-devel （ubuntu中为libpopt-dev），redheat中都已经有了。
配置如下：
主：
! Configuration File for keepalived global_defs { router_id test } vrrp_script chk_redis { script "/etc/keepalived/scripts/redis_check.sh 127.0.0.1 7001" interval 2
timeout 2
fall 3 } vrrp_script chk_redis2 { script "/etc/keepalived/scripts/redis_check.sh 127.0.0.1 7002" interval 2
timeout 2
fall 3
} vrrp_instance redis { state BACKUP # 主也配置为SLAVE interface wlan0} virtual_router_id 51 priority 150 nopreempt # 不抢占，注意加上 advert_int 1 authentication { auth_type PASS auth_pass lulu } virtual_ipaddress { 192.168.1.144 } track_script { chk_redis } notify_master "/etc/keepalived/scripts/redis_master_slave.sh 127.0.0.1 192.168.1.126 7001" notify_backup "/etc/keepalived/scripts/redis_backup_slave.sh 127.0.0.1 192.168.1.126 7001" notify_fault /etc/keepalived/scripts/redis_fault.sh notify_stop /etc/keepalived/scripts/redis_stop.sh } vrrp_instance redis2 { state BACKUP interface wlan0 } virtual_router_id 50 priority 150 advert_int 1 nopreempt authentication { auth_type PASS auth_pass lulu } virtual_ipaddress { 192.168.1.145 } track_script { chk_redis2 } notify_master "/etc/keepalived/scripts/redis_master_slave.sh 127.0.0.1 192.168.1.126 7002" notify_backup "/etc/keepalived/scripts/redis_backup_slave.sh 127.0.0.1 192.168.1.126 7002" notify_fault /etc/keepalived/scripts/redis_fault.sh notify_stop /etc/keepalived/scripts/redis_stop.sh }
备：
! Configuration File for keepalived global_defs { router_id test } vrrp_script chk_redis { script "/etc/keepalived/scripts/redis_check.sh 127.0.0.1 7001" interval 2
timeout 2
fall 3 } vrrp_script chk_redis2 { script "/etc/keepalived/scripts/redis_check.sh 127.0.0.1 7002" interval 2
timeout 2
fall 3 } vrrp_instance redis { state BACKUP interface wlan0 } virtual_router_id 51 priority 100 advert_int 1 authentication { auth_type PASS auth_pass test } virtual_ipaddress { 192.168.1.144 } track_script { chk_redis } notify_master "/etc/keepalived/scripts/redis_master_slave.sh 127.0.0.1 192.168.1.126 7001" notify_backup "/etc/keepalived/scripts/redis_backup_slave.sh 127.0.0.1 192.168.1.126 7001" notify_fault /etc/keepalived/scripts/redis_fault.sh notify_stop /etc/keepalived/scripts/redis_stop.sh } vrrp_instance redis2 { state BACKUP interface wlan0 } virtual_router_id 50 advert_int 1 priority 100 authentication { auth_type PASS auth_pass test } virtual_ipaddress { 192.168.1.145 } track_script { chk_redis2 } notify_master "/etc/keepalived/scripts/redis_master_slave.sh 127.0.0.1 192.168.1.126 7002" notify_backup "/etc/keepalived/scripts/redis_backup_slave.sh 127.0.0.1 192.168.1.126 7002" notify_fault /etc/keepalived/scripts/redis_fault.sh notify_stop /etc/keepalived/scripts/redis_stop.sh }
其他脚本见附件：http://files.cnblogs.com/lulu/scripts.rar