Redis 安装主从配置及Sentinel配置自动Failover

xiu12

　　http://www.linuxidc.com/Linux/2013-11/92287p3.htm
　　1、安装redis
　　--------------------------------------------------------------------------------
　　首页地址：http://redis.io/
　　下载地址：http://download.redis.io/ 下载最新的源码包（当前最新稳定版2.6.16）
　　#wget http://download.redis.io/releases/redis-2.6.16.tar.gz
　　#tar xzf redis-2.6.16.tar.gz
　　#cd redis-2.6.16
　　#make
　　#make install
　　#cp redis.conf /etc/
　　--------------------------------------------------------------------------------
　　2、运行
　　--------------------------------------------------------------------------------
　　#redis-server /etc/redis.conf & （加载配置文件并后台运行）
　　#redis-cli （命令行工具）
　　--------------------------------------------------------------------------------
　　3、测试
　　--------------------------------------------------------------------------------
　　# redis-benchmark --help 获取帮助信息，其中包含压力测试命令样例
　　Examples:
　　Run the benchmark with the default configuration against 127.0.0.1:6379:
　　$ redis-benchmark
　　Fill 127.0.0.1:6379 with about 1 million keys only using the SET test:
　　$ redis-benchmark -t set -n 1000000 -r 100000000
　　[root@php1 ~]# redis-benchmark -t set -c 20 -n 1000000 -r 100000000
　　====== SET ======
　　1000000 requests completed in 8.92 seconds
　　20 parallel clients
　　3 bytes payload
　　keep alive: 1
　　100.00%  set a 123
　　OK
　　redis 127.0.0.1:6379> get a
　　"123"
　　#从服务器
　　[root@mysql1 ~]# redis-cli -h 127.0.0.1 -a redispass
　　redis 127.0.0.1:6379> get a
　　"123"
　　[root@mysql3 ~]# redis-cli -h 127.0.0.1 -a redispass
　　redis 127.0.0.1:6379> get a
　　"123"
　　redis 127.0.0.1:6379> set b 234
　　(error) READONLY You can't write against a read only slave. （开起了只读模式，所以从将不能写入数据，可以保证数据只从主服务器同步至从服务器）
　　#故障测试
　　#从服务器宕机
　　将自动从主服务器的主从关系中解除
　　#主服务器宕机，从服务器显示，并且不能自动切换
　　[15932] 03 Nov 09:46:25.465 * Connecting to MASTER...
　　[15932] 03 Nov 09:46:25.465 * MASTER  SLAVE sync started
　　[15932] 03 Nov 09:46:25.466 # Error condition on socket for SYNC: Connection refused
　　5、Redis sentinel配置
　　--------------------------------------------------------------------------------
　　主页地址：http://redis.io/topics/sentinel
　　sentinel主要功能就是为Redis M-S(master,slaves)集群提供了
　　1)master存活检测
　　2)集群中M-S服务监控
　　3) 自动故障转移,M-S角色转换等能力,
　　从一个方面说是提高了redis集群的可用性.
　　#添加并编辑配置文件/etc/sentinel.conf，新增内容如下，也可在安装文件中负责sentinel配置文件并作适当修改：
　　[root@php1 ~]# cat /etc/sentinel.conf
　　port 26379
　　sentinel monitor mymaster 172.17.16.7 6379 2
　　sentinel auth-pass mymaster redispass
　　sentinel down-after-milliseconds mymaster 5000
　　sentinel failover-timeout mymaster 60000
　　sentinel parallel-syncs mymaster 1
　　[root@mysql1 ~]# cat /etc/sentinel.conf
　　port 26479
　　sentinel monitor mymaster 172.17.16.7 6379 2
　　sentinel auth-pass mymaster redispass
　　sentinel down-after-milliseconds mymaster 30000
　　sentinel failover-timeout mymaster 900000
　　sentinel parallel-syncs mymaster 1
　　[root@mysql3 ~]# cat /etc/sentinel.conf
　　port 26579
　　sentinel monitor mymaster 172.17.16.7 6379 2
　　sentinel auth-pass mymaster redispass
　　sentinel down-after-milliseconds mymaster 30000
　　sentinel failover-timeout mymaster 900000
　　sentinel parallel-syncs mymaster 1
　　#在三个服务器中以sentinel模式启动redis-server
　　[root@php1 ~]# redis-server /etc/sentinel.conf --sentinel
　　[root@mysql1 ~]# redis-server /etc/sentinel.conf --sentinel
　　[root@mysql3 ~]# redis-server /etc/sentinel.conf --sentinel
　　#最后三台服务器sentinel输出如下：（注意每个机器输出对应的都是另外两台机器的IP）
　　#php1
　　[14047] 03 Nov 10:18:55.109 * +slave slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[14047] 03 Nov 10:18:55.109 * +slave slave 172.17.16.21:6379 172.17.16.21 6379 @ mymaster 172.17.16.7 6379
　　[14047] 03 Nov 10:19:42.272 * +sentinel sentinel 172.17.16.21:26479 172.17.16.21 26479 @ mymaster 172.17.16.7 6379
　　[14047] 03 Nov 10:20:46.338 * +sentinel sentinel 172.17.16.22:26579 172.17.16.22 26579 @ mymaster 172.17.16.7 6379
　　#mysql1
　　[3512] 03 Nov 10:23:36.473 * +slave slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:23:36.473 * +slave slave 172.17.16.21:6379 172.17.16.21 6379 @ mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:23:40.446 * +sentinel sentinel 172.17.16.7:26379 172.17.16.7 26379 @ mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:24:45.567 * +sentinel sentinel 172.17.16.22:26579 172.17.16.22 26579 @ mymaster 172.17.16.7 6379
　　#mysql3
　　[21428] 03 Nov 10:24:40.561 * +slave slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[21428] 03 Nov 10:24:40.561 * +slave slave 172.17.16.21:6379 172.17.16.21 6379 @ mymaster 172.17.16.7 6379
　　[21428] 03 Nov 10:24:41.367 * +sentinel sentinel 172.17.16.7:26379 172.17.16.7 26379 @ mymaster 172.17.16.7 6379
　　[21428] 03 Nov 10:24:42.280 * +sentinel sentinel 172.17.16.21:26479 172.17.16.21 26479 @ mymaster 172.17.16.7 6379
　　#关掉mysql3上的redis-server服务
　　#每个sentinel将输出如下信息，表示认为slave 22已经宕机，剔除
　　[3512] 03 Nov 10:30:08.359 # +sdown slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　#重新启动mysql3上的redis-server服务
　　#每个sentinel将输出如下信息，表示认为slave 22已经重新启动，新增到集群中
　　[3512] 03 Nov 10:31:28.901 * +reboot slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:31:29.115 # -sdown slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　#关掉主php1的redis-server服务
　　#将重新选举主服务器并重新配置连接到新的主服务器：
　　#php1上输出：
　　[14047] 03 Nov 10:29:49.888 # +sdown master mymaster 172.17.16.7 6379
　　[14047] 03 Nov 10:29:51.188 # +odown master mymaster 172.17.16.7 6379 #quorum 3/2
　　[14047] 03 Nov 10:29:59.189 # +failover-detected master mymaster 172.17.16.7 6379
　　[14047] 03 Nov 10:29:59.689 * +slave-reconf-inprog slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[14047] 03 Nov 10:30:00.789 * +slave-reconf-done slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[14047] 03 Nov 10:30:00.888 # +failover-end master mymaster 172.17.16.7 6379
　　[14047] 03 Nov 10:30:00.888 # +switch-master mymaster 172.17.16.7 6379 172.17.16.21 6379
　　[14047] 03 Nov 10:30:00.990 * +slave slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.21 6379
　　[14047] 03 Nov 10:30:01.058 * +sentinel sentinel 172.17.16.21:26479 172.17.16.21 26479 @ mymaster 172.17.16.21 6379
　　[14047] 03 Nov 10:30:05.956 * +sentinel sentinel 172.17.16.22:26579 172.17.16.22 26579 @ mymaster 172.17.16.21 6379
　　[14047] 03 Nov 10:30:30.988 # +sdown slave 172.17.16.7:6379 172.17.16.7 6379 @ mymaster 172.17.16.21 6379
　　#mysql1上输出：
　　[3512] 03 Nov 10:33:49.516 # +sdown master mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:33:49.716 # +odown master mymaster 172.17.16.7 6379 #quorum 3/2
　　[3512] 03 Nov 10:33:49.716 # +failover-triggered master mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:33:49.716 # +failover-state-wait-start master mymaster 172.17.16.7 6379 #starting in 8139 milliseconds
　　[3512] 03 Nov 10:33:57.893 # +failover-state-select-slave master mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:33:57.997 # +selected-slave slave 172.17.16.21:6379 172.17.16.21 6379 @ mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:33:57.997 * +failover-state-send-slaveof-noone slave 172.17.16.21:6379 172.17.16.21 6379 @ mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:33:58.101 * +failover-state-wait-promotion slave 172.17.16.21:6379 172.17.16.21 6379 @ mymaster 172.17.16.7 6379
　　[3496] 03 Nov 10:33:58.101 * MASTER MODE enabled (user request)
　　[3512] 03 Nov 10:33:58.741 # +promoted-slave slave 172.17.16.21:6379 172.17.16.21 6379 @ mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:33:58.741 # +failover-state-reconf-slaves master mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:33:58.849 * +slave-reconf-sent slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:33:58.957 * +slave-reconf-inprog slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[3496] 03 Nov 10:33:59.080 * Slave ask for synchronization
　　[3496] 03 Nov 10:33:59.080 * Starting BGSAVE for SYNC
　　[3496] 03 Nov 10:33:59.081 * Background saving started by pid 3515
　　[3515] 03 Nov 10:33:59.086 * DB saved on disk
　　[3515] 03 Nov 10:33:59.087 * RDB: 0 MB of memory used by copy-on-write
　　[3496] 03 Nov 10:33:59.188 * Background saving terminated with success
　　[3496] 03 Nov 10:33:59.189 * Synchronization with slave succeeded
　　[3512] 03 Nov 10:33:59.994 * +slave-reconf-done slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:34:00.093 # +failover-end master mymaster 172.17.16.7 6379
　　[3512] 03 Nov 10:34:00.093 # +switch-master mymaster 172.17.16.7 6379 172.17.16.21 6379
　　[3512] 03 Nov 10:34:00.194 * +slave slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.21 6379
　　[3512] 03 Nov 10:34:00.325 * +sentinel sentinel 172.17.16.7:26379 172.17.16.7 26379 @ mymaster 172.17.16.21 6379
　　[3512] 03 Nov 10:34:05.192 * +sentinel sentinel 172.17.16.22:26579 172.17.16.22 26579 @ mymaster 172.17.16.21 6379
　　[3512] 03 Nov 10:34:30.134 # +sdown slave 172.17.16.7:6379 172.17.16.7 6379 @ mymaster 172.17.16.21 6379
　　#mysql3上输出：
　　[21428] 03 Nov 10:33:49.343 # +sdown master mymaster 172.17.16.7 6379
　　[21428] 03 Nov 10:33:49.543 # +odown master mymaster 172.17.16.7 6379 #quorum 2/2
　　[21428] 03 Nov 10:33:58.896 # +failover-detected master mymaster 172.17.16.7 6379
　　[21428] 03 Nov 10:33:59.836 * +slave-reconf-inprog slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[21428] 03 Nov 10:33:59.836 * +slave-reconf-done slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.7 6379
　　[21428] 03 Nov 10:33:59.940 # +failover-end master mymaster 172.17.16.7 6379
　　[21428] 03 Nov 10:33:59.940 # +switch-master mymaster 172.17.16.7 6379 172.17.16.21 6379
　　[21428] 03 Nov 10:34:00.045 * +slave slave 172.17.16.22:6379 172.17.16.22 6379 @ mymaster 172.17.16.21 6379
　　[21428] 03 Nov 10:34:00.293 * +sentinel sentinel 172.17.16.21:26479 172.17.16.21 26479 @ mymaster 172.17.16.21 6379
　　[21428] 03 Nov 10:34:00.325 * +sentinel sentinel 172.17.16.7:26379 172.17.16.7 26379 @ mymaster 172.17.16.21 6379
　　[21428] 03 Nov 10:34:29.989 # +sdown slave 172.17.16.7:6379 172.17.16.7 6379 @ mymaster 172.17.16.21 6379
　　#查看主从关系：（mysql1为主，mysql3为从，如果php1重新启动，也将为从服务器加入到新的集群）
　　[root@mysql1 ~]# redis-cli -h 127.0.0.1 -p 6379 -a redispass info replication
　　# Replication
　　role:master
　　connected_slaves:1
　　slave0:172.17.16.22,6379,online
　　[root@mysql3 ~]# redis-cli -h 127.0.0.1 -p 6379 -a redispass info replication
　　# Replication
　　role:slave
　　master_host:172.17.16.21
　　master_port:6379
　　master_link_status:up
　　master_last_io_seconds_ago:1
　　master_sync_in_progress:0
　　slave_priority:100
　　slave_read_only:1
　　connected_slaves:0
　　#重新启动php1，检查mysql1的主从关系
　　[root@mysql1 ~]# redis-cli -h 127.0.0.1 -p 6379 -a redispass info replication
　　# Replication
　　role:master
　　connected_slaves:2
　　slave0:172.17.16.22,6379,online
　　slave1:172.17.16.7,6379,online
　　切换成功。
　　注意：
　　如果原来的 redis master停止后，并已经选取出新的master后，原来停止的master启动前需要指定
　　slaveof {目前master_ip}  6379