Redis序列之Sentinel

曾经终败给现在 2022-07-13 07:49 168阅读 0赞

前段时间一直在研究Redis的Sentinel集群，实验几次也没成功过，要么是Master有问题，要么是Mster宕机后，Slave没能自动切换成新的Master，今天突发灵感，实验一次成功，集群配置如上：  
Master：127.0.0.1:10001  
Slave：127.0.0.1:10002,127.0.0.1:10003  
Sentinel：127.0.0.1:26389,127.0.0.1:26489

首先新建10001，10002，10003文件夹，将redis.conf,redid-server分别Copy到文件夹中

longwentaodeMacBook-Pro:redis-cluster longwentao$ ls
    10001       10002       10003       sentinel
    longwentaodeMacBook-Pro:redis-cluster longwentao$ ls 10001
    redis-server    redis.conf

修改redis.conf文件中的端口号和文件夹保持一致

port 10001
    port 10002
    port 10003

修改10002,10003文件下的redis.conf文件，指定master为127.0.0.1:10001，其他使用默认配置

# slaveof <masterip> <masterport>
    slaveof 127.0.0.1 10001

再创建sentinel文件夹，将redis-sentinel,sentinel.conf文件Copy到文件夹中，将sentinel.conf文件重命名为sentinel\_26379.conf，修改端口号及需要监控的master

port 26379
    sentinel monitor mymaster 127.0.0.1 10001 1

1表示至少一个sentinel同意master宕机后，切换新的master

将sentinel\_26379.conf Copy一份成sentinel\_26479.conf，并修改端口号为26479，其他不变，最后sentinel文件夹中的内容如下

longwentaodeMacBook-Pro:sentinel longwentao$ ls
    redis-sentinel      sentinel_26379.conf sentinel_26479.conf

一个conf文件代表一个sentinel

到这里整个集群配置就完成了，接下来先启动master和sentinel

longwentaodeMacBook-Pro:10001 longwentao$ ./redis-server redis.conf 
    longwentaodeMacBook-Pro:sentinel longwentao$ ./redis-sentinel sentinel_26379.conf 
    longwentaodeMacBook-Pro:sentinel longwentao$ ./redis-sentinel sentinel_26479.conf

再启动10002和10003

longwentaodeMacBook-Pro:10002 longwentao$ ./redis-server redis.conf 
    longwentaodeMacBook-Pro:10003 longwentao$ ./redis-server redis.conf

可以看到，sentinel中已经监控到增加了两个slave

588:X 31 Dec 12:54:52.469 # Sentinel ID is 75e84135f9dddd13910b3e835f6836b0d861dbdb
    588:X 31 Dec 12:54:52.469 # +monitor master mymaster 127.0.0.1 10001 quorum 1
    588:X 31 Dec 12:54:52.470 * +slave slave 127.0.0.1:10002 127.0.0.1 10002 @ mymaster 127.0.0.1 10001 588:X 31 Dec 12:54:52.470 * +slave slave 127.0.0.1:10003 127.0.0.1 10003 @ mymaster 127.0.0.1 10001

在10001中查看群集信息，10001为Master，有两个slave10002，10003

longwentaodeMacBook-Pro:src longwentao$ ./redis-cli -p 10001
    127.0.0.1:10001> info replication
    # Replication
    role:master
    connected_slaves:2
    slave0:ip=127.0.0.1,port=10002,state=online,offset=417020,lag=1
    slave1:ip=127.0.0.1,port=10003,state=online,offset=417020,lag=1
    master_repl_offset:417020
    repl_backlog_active:1
    repl_backlog_size:1048576
    repl_backlog_first_byte_offset:2
    repl_backlog_histlen:417019

在10002中查看集群信息

longwentaodeMacBook-Pro:src longwentao$ ./redis-cli -p 10002
    127.0.0.1:10002> info replication
    # Replication
    role:slave
    master_host:127.0.0.1
    master_port:10001
    master_link_status:up
    master_last_io_seconds_ago:2
    master_sync_in_progress:0
    slave_repl_offset:424608
    slave_priority:100
    slave_read_only:1
    connected_slaves:0
    master_repl_offset:0
    repl_backlog_active:0
    repl_backlog_size:1048576
    repl_backlog_first_byte_offset:0
    repl_backlog_histlen:0
    127.0.0.1:10002>

宕机验证，将master停掉,sentinel中显示10003自动变成了master

588:X 31 Dec 13:57:38.194 # +switch-master mymaster 127.0.0.1 10001 127.0.0.1 10003
    588:X 31 Dec 13:57:38.195 * +slave slave 127.0.0.1:10002 127.0.0.1 10002 @ mymaster 127.0.0.1 10003 588:X 31 Dec 13:57:38.195 * +slave slave 127.0.0.1:10001 127.0.0.1 10001 @ mymaster 127.0.0.1 10003 588:X 31 Dec 13:58:08.268 # +sdown slave 127.0.0.1:10001 127.0.0.1 10001 @ mymaster 127.0.0.1 10003

再查看集群信息，10003为master,有一个10002的slave节点

longwentaodeMacBook-Pro:src longwentao$ ./redis-cli -p 10003
    127.0.0.1:10003> info replication
    # Replication
    role:master
    connected_slaves:1
    slave0:ip=127.0.0.1,port=10002,state=online,offset=28740,lag=1
    master_repl_offset:28874
    repl_backlog_active:1
    repl_backlog_size:1048576
    repl_backlog_first_byte_offset:2
    repl_backlog_histlen:28873

再将10001启动,sentinel显示增加了一个10001的slave，master为10003

588:X 31 Dec 14:03:23.277 # -sdown slave 127.0.0.1:10001 127.0.0.1 10001 @ mymaster 127.0.0.1 10003 588:X 31 Dec 14:03:33.256 * +convert-to-slave slave 127.0.0.1:10001 127.0.0.1 10001 @ mymaster 127.0.0.1 10003

查看集群信息，master还是10003，10001已经变成了slave

127.0.0.1:10003> info replication
    # Replication
    role:master
    connected_slaves:2
    slave0:ip=127.0.0.1,port=10002,state=online,offset=60039,lag=1
    slave1:ip=127.0.0.1,port=10001,state=online,offset=60039,lag=1
    master_repl_offset:60039
    repl_backlog_active:1
    repl_backlog_size:1048576
    repl_backlog_first_byte_offset:2
    repl_backlog_histlen:60038
    127.0.0.1:10003>