当前位置：网站首页>Redis sentinel cluster main database failure data recovery ideas ＃ yyds dry goods inventory ＃

Redis sentinel cluster main database failure data recovery ideas ＃ yyds dry goods inventory ＃

2022-06-28 09:41:00 【Jiangxiaolong's technology blog】

Redis Sentinel cluster main database fault data recovery

When the main database is repaired and re launched, first know who is the current main database through the sentry , Then you will go to the main database to synchronize data , And it will automatically modify the configuration file , After data synchronization , If you want to restore the primary database to become the primary database again, you need to increase the weight of the primary database , And then re elect , At this time, the original master database can become a new master database , After adjustment, adjust the weight value of the main library to the default value

Realize the idea ：

1. Restore the failed primary library

2. View the current master-slave status , Verify that the main library is down , Whether it is synchronized with the data generated from the library

3. Adjust the weight value

4. Re election , Change the original master database into a new master database

5. After the restored primary database becomes the new primary database again , To change all the adjusted weight values to the default values

The premise that the main library can rejoin the sentinel cluster ： One of the remaining two nodes must be master, And the two node configuration files have specified new nodes master Address

1. Recover the damaged main library

1. Restore main library 
[[email protected] ~]# redis-server /data/redis_cluster/redis_6379/conf/redis_6379.conf 
[[email protected] ~]# 
[[email protected] ~]# redis-sentinel /data/redis_cluster/redis_26379/conf/redis_26379.conf

2. View the log output of the other two nodes , Any node will output , Express redis-1 Has joined the cluster 
tail -f /data/redis_cluster/redis_26379/logs/redis_26379.log 
78223:X 30 Jan 12:05:09.073 # -sdown sentinel ac621a57296db0cead07751a4f0a19c570daa7f9 192.168.81.210 26379 @ mymaster 192.168.81.220 6379,

     
      1.
      2.
      3.
      4.
      5.
      6.
      7.
      8.

Redis The idea of data recovery from the main database of sentry cluster #yyds Dry inventory #_ priority

2. View the recovered Primary Library redis-1 The configuration file

[[email protected] ~]# cat /data/redis_cluster/redis_26379/conf/redis_26379.conf
bind 192.168.81.210
port 26379
daemonize yes
logfile "/data/redis_cluster/redis_26379/logs/redis_26379.log"
dir "/data/redis_cluster/redis_26379/data"
sentinel myid ac621a57296db0cead07751a4f0a19c570daa7f9
sentinel monitor mymaster 192.168.81.220 6379 2

     
      1.
      2.
      3.
      4.
      5.
      6.
      7.
      8.

You can see that it has been automatically modified to the address of the current library

Redis The idea of data recovery from the main database of sentry cluster #yyds Dry inventory #_ priority _02

3. View the recovered Primary Library redis-1 The master-slave relationship of

1. The current master library has been synchronized redis-2
[[email protected] ~]# redis-cli 
127.0.0.1:6379> CONFIG GET slaveof
1) "slaveof"
2) "192.168.81.220 6379"

2. You can already see the main library downtime phase , The latest data generated from the library to the main library 
127.0.0.1:6379> get gzzy_test
"guzhangzhuanyi"

     
      1.
      2.
      3.
      4.
      5.
      6.
      7.
      8.
      9.

4. Configure the weight value of the recovered primary library , Make it re elected as the main library

The sentinel election is first of all

1. Check who has higher weight priority is selected as the main library

2. Weight and priority are consistent , On the comparison id,id Big election

1. View the weight values of the other two nodes 
[[email protected] ~]# redis-cli -h 192.168.81.220 -p 6379 config get slave-priority 
1) "slave-priority"
2) "100"
 
[[email protected] ~]# redis-cli -h 192.168.81.230 -p 6379 config get slave-priority 
1) "slave-priority"
2) "100"

2. Change the weight value of the other two nodes to 0
[[email protected] ~]# redis-cli -h 192.168.81.220 -p 6379 config set slave-priority 0
OK
[[email protected] ~]# redis-cli -h 192.168.81.230 -p 6379 config set slave-priority 0
OK

3. Set the permission priority of the recovered primary database to be higher than the other two nodes 
[[email protected] ~]# redis-cli -h 192.168.81.210 -p 6379 config set slave-priority 150
OK
[[email protected] ~]# redis-cli -h 192.168.81.210 -p 6379 config get slave-priority 
1) "slave-priority"
2) "150"


4. Re election 
[[email protected] ~]# redis-cli -h 192.168.81.210 -p 26379 sentinel failover mymaster


5. View other nodes sentinel Output log 
[[email protected] ~]# tail -f /data/redis_cluster/redis_26379/logs/redis_26379.log
78223:X 30 Jan 12:32:27.591 * +convert-to-slave slave 192.168.81.220:6379 192.168.81.220 6379 @ mymaster 192.168.81.210 6379


     
      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.

According to the output of the log , It can be clearly seen that the adjustment redis-1 The weight priority of is 150, Higher than the other two nodes , therefore redis-1 It becomes the main library

Redis The idea of data recovery from the main database of sentry cluster #yyds Dry inventory #_ priority _03

View the master-slave replication relationship of the node

There is no synchronized Library in the main library , The other two nodes are synchronized redis-1 Main library

[[email protected] ~]# redis-cli -h 192.168.81.210 -p 6379 config get slaveof
1) "slaveof"
2) ""

[[email protected] ~]# redis-cli -h 192.168.81.220 -p 6379 config get slaveof
1) "slaveof"
2) "192.168.81.210 6379"

[[email protected] ~]# redis-cli -h 192.168.81.230 -p 6379 config get slaveof
1) "slaveof"
2) "192.168.81.210 6379"


     
      1.
      2.
      3.
      4.
      5.
      6.
      7.
      8.
      9.
      10.
      11.
      12.

5. Adjust the weight value to the default value

Adjust the weight value to the default value , As a condition for judgment in the next election

[[email protected] ~]# redis-cli -h 192.168.81.210 -p 6379 config set slave-priority 100
OK
[[email protected] ~]# redis-cli -h 192.168.81.220 -p 6379 config set slave-priority 100
OK
[[email protected] ~]# redis-cli -h 192.168.81.230 -p 6379 config set slave-priority 100
OK

     
      1.
      2.
      3.
      4.
      5.
      6.

原网站

版权声明
本文为[Jiangxiaolong's technology blog]所创，转载请带上原文链接，感谢
https://yzsam.com/2022/02/202202161346099093.html