Causes the exception condition:
1, three Redis node data cannot be persisted by an RDB
2. Redis data can only be read and cannot be written (caused by problem 1), resulting in data cannot be added and updated directly
Current temporary processing methods:
1, config set stop-writes-on-bgsave-error no first let data can write Redis, do not affect the data read and write operations on the line
2. Adjustment vm.overcommit_memory = 2 , This configuration is currently ineffective because the Redis master process memory usage is already high
0 Direct and Idle physical memory contrast, enough to release
1 Direct Release
2 physical Memory 50% (can be adjusted ) plus swap, if enough to release, abnormal will not turn off the process
3, enter three nodes, respectively, delete the relatively small amount of access, large capacity of data, control the amount of Redis master process memory
At present the idea, still need to do the thing,
1, continue to study Redis persistence , think there is no better alternative
2, vm.overcommit_memory and other core parameters of the actual principle
Redis RDB Persistence exception