Redis master-slave structure and application scenarios

summary

Generally speaking, the master-slave structure is used to achieve high availability , That is, when the primary server goes down , Switch to the slave server immediately . The master-slave structure has many advantages , such as : High availability , Read write pressure separation ; It also brings problems , The master-slave data may be inconsistent , Cause dirty data to be read .redis Since the master-slave structure is supported , Naturally, these problems also exist , We can see redis How to solve these problems .

principle

explain : Normal master-slave synchronization structure , The slave node simply executes replicaof Command is enough

explain : One of them salve Dropped the line or hung up ( Generally speaking , The master and slave are in a network segment , It is unlikely to be disconnected , There is a high probability of downtime ), This
When master The newly accepted log cannot be sent to this slave, Lead to master-slave synchronization delay ,master Will temporarily put the newly accepted command into
repl_backlog_buffer( Ring buffer ),master_repl_offset As the number of requests increases
Long ,slave_repl_offset The record is slave Synchronized location , wait for salve Network recovery , Continue to synchronize the remaining logs .

explain : The master-slave structure is generally designed to prevent master Downtime , Then it involves choosing a new master, Resynchronization issues .redis It is the use of sentinels to
New election master, The election process is probably to eliminate the nodes with poor network first , Then, according to the priority of the node configuration , Then judge the synchronization progress , The last is
Find the first to go online, that is ID The smallest . After the election ,