Redis master-slave mechanism

One 、 Cluster structure

A single node Redis There is an upper limit to the concurrency of , We should further improve Redis Concurrency capability , You need to build a master-slave cluster , Read and write separation .

The master-slave structure is as follows ：

  There are three nodes , A master node , Two slave nodes . Start in the same virtual machine 3 individual redis example , Simulate master-slave cluster ：

 

Two 、 Steps to build

1、 Create directory ：

Create three folders , The names are 7001、7002、7003：

#  Get into /tmp Catalog 
cd /tmp
#  Create directory 
mkdir 7001 7002 7003

2、 To write redis.conf The configuration file :

take redis-6.2.6/redis.conf Copy the files into three directories , Persistence mode is the default RDB Pattern . And modify the configuration file in each folder , Change the ports to 7001、7002、7003, take rdb Change the file storage location to your own directory ：

#  Turn on RDB
# save ""
save 3600 1
save 300 100
save 60 10000

#  close AOF
appendonly no

sed -i -e 's/6379/7001/g' -e 's/dir .\//dir \/tmp\/7001\//g' 7001/redis.conf
sed -i -e 's/6379/7002/g' -e 's/dir .\//dir \/tmp\/7002\//g' 7002/redis.conf
sed -i -e 's/6379/7003/g' -e 's/dir .\//dir \/tmp\/7003\//g' 7003/redis.conf

3、 Modify the declaration in the configuration file IP：

The virtual machine itself has multiple IP, To avoid future chaos , We need to be in redis.conf Specify the binding of each instance in the file ip Information , The format is as follows ：

# redis Instance declaration  IP
replica-announce-ip 192.168.1.103

#  Do it one by one 
sed -i '1a replica-announce-ip 192.168.150.101' 7001/redis.conf
sed -i '1a replica-announce-ip 192.168.150.101' 7002/redis.conf
sed -i '1a replica-announce-ip 192.168.150.101' 7003/redis.conf

#  Or one click modification 
printf '%s\n' 7001 7002 7003 | xargs -I{} -t sed -i '1a replica-announce-ip 192.168.150.101' {}/redis.conf

3、 ... and 、 start-up

For the convenience of viewing the log , stay mobaxterm Open in 3 individual redis window , Respectively in tmp Start the command under the directory ：

#  The first 1 individual 
redis-server 7001/redis.conf
#  The first 2 individual 
redis-server 7002/redis.conf
#  The first 3 individual 
redis-server 7003/redis.conf

After starting ：

 

Four 、 Open the master-slave relationship

To configure the master-slave relationship, you can use replicaof perhaps slaveof（redis5.0 before ） command ：

# slaveof <masterip> <masterport>
slaveof 192.168.1.103 7001

temporary ： Use redis-cli The client connects to redis service , perform slaveof command （ Failure after restart ）;

permanent ： Add this command in redis.conf in .

In order to facilitate the demonstration, we use the temporary method .

adopt redis-cli Command connection 7002, Execute the following command ：

#  stay tmp Connection under directory  7002
redis-cli -p 7002
#  perform slaveof
slaveof 192.168.1.103 7001

adopt redis-cli Command connection 7003, Execute the following command ：

#  stay tmp Connection under directory  7003
redis-cli -p 7003
#  perform slaveof
slaveof 192.168.1.103 7001

Connect 7001, View the cluster status ：

#  Connect  7001
redis-cli -p 7001
#  Check the status 
info replication

result ：

 

5、 ... and 、 test

 

6、 ... and 、 principle

1、 Full amount of synchronization

master How to determine slave Is it the first time to synchronize data ？ Two very important concepts will be used here ：

Replication Id：

abbreviation replid, Is the tag of the dataset ,id Consistent means the same data set . every last master There's only one replid,slave Will inherit master Node replid;

offset：

Offset , As recorded in repl_baklog The data in increases gradually .slave When the synchronization is completed, the current synchronization will also be recorded offset. If slave Of offset Less than master Of offset, explain slave The data lag behind master, You need to update . therefore slave Do data synchronization , You have to go to master Declare your own replication id and offset,master To determine which data needs to be synchronized .

Full synchronous process ：

 

 2、 The incremental synchronization

among repl_baklog There is an upper limit on size , When full, the oldest data will be overwritten . If slave Disconnected for too long , The data that has not been backed up is overwritten , Cannot be based on log Do incremental synchronization , Only full synchronization can be performed again .

3、 Optimize

It can be optimized from the following aspects Redis Master-slave cluster ：

4、 summary