Introduction to redis high availability

High availability base - Master slave copy

Redis The replication function of is to support data synchronization between multiple databases ,
The main database can be read and write , When the master database data changes, it will be automatically synchronized to the slave database ,
The slave database is generally read-only , It will receive the data synchronized from the note database .
A master database can have more than one slave database , A slave database can only have one master database

By default , Each server is the master node , Can be set by configuration Redis Nodes are called slave databases

The principle of replication

1、 When starting a slave node , It will send one psync To the master node
2、 If the slave node is connected to the master node for the first time , Will trigger a full copy , At this point, the master node will start a background process , Start generating a RDB Snapshot file
3、 At the same time, new commands received from the client will be written into the cache memory ,RDB After file generation , The master node will RDB The file is sent to the slave node , The slave node will first RDB File write to local disk , Then it is loaded into memory from the local disk
4、 Then the master node will send the cache write command in memory to the slave node , Synchronize this data from the node
5、 If a network failure occurs before the slave node and the master node , Connection is broken , Will automatically reconnect , After connection, the master node will synchronize some missing data to the slave node

Master slave replication configuration

By default , Each node is the master node , Just configure the slave node
You can copy Redis.conf The configuration file , Modify the main information

Port number
pid name
log File name
dump.rdb name

You can modify the configuration , The configuration is permanently valid , It can also be processed in the form of commands

command ：info replication: View current library information

command ：slaveof ip port : Execute the name... On the slave server , Given the port and address of the primary server IP
command ：Slave no one : You can restore from a node to a host

Start the three Redis service , Given port 6380、6381、6382
Given 6380 Master node
1、 establish Redis example

nohup redis-server --port 6380 >> /usr/local/redis/data/log/6380.log 2>&1 &
nohup redis-server --port 6381 >> /usr/local/redis/data/log/6381.log 2>&1 &
nohup redis-server --port 6382 >> /usr/local/redis/data/log/6382.log 2>&1 &


[root@redis ~]# ps -ef |grep redis
root      16421  16314  0 03:01 pts/1    00:00:00 redis-server *:6380
root      16427  16314  0 03:01 pts/1    00:00:00 redis-server *:6381
root      16431  16314  0 03:01 pts/1    00:00:00 redis-server *:6382

2、 Connect to the database and set up master-slave replication

Gong-Dezhe-MBP:log gongdezhe$ redis-cli -p 6380
127.0.0.1:6380> 

Gong-Dezhe-MBP:~ gongdezhe$ redis-cli -p 6381
127.0.0.1:6381> slaveof 127.0.0.1 6380
OK

Gong-Dezhe-MBP:~ gongdezhe$ redis-cli -p 6382
127.0.0.1:6382> slaveof 127.0.0.1 6380
OK

After the master-slave configuration is complete , Master node information

From node information ：

Data synchronization test ：

High availability solution - Sentinel mode sentinel

The built master-slave replication scheme can achieve the purpose of data synchronization , But when the primary server goes down , You need to manually switch one from the server to the master server , This process requires manual intervention , Switching at the same time will lead to Redis The write function for is not available .
Need a way to automatically complete master Fault detection and be able to put a Slave Switch to a master, Sentinels are needed at this time sentinel Pattern , Sentinel mode can automatically switch between master and slave nodes .

Introduction to sentinel mode

When we set up redis A problem will be found after the master-slave replication scheme , That is, when the primary server goes down , You need to manually switch one from the server to the primary server , This requires human intervention , laborious , At the same time, it will also cause redis Server write not available . So we need a way to accomplish Master After a fault, it can automatically send a Slave Switch to a Master, This is the time sentinel Sentinel mode . The sentinel mechanism can automatically switch between master and slave nodes .( Automatic election of the boss )

Introduction to sentinel mode

sentinel It is an official high availability solution , The principle is that the sentry sends orders , wait for Redis Server response , So as to monitor the operation of multiple Redis example . meanwhile sentinel It's a distributed system , You can run multiple Sentinel process , It can be done sentinel High availability .

sentinel working process ：

● By sending to the master and slave servers ping command , Return the server to running state .
● When the Sentinels detect master Downtime , Will automatically send a slave Switch to master, Then the other slave servers are informed through publish subscribe mode , Modify the configuration file , Let them switch hosts .

About sentinel Three scheduled tasks ：

● Every time 1 Every second sentinel For others sentinel and redis Node execution ping operation , The heartbeat detection .
● Every time 10 Every second sentinel Would be right master and slave perform info command , The purpose is to find slave node , Determine the master-slave relationship .
● Every time 2 Every second sentinel adopt master Node channel Exchange information （pub/sub）.master There is a publish subscribe channel on the node (sentinel:hello).sentinel Node passing __sentinel__:hello Channels exchange information ( For nodes " view " And your own information ), Reach a consensus .

The sentinel principle

But a sentinel process is right Redis The server monitors , There may be problems , So , We can use multiple sentinels to monitor , There will be surveillance between the Sentinels , In this way, the multi sentry mode is formed
sentinel It's a distributed system , You can run multiple Sentinel process . So monitor the same Master Of Sentinel Will automatically connect , Make up a distributed Sentinel The Internet , Communicate with each other and exchange information about the monitored server .

Sentinel troubleshooting principle
1. Subjective offline ：
When the primary server fails , One at this time sentinel Fault found , The system doesn't work right away failover The process （ This phenomenon is called subjective offline ）, It's going to go to the rest of the network Sentinel Confirm .
2. Objective offline ：
And then the rest Sentinel Faults have also been found one after another , At this time, one of them Sentinel Will vote . A certain number of sentinels ( Specified in the configuration file ) confirm Master Marked as subjective offline , At this time will be Master Mark as objective .
3.sentinel Of leader The election ：
To complete failover （ Will malfunction master To eliminate , And will be one slave Upgrade to master） You have to elect one first leader. The first to find the fault sentinel Make a request to other sentinels to become leader, The other sentinels did not agree with the other sentinels leader When asked , Will vote for the sentinel. When more than half sentinel After the vote is passed, it is determined that sentinel by leader. The next failover has this leader complete .
4.master The election ：
leader After selection, the fault will master To eliminate , from slave Choose one of them to be master. The principles followed are as follows ：
● slave The priority of the
● slave from master The amount of data synchronized , that slave More is preferred .
5. new Master Then notify all users through publish subscribe mode sentinel Update monitoring host information .
6. After the failed master server is repaired, it will become the slave server to continue working .

The example is as follows ：
Failure occurs
Fail over

Master After re launching

The advantages of sentinel mode

● The sentry cluster 、 Based on master-slave replication mode , All the advantages of master-slave replication , It will have
● Master slave can switch , Faults can be transferred , The availability of the system will be better
● Sentinel mode is the upgrade of master-slave mode , Manual to automatic , More robust

Disadvantages of sentinel mode

● Redis Bad online expansion , Once the cluster capacity reaches the upper limit , Online expansion is very troublesome
● Implementing the configuration of sentinel mode is cumbersome , There are many choices