Redis cluster

Redis And cluster colony

1. Cluster introduction

Cluster is to use the network to connect several computers , And provide a unified way of Management , Make it present the service effect of single machine

2. Cluster function

1. Disperse the access pressure of a single server , Load balancing
2. Disperse the storage pressure of a single server , Achieve scalability
3. Reduce the business disaster caused by single server downtime

3. Cluster data storage design

%16384 It's modular operation
If you add another server , Or a server suddenly goes down , What do I do ？

37 The small box is called ” Slot “

If you add another server now , Or a server suddenly goes down , What do I do ？
The following example is 3 Servers , Another server has been added , Turned into 4 individual , Then each of the other three servers takes out a part key To the new server , To optimize .
The so-called adding nodes and removing nodes actually change the location of the slot

4. The design of communication within the cluster

1. Now there are clients to access the database , Conduct key Lookup ,key After being calculated by two algorithms , obtain key Location of the corresponding storage slot , Suppose the result of the calculation is A Inside 35 Slot No , If one hit, it will return directly , If you miss ,35 Slot number will be based on this key Search for , Tell the client that it is B In that slot , The client will go again B Find the slot inside （ Be careful ： No A Go find , Is the client to find , Avoid clients from going to other places for many times and looking for them one by one ）
2. You can hit it twice at most
Be careful ： The number of slots in each database is not necessarily continuous , It's scattered

5. Cluster building method

1. Native installation （ Single command ）
1) Configure server （3 Lord 3 from ）
2) Establish communication （Meet）
3) Slotting （Slot）
4) Build master slave （master-slave）
2. Tool installation （ The batch )

6.cluster To configure

1.cluster-enabled yes|no
Add the server as a node
2.cluster-config-file
cluster Profile name , This file is automatically generated .
If we give you a name , Will automatically generate the specified name cluster-config-file ;
If we don't give you a name , A default name will be automatically generated cluster-config-file ;
give an example ：cluster-config-file nodes-6379.conf
And it's only used to quickly find files .
（ notes ： We try to assign a name , Because it is generated by default in the case of a large number of nodes cluster-config-file The file names are the same , Easy to confuse ）
3.cluster-node-timeout
Node service response timeout , Used to determine whether the node is offline or switched to a slave node
4.cluster-migration-barrier
master Connected slave Minimum quantity

7.cluster Node operation command

1. View the cluster node information

cluster nodes

2. Enter a slave node redis, Switch its master node

cluster replicate <master-id>

3. Find a new node , Add master node

cluster meet ip:port

4. Ignore one without solt The node of

cluster forget <id>

5. Manual failover

cluster failover

8.redis-trib command

redis-trib Equivalent to a script , hold cluster Some of the commands are optimized , In practice cluster Own command or redis-trib Everything can be ordered

1. Add a node

redis-trib.rb add-node

2. Delete node

redis-trib.rb del-node

3. The shard

redis-trib.rb reshard