Project practice, redis cluster technology learning (6)

4. Cluster integrity check

Cluster integrity means that all slots are allocated to the surviving master node , as long as 16384 If one of the slots is not assigned to a node, the cluster is incomplete . have access to redis-trib.rb check The command checks whether the two clusters created before are successful ,check The command only needs to give the address of any node in the cluster to complete the inspection of the whole cluster , The order is as follows ：

redis-trib.rb check 127.0.0.1:6379

redis-trib.rb check 127.0.0.1:6481

When the following information is finally output , Prompt that all slots in the cluster have been assigned to nodes ：

[OK] All nodes agree about slots configuration.

>>> Check for open slots...

>>> Check slots coverage...

[OK] All 16384 slots covered.

Redis.3 Node communication

Redis.3.1 Communication process

In distributed storage, we need to provide a mechanism to maintain node metadata information , Metadata means ： node

What data are you responsible for , Whether there is fault and other status information . Common metadata maintenance methods include ： focus

The type and P2P The way .Redis The cluster using P2P Of Gossip（ Gossip ） agreement ,Gossip agreement

The working principle is that nodes constantly communicate with each other and exchange information , After a period of time, all nodes will know the set

Group complete information , This way is similar to spreading rumors

Communication process description ：

1） Each node in the cluster will have a separate TCP passageway , For communication between nodes ,

The communication port number is added to the basic port 10000.

2） Each node selects several nodes to send through specific rules in a fixed period ping news .

3） Received ping The node of the message uses pong Message in response .

Each node in the cluster selects the node to communicate through certain rules , Each node may know all sections

spot , Or maybe only some nodes are known , As long as these nodes can communicate with each other normally , Eventually they will reach

To a consistent state . When a node fails 、 New nodes join 、 The role of master and slave changes 、 Slot information change and so on

When this happens , Through constant ping/pong Message communication , After a period of time, all nodes will know the latest status of all nodes in the whole cluster , So as to achieve the purpose of cluster state synchronization .