Redis (9), redis Cluster use of redis cluster

Premise: redis cluster sharding, currently there are two types of redis solutions

Hash slot (hash slot), representing the scheme: redis cluster

Consistent hash, representative scheme: twemproxy, codis

This article is for redis cluster configuration implementation, the principle will be analyzed in another article. By haoran-10.iteye.com

First understand a few key words:

Node: When a single redis is running, it is isolated. Once a failure occurs, the data will not be lost when the master-slave replication occurs at most, but the redis is unavailable. Redis stand-alone, one node.

Cluster: Many nodes are connected by some scheme to provide redis services together, which is called redis cluster. When a single node fails, the data will be transferred (no strong consistency is guaranteed) to a safe backup node to provide highly available redis services. The disadvantage is that commands that do not support processing multiple keys cannot be used.

Sharding: When there is a lot of redis data, a single redis cannot fit. At this time, the redis cluster distributes the data evenly to different nodes through the sharding technology.

Hash slot: Redis cluster distributes data to different nodes . Redis Cluster has 16384 hash slots, we just use the key's CRC16 encoding to modulo 16384 to calculate the hash slot to which a given key belongs.

1. Key configuration of redis node

daemonize yes
pidfile /mytest/redis_dev/9001/redis.pid
port 9001
appendonly yes
appendfsync everysec
cluster-enabled yes
cluster-config-file nodes.conf  
cluster-node-timeout 15000

For the first configuration, it is strongly recommended to use 6 nodes, so that the configuration and advantages of the redis cluster can be more intuitively understood.

Copy into 6 points, 9001, 9002, 9003, 9004, 9005, 9006 (note to change to different ports, different pid files, ignore nodes.conf, redis will automatically create and manage)

2. Start each node

3. Use the redis-trib tool to create a cluster

After the startup is successful, use the tools that come with redis to create a cluster

Note, here to install tools such as ruby

(1), install ruby

http://jingyan.baidu.com/article/b7001fe173fe9a0e7382dd57.html

(2), install ruby gem

https://rubygems.org/pages/download#formats

(3), install gem redis

https://rubygems.org/gems/redis/versions/3.2.2

The installation process is relatively simple, skip it.

Type "yes" to continue

Hash slots have been allocated to 9001, 9002, 9003 nodes.

4. Test with redis-cli

Note: here is to use the src/redis-cli -c command to start the client cluster operation mode

Get operational tests:

As can be seen from the data, the client enters from port 9002. When set age 27 is entered, it is automatically transferred to node 9001.

When entering get age, data is also obtained from node 9001

5. Fault-tolerant processing test

Manually kill the 9001 node

Continue to operate the redis cluster at this point

Automatically obtain data from the original 9001 slave node 9004 node (9001 is the master node, 9004 is the slave node of 9001), indicating that 9004 automatically becomes the master node

6. Add node test

Create 9007 node, start 9007

And set the 9007 node as the slave node of 9004

src/redis-trib.rb add-node --slave 127.0.0.1:9007 127.0.0.1:9004

From the output, you can see that 9007 is a replica node of 9004, that is, the slave node.

7. Add a master node and re-shard the cluster

redis-trib.rb reshard 127.0.0.1:9008

Mainly to reallocate the hash slot

8. java client

It seems that only sharedjedis can be used. Spring-data-redis does not support the new cluster features of redis. I don't know if the posture is wrong.

package wang.conge.init;

import java.util.HashSet;
import java.util.Set;

import redis.clients.jedis.HostAndPort;
import redis.clients.jedis.JedisCluster;

public class SharedRedisClient {
	public static void main(String[] args) {
		Set<HostAndPort> jedisClusterNodes = new HashSet<HostAndPort>();  
        jedisClusterNodes.add(new HostAndPort("127.0.0.1", 9001));  
        
        JedisCluster jedis = new JedisCluster(jedisClusterNodes);
        
        jedis.set("nginx", "yes");
		jedis.set("tomcat", "yes");
		jedis.set("keepalived", "yes");
		jedis.set("vip", "yes");
		jedis.set("redis", "yes");
		jedis.set("mysql", "yes");
		
		String key = jedis.get("redis");
		System.out.println(key);
	}
}

9. Summary

The configuration of the cluster solution officially provided by redis is too cumbersome, but the performance is relatively strong. The company has not yet used it in production, so we will wait and see.

Redis (9), redis Cluster use of redis cluster

Guess you like