1 Redis集群
1.1 redis-cluster架构图
架构细节:
(1)所有的redis节点彼此互联(PING-PONG机制),内部使用二进制协议优化传输速度和带宽.
(2)节点的fail是通过集群中超过半数的节点检测失效时才生效.
(3)客户端与redis节点直连,不需要中间proxy层.客户端不需要连接集群所有节点,连接集群中任何一个可用节点即可
(4)redis-cluster把所有的物理节点映射到[0-16383]slot上,cluster 负责维护node<->slot<->value
Redis 集群中内置了 16384 个哈希槽，当需要在 Redis 集群中放置一个 key-value 时，redis 先对 key 使用 crc16 算法算出一个结果，然后把结果对 16384 求余数，这样每个 key 都会对应一个编号在 0-16383 之间的哈希槽，redis 会根据节点数量大致均等的将哈希槽映射到不同的节点
示例如下：
1.2 redis-cluster投票:容错
(1)集群中所有master参与投票,如果半数以上master节点与其中一个master节点通信超过(cluster-node-timeout),认为该master节点挂掉.
(2):什么时候整个集群不可用(cluster_state:fail)?
- 如果集群任意master挂掉,且当前master没有slave，则集群进入fail状态。也可以理解成集群的[0-16383]slot映射不完全时进入fail状态。
- 如果集群超过半数以上master挂掉，无论是否有slave，集群进入fail状态。
1.3 安装ruby
集群管理工具（redis-trib.rb）是使用ruby脚本语言编写的。
第一步：安装ruby
[root@qfjava bin2]# yum install ruby
[root@qfjava bin2]# yum install rubygems
第二步：将以下文件上传到linux系统
第三步：安装ruby和redis接口
[root@qfjava ~]# gem install redis-3.0.0.gem
第四步：将redis-3.0.0包下src目录中的以下文件拷贝到redis19/redis-cluster/
[root@qfjava src]# cd /usr/local/redis19/
[root@qfjava redis19]# mkdir redis-cluster
[root@qfjava redis19]# cd /root/redis-3.0.0/src/
[root@qfjava src]# cp redis-trib.rb /usr/local/redis19/redis-cluster
第五步：查看是否拷贝成功
1.4 搭建集群
搭建集群最少也得需要3台主机，如果每台主机再配置一台从机的话，则最少需要6台机器。
端口设计如下：7001-7006
第一步：复制出一个7001机器
[root@qfjava redis19]# cp bin ./redis-cluster/7001 –r
第二步：如果存在持久化文件，则删除
[root@qfjava 7001]# rm -rf appendonly.aof dump.rdb
第三步：设置集群参数
第四步：修改端口
第五步：复制出7002-7006机器
[root@qfjava redis-cluster]# cp 7001/ 7002 -r
[root@qfjava redis-cluster]# cp 7001/ 7003 -r
[root@qfjava redis-cluster]# cp 7001/ 7004 -r
[root@qfjava redis-cluster]# cp 7001/ 7005 -r
[root@qfjava redis-cluster]# cp 7001/ 7006 –r
第六步：修改7002-7006机器的端口
第七步：启动7001-7006这六台机器
第九步：创建集群
|
[root@qfjava redis-cluster]# ./redis-trib.rb create --replicas 1 192.168.242.137:7001 192.168.242.137:7002 192.168.242.137:7003 192.168.242.137:7004 192.168.242.137:7005 192.168.242.137:7006 >>> Creating cluster Connecting to node 192.168.242.137:7001: OK Connecting to node 192.168.242.137:7002: OK Connecting to node 192.168.242.137:7003: OK Connecting to node 192.168.242.137:7004: OK Connecting to node 192.168.242.137:7005: OK Connecting to node 192.168.242.137:7006: OK >>> Performing hash slots allocation on 6 nodes... Using 3 masters: 192.168.242.137:7001 192.168.242.137:7002 192.168.242.137:7003 Adding replica 192.168.242.137:7004 to 192.168.242.137:7001 Adding replica 192.168.242.137:7005 to 192.168.242.137:7002 Adding replica 192.168.242.137:7006 to 192.168.242.137:7003 M: 8240cd0fe6d6f842faa42b0174fe7c5ddcf7ae24 192.168.242.137:7001 slots:0-5460 (5461 slots) master M: 4f52a974f64343fd9f1ee0388490b3c0647a4db7 192.168.242.137:7002 slots:5461-10922 (5462 slots) master M: cb7c5def8f61df2016b38972396a8d1f349208c2 192.168.242.137:7003 slots:10923-16383 (5461 slots) master S: 66adf006fed43b3b5e499ce2ff1949a756504a16 192.168.242.137:7004 replicates 8240cd0fe6d6f842faa42b0174fe7c5ddcf7ae24 S: cbb0c9bc4b27dd85511a7ef2d01bec90e692793b 192.168.242.137:7005 replicates 4f52a974f64343fd9f1ee0388490b3c0647a4db7 S: a908736eadd1cd06e86fdff8b2749a6f46b38c00 192.168.242.137:7006 replicates cb7c5def8f61df2016b38972396a8d1f349208c2 Can I set the above configuration? (type 'yes' to accept): yes >>> Nodes configuration updated >>> Assign a different config epoch to each node >>> Sending CLUSTER MEET messages to join the cluster Waiting for the cluster to join.. >>> Performing Cluster Check (using node 192.168.242.137:7001) M: 8240cd0fe6d6f842faa42b0174fe7c5ddcf7ae24 192.168.242.137:7001 slots:0-5460 (5461 slots) master M: 4f52a974f64343fd9f1ee0388490b3c0647a4db7 192.168.242.137:7002 slots:5461-10922 (5462 slots) master M: cb7c5def8f61df2016b38972396a8d1f349208c2 192.168.242.137:7003 slots:10923-16383 (5461 slots) master M: 66adf006fed43b3b5e499ce2ff1949a756504a16 192.168.242.137:7004 slots: (0 slots) master replicates 8240cd0fe6d6f842faa42b0174fe7c5ddcf7ae24 M: cbb0c9bc4b27dd85511a7ef2d01bec90e692793b 192.168.242.137:7005 slots: (0 slots) master replicates 4f52a974f64343fd9f1ee0388490b3c0647a4db7 M: a908736eadd1cd06e86fdff8b2749a6f46b38c00 192.168.242.137:7006 slots: (0 slots) master replicates cb7c5def8f61df2016b38972396a8d1f349208c2 [OK] All nodes agree about slots configuration. >>> Check for open slots... >>> Check slots coverage... [OK] All 16384 slots covered. [root@qfjava redis-cluster]# |
1.5 连接集群
[root@qfjava 7001]# ./redis-cli -h 192.168.242.137 -p 7001 –c
-c：指定是集群连接
1.6 查看集群信息
l 查看集群信息
|
l 192.168.242.137:7002> cluster info l cluster_state:ok l cluster_slots_assigned:16384 l cluster_slots_ok:16384 l cluster_slots_pfail:0 l cluster_slots_fail:0 l cluster_known_nodes:6 l cluster_size:3 l cluster_current_epoch:6 l cluster_my_epoch:2 l cluster_stats_messages_sent:2372 l cluster_stats_messages_received:2372 l 192.168.242.137:7002> |
l 查看集群节点
|
l 192.168.242.137:7002> cluster nodes l 8240cd0fe6d6f842faa42b0174fe7c5ddcf7ae24 192.168.242.137:7001 master - 0 1451581348093 1 connected 0-5460 l cb7c5def8f61df2016b38972396a8d1f349208c2 192.168.242.137:7003 master - 0 1451581344062 3 connected 10923-16383 l 66adf006fed43b3b5e499ce2ff1949a756504a16 192.168.242.137:7004 slave 8240cd0fe6d6f842faa42b0174fe7c5ddcf7ae24 0 1451581351115 1 connected l a908736eadd1cd06e86fdff8b2749a6f46b38c00 192.168.242.137:7006 slave cb7c5def8f61df2016b38972396a8d1f349208c2 0 1451581349101 3 connected l 4f52a974f64343fd9f1ee0388490b3c0647a4db7 192.168.242.137:7002 myself,master - 0 0 2 connected 5461-10922 l cbb0c9bc4b27dd85511a7ef2d01bec90e692793b 192.168.242.137:7005 slave 4f52a974f64343fd9f1ee0388490b3c0647a4db7 0 1451581350108 5 connected |
2 jedis连接集群
2.1 设置防火墙
|
[root@qfjava redis-cluster]# vim /etc/sysconfig/iptables -A INPUT -m state --state NEW -m tcp -p tcp --dport 6379 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 6379 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 6379 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 6379 -j ACCEPT # Firewall configuration written by system-config-firewall # Manual customization of this file is not recommended. *filter :INPUT ACCEPT [0:0] :FORWARD ACCEPT [0:0] :OUTPUT ACCEPT [0:0] -A INPUT -m state --state ESTABLISHED,RELATED -j ACCEPT -A INPUT -p icmp -j ACCEPT -A INPUT -i lo -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 22 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 3306 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 8080 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 6379 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 7001 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 7002 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 7003 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 7004 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 7005 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 7006 -j ACCEPT -A INPUT -m state --state NEW -m tcp -p tcp --dport 7007 -j ACCEPT -A INPUT -j REJECT --reject-with icmp-host-prohibited -A FORWARD -j REJECT --reject-with icmp-host-prohibited COMMIT ~ ~ ~ ~ "/etc/sysconfig/iptables" 23L, 1146C 已写入 [root@qfjava redis-cluster]# service iptables restart iptables：清除防火墙规则： [确定] iptables：将链设置为政策 ACCEPT：filter [确定] iptables：正在卸载模块： [确定] iptables：应用防火墙规则： [确定] [root@qfjava redis-cluster]# |
2.2 使用spring
- 配置applicationContext.xml
|
<!-- 连接池配置 --> <bean id="jedisPoolConfig" class="redis.clients.jedis.JedisPoolConfig"> <!-- 最大连接数 --> <property name="maxTotal" value="30" /> <!-- 最大空闲连接数 --> <property name="maxIdle" value="10" /> <!-- 每次释放连接的最大数目 --> <property name="numTestsPerEvictionRun" value="1024" /> <!-- 释放连接的扫描间隔（毫秒） --> <property name="timeBetweenEvictionRunsMillis" value="30000" /> <!-- 连接最小空闲时间 --> <property name="minEvictableIdleTimeMillis" value="1800000" /> <!-- 连接空闲多久后释放, 当空闲时间>该值 且 空闲连接>最大空闲连接数 时直接释放 --> <property name="softMinEvictableIdleTimeMillis" value="10000" /> <!-- 获取连接时的最大等待毫秒数,小于零:阻塞不确定的时间,默认-1 --> <property name="maxWaitMillis" value="1500" /> <!-- 在获取连接的时候检查有效性, 默认false --> <property name="testOnBorrow" value="true" /> <!-- 在空闲时检查有效性, 默认false --> <property name="testWhileIdle" value="true" /> <!-- 连接耗尽时是否阻塞, false报异常,ture阻塞直到超时, 默认true --> <property name="blockWhenExhausted" value="false" /> </bean> <!-- redis集群 --> <bean id="jedisCluster" class="redis.clients.jedis.JedisCluster"> <constructor-arg index="0"> <set> <bean class="redis.clients.jedis.HostAndPort"> <constructor-arg index="0" value="192.168.101.3"></constructor-arg> <constructor-arg index="1" value="7001"></constructor-arg> </bean> <bean class="redis.clients.jedis.HostAndPort"> <constructor-arg index="0" value="192.168.101.3"></constructor-arg> <constructor-arg index="1" value="7002"></constructor-arg> </bean> <bean class="redis.clients.jedis.HostAndPort"> <constructor-arg index="0" value="192.168.101.3"></constructor-arg> <constructor-arg index="1" value="7003"></constructor-arg> </bean> <bean class="redis.clients.jedis.HostAndPort"> <constructor-arg index="0" value="192.168.101.3"></constructor-arg> <constructor-arg index="1" value="7004"></constructor-arg> </bean> <bean class="redis.clients.jedis.HostAndPort"> <constructor-arg index="0" value="192.168.101.3"></constructor-arg> <constructor-arg index="1" value="7005"></constructor-arg> </bean> <bean class="redis.clients.jedis.HostAndPort"> <constructor-arg index="0" value="192.168.101.3"></constructor-arg> <constructor-arg index="1" value="7006"></constructor-arg> </bean> </set> </constructor-arg> <constructor-arg index="1" ref="jedisPoolConfig"></constructor-arg> </bean> |
- 测试代码
|
private ApplicationContext applicationContext; @Before public void init() { applicationContext = new ClassPathXmlApplicationContext( "classpath:applicationContext.xml"); }
// redis集群 @Test public void testJedisCluster() { JedisCluster jedisCluster = (JedisCluster) applicationContext .getBean("jedisCluster");
jedisCluster.set("name", "zhangsan"); String value = jedisCluster.get("name"); System.out.println(value); } |
关于Redis的一些面试题：
http://blog.itpub.net/31545684/viewspace-2213990/
Redis高效原因--epoll多路复用模型
补充：
1、Java bio 模型，阻塞进程式 NIO
2、Linux select 模型，变更触发轮询查找，有1024个数量的限制
3、epoll模型，变更触发回调直接读取，理论无上限
https://blog.csdn.net/wxy941011/article/details/80274233
NIO, Redis,Nginx,Netty 高效的原因都是因为底层使用了epoll VS select/poll