|
Shrio+Redis实现tomcat集群session共享
一、背景
当我们使用了nginx做项目集群以后,就会出现一个很严重的问题亟待解决,那就是:tomcat集群之间如何实现session共享的问题,如果这个问题不解决,就会出现登陆过后再次请求资源依旧需要登陆的问题。这篇文章我们就解决这个问题。
二、实现步骤
说明:本篇是在spring+shiro集成的基础上进行改进的,如果不知道spring和shiro怎么集成,请移步:spring集成shiro做登陆认证
1.在pom.xml中添加shiro-redis和jedis的依赖
org.crazycake
shiro-redis
2.4.2.1-RELEASE
redis.clients
jedis
2.7.2
2.首先我们需要对redis进行集成,在resources下新建config.properties
#redis pool config
redis.pool.maxActive=200
redis.pool.maxIdle=100
redis.pool.maxWait=100
redis.pool.testOnBorrow=true
#redis config
redis.host=192.168.85.129
redis.port=6379
redis.timeout=2000
redis.password=123456
redis.dbindex=8
redis.default.expire=1800000
3.在resources/spring文件夹下新建spring-redis.xml来集成redis操作客户端
Redis configuration
4.添加redisClient.java作为访问redis的客户端
package com.hafiz.www.redis;
import org.crazycake.shiro.RedisManager;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import redis.clients.jedis.Jedis;
import redis.clients.jedis.JedisPool;
import java.util.Set;
/**
* Desc: Jedis 操作客户端
* Created by hafiz.zhang on 2017/7/21.
*/
public> private static final Logger LOGGER = LoggerFactory.getLogger(RedisClient.class);
private static JedisPool jedisPool = null;
public RedisClient(JedisPool jedisPool) {
this.jedisPool = jedisPool;
}
public void init() {
super.init();
}
@Override
public byte[] get(byte[] key) {
Jedis jedis = jedisPool.getResource();
byte[] value;
try {
value = jedis.get(key);
} catch (Exception e) {
LOGGER.error("redis key:{} get value occur exception", new String(key));
throw new RuntimeException("redis operation error:", e);
} finally {
jedis.close();
}
return value;
}
@Override
public byte[] set(byte[] key, byte[] value) {
Jedis jedis = jedisPool.getResource();
try {
jedis.set(key, value);
Integer expire = getExpire();
if(expire != 0) {
jedis.expire(key, expire);
}
} catch (Exception e) {
LOGGER.error("redis key:{} set value:{} occur exception", new String(key), new String(value));
throw new RuntimeException("redis operation error:", e);
} finally {
jedis.close();
}
return value;
}
@Override
public byte[] set(byte[] key, byte[] value, int expire) {
Jedis jedis = jedisPool.getResource();
try {
jedis.set(key, value);
if(expire != 0) {
jedis.expire(key, expire);
}
} catch (Exception e) {
LOGGER.error("redis key:{} set value:{} in expire:{} occur exception", new String(key), new String(value), expire);
throw new RuntimeException("redis operation error:", e);
} finally {
jedis.close();
}
return value;
}
@Override
public void del(byte[] key) {
Jedis jedis = jedisPool.getResource();
try {
jedis.del(key);
} catch (Exception e) {
LOGGER.error("redis key:{} del value occur exception", new String(key));
throw new RuntimeException("redis operation error:", e);
} finally {
jedis.close();
}
}
@Override
public void flushDB() {
Jedis jedis = jedisPool.getResource();
try {
jedis.flushDB();
} catch (Exception e) {
LOGGER.error("redis flushDB occur exception");
throw new RuntimeException("redis operation error:", e);
} finally {
jedis.close();
}
}
@Override
public Long dbSize() {
Long dbSize = Long.valueOf(0L);
Jedis jedis = jedisPool.getResource();
try {
dbSize = jedis.dbSize();
} catch (Exception e) {
LOGGER.error("redis get dbSize occur exception");
throw new RuntimeException("redis operation error:", e);
} finally {
jedis.close();
}
return dbSize;
}
@Override
public Set keys(String pattern) {
Set keys = null;
Jedis jedis = jedisPool.getResource();
try {
keys = jedis.keys(pattern.getBytes());
} catch (Exception e) {
LOGGER.error("redis get keys in pattern:{} occur exception", pattern);
throw new RuntimeException("redis operation error:", e);
} finally {
jedis.close();
}
return keys;
}
}
5.接着,我们对spring-shiro.xml做如下修改(红色标记的部分)
Shiro Configuration
/login.json = anon
/logout.json = anon
/js/** = anon
/ = authc
/** = auth
6.spring.xml做如下修改
到此我们就完了shiro+redis实现session共享的问题,其实也很简单,其中的实现逻辑也很简单,就是shiro的拦截器会首先去redis里面获取session,作为当本次请求的session.
其他代码以及简单测试代码不再贴出,给出github地址:https://github.com/hafizzhang/shiro-session-cluster.git
三、总结
通过本文,我们就完成了spring+shiro+redis实现集群session共享的问题,经过亲测可行。但有一点遗憾的地方,就是每次请求至少会有8次redis的读操作,一次写操作,这个问题还没有找到很好的解决办法,本地使用ehcahe也不太现实,因为涉及到本地缓存和redis缓存同步的问题,如果你有好的办法,欢迎讨论交流学习!
|
|
|