前面的两篇文章(Redis的持久化方案, 一文掌握Redis的三种集群方案)分别介绍了Redis的持久化与集群方案 —— 包括主从复制模式、哨兵模式、Cluster模式,其中主从复制模式由于不能自动做故障转移,当节点出现故障时需要人为干预,不满足生产环境的高可用需求,所以在生产环境一般使用哨兵模式或Cluster模式。那么在Spring Boot项目中,如何访问这两种模式的Redis集群,可能遇到哪些问题,是本文即将介绍的内容。
Spring Boot 2 整合Redis
spring boot中整合Redis非常简单,在pom.xml中添加依赖
<dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-data-redis</artifactId>
</dependency>
spring boot 2的spring-boot-starter-data-redis
中,默认使用的是lettuce作为redis客户端,它与jedis的主要区别如下:
- Jedis是同步的,不支持异步,Jedis客户端实例不是线程安全的,需要每个线程一个Jedis实例,所以一般通过连接池来使用Jedis
- Lettuce是基于Netty框架的事件驱动的Redis客户端,其方法调用是异步的,Lettuce的API也是线程安全的,所以多个线程可以操作单个Lettuce连接来完成各种操作,同时Lettuce也支持连接池
如果不使用默认的lettuce,使用jedis的话,可以排除lettuce的依赖,手动加入jedis依赖,配置如下
<dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-data-redis</artifactId>
<exclusions>
<exclusion>
<groupId>io.lettuce</groupId>
<artifactId>lettuce-core</artifactId>
</exclusion>
</exclusions>
</dependency>
<dependency>
<groupId>redis.clients</groupId>
<artifactId>jedis</artifactId>
<version>2.9.0</version>
</dependency>
在配置文件application.yml中添加配置(针对单实例)
spring:
redis:
host: 192.168.40.201
port: 6379
password: passw0rd
database: 0 # 数据库索引,默认0
timeout: 5000 # 连接超时,单位ms
jedis: # 或lettuce, 连接池配置,springboot2.0中使用jedis或者lettuce配置连接池,默认为lettuce连接池
pool:
max-active: 8 # 连接池最大连接数(使用负值表示没有限制)
max-wait: -1 # 连接池分配连接最大阻塞等待时间(阻塞时间到,抛出异常。使用负值表示无限期阻塞)
max-idle: 8 # 连接池中的最大空闲连接数
min-idle: 0 # 连接池中的最小空闲连接数
然后添加配置类。其中@EnableCaching注解是为了使@Cacheable、@CacheEvict、@CachePut、@Caching注解生效
@Configuration
@EnableCaching
public class RedisConfig {
@Bean
public RedisTemplate<String, Object> redisTemplate(RedisConnectionFactory factory) {
RedisTemplate<String, Object> template = new RedisTemplate<>();
template.setConnectionFactory(factory);
// 使用Jackson2JsonRedisSerialize 替换默认的jdkSerializeable序列化
Jackson2JsonRedisSerializer jackson2JsonRedisSerializer = new Jackson2JsonRedisSerializer(Object.class);
ObjectMapper om = new ObjectMapper();
om.setVisibility(PropertyAccessor.ALL, JsonAutoDetect.Visibility.ANY);
om.enableDefaultTyping(ObjectMapper.DefaultTyping.NON_FINAL);
jackson2JsonRedisSerializer.setObjectMapper(om);
StringRedisSerializer stringRedisSerializer = new StringRedisSerializer();
// key采用String的序列化方式
template.setKeySerializer(stringRedisSerializer);
// hash的key也采用String的序列化方式
template.setHashKeySerializer(stringRedisSerializer);
// value序列化方式采用jackson
template.setValueSerializer(jackson2JsonRedisSerializer);
// hash的value序列化方式采用jackson
template.setHashValueSerializer(jackson2JsonRedisSerializer);
template.afterPropertiesSet();
return template;
}
}
上述配置类注入了自定义的RedisTemplate<String, Object>, 替换RedisAutoConfiguration中自动配置的RedisTemplate<Object, Object>类(RedisAutoConfiguration另外还自动配置了StringRedisTemplate)。
此时,我们可以通过定义一个基于RedisTemplate的工具类,或通过在Service层添加@Cacheable、@CacheEvict、@CachePut、@Caching注解来使用缓存。比如定义一个RedisService类,封装常用的Redis操作方法,
@Component
@Slf4j
public class RedisService {
@Autowired
private RedisTemplate<String, Object> redisTemplate;
/**
* 指定缓存失效时间
*
* @param key 键
* @param time 时间(秒)
* @return
*/
public boolean expire(String key, long time) {
try {
if (time > 0) {
redisTemplate.expire(key, time, TimeUnit.SECONDS);
}
return true;
} catch (Exception e) {
log.error("exception when expire key {}. ", key, e);
return false;
}
}
/**
* 根据key获取过期时间
*
* @param key 键 不能为null
* @return 时间(秒) 返回0代表为永久有效
*/
public long getExpire(String key) {
return redisTemplate.getExpire(key, TimeUnit.SECONDS);
}
/**
* 判断key是否存在
*
* @param key 键
* @return true 存在 false不存在
*/
public boolean hasKey(String key) {
try {
return redisTemplate.hasKey(key);
} catch (Exception e) {
log.error("exception when check key {}. ", key, e);
return false;
}
}
...
}
出于篇幅,完整代码请查阅本文示例源码: https://github.com/ronwxy/springboot-demos/tree/master/springboot-redis-sentinel
或在Service层使用注解,如
@Service
@CacheConfig(cacheNames = "users")
public class UserService {
private static Map<String, User> userMap = new HashMap<>();
@CachePut(key = "#user.username")
public User addUser(User user){
user.setUid(UUID.randomUUID().toString());
System.out.println("add user: " + user);
userMap.put(user.getUsername(), user);
return user;
}
@Caching(put = {
@CachePut( key = "#user.username"),
@CachePut( key = "#user.uid")
})
public User addUser2(User user) {
user.setUid(UUID.randomUUID().toString());
System.out.println("add user2: " + user);
userMap.put(user.getUsername(), user);
return user;
}
...
}
Spring Boot 2 整合Redis哨兵模式
Spring Boot 2 整合Redis哨兵模式除了配置稍有差异,其它与整合单实例模式类似,配置示例为
spring:
redis:
password: passw0rd
timeout: 5000
sentinel:
master: mymaster
nodes: 192.168.40.201:26379,192.168.40.201:36379,192.168.40.201:46379 # 哨兵的IP:Port列表
jedis: # 或lettuce
pool:
max-active: 8
max-wait: -1
max-idle: 8
min-idle: 0
上述配置只指定了哨兵节点的地址与master的名称,但Redis客户端最终访问操作的是master节点,那么Redis客户端是如何获取master节点的地址,并在发生故障转移时,如何自动切换master地址的呢?我们以Jedis连接池为例,通过源码来揭开其内部实现的神秘面纱。
在 JedisSentinelPool 类的构造函数中,对连接池做了初始化,如下
public JedisSentinelPool(String masterName, Set<String> sentinels,
final GenericObjectPoolConfig poolConfig, final int connectionTimeout, final int soTimeout,
final String password, final int database, final String clientName) {
this.poolConfig = poolConfig;
this.connectionTimeout = connectionTimeout;
this.soTimeout = soTimeout;
this.password = password;
this.database = database;
this.clientName = clientName;
HostAndPort master = initSentinels(sentinels, masterName);
initPool(master);
}
private HostAndPort initSentinels(Set<String> sentinels, final String masterName) {
for (String sentinel : sentinels) {
final HostAndPort hap = HostAndPort.parseString(sentinel);
log.fine("Connecting to Sentinel " + hap);
Jedis jedis = null;
try {
jedis = new Jedis(hap.getHost(), hap.getPort());
List<String> masterAddr = jedis.sentinelGetMasterAddrByName(masterName);
// connected to sentinel...
sentinelAvailable = true;
if (masterAddr == null || masterAddr.size() != 2) {
log.warning("Can not get master addr, master name: " + masterName + ". Sentinel: " + hap
+ ".");
continue;
}
master = toHostAndPort(masterAddr);
log.fine("Found Redis master at " + master);
break;
} catch (JedisException e) {
// resolves #1036, it should handle JedisException there's another chance
// of raising JedisDataException
log.warning("Cannot get master address from sentinel running @ " + hap + ". Reason: " + e
+ ". Trying next one.");
} finally {
if (jedis != null) {
jedis.close();
}
}
}
//省略了非关键代码
for (String sentinel : sentinels) {
final HostAndPort hap = HostAndPort.parseString(sentinel);
MasterListener masterListener = new MasterListener(masterName, hap.getHost(), hap.getPort());
// whether MasterListener threads are alive or not, process can be stopped
masterListener.setDaemon(true);
masterListeners.add(masterListener);
masterListener.start();
}
return master;
}
initSentinels
方法中主要干了两件事:
- 遍历哨兵节点,通过
get-master-addr-by-name
命令获取master节点的地址信息,找到了就退出循环。get-master-addr-by-name
命令执行结果如下所示
[root@dev-server-1 master-slave]# redis-cli -p 26379
127.0.0.1:26379> sentinel get-master-addr-by-name mymaster
1) "192.168.40.201"
2) "7001"
127.0.0.1:26379>
- 对每一个哨兵节点通过一个 MasterListener 进行监听(Redis的发布订阅功能),订阅哨兵节点
+switch-master
频道,当发生故障转移时,客户端能收到哨兵的通知,通过重新初始化连接池,完成主节点的切换。
MasterListener.run方法中监听哨兵部分代码如下
j.subscribe(new JedisPubSub() {
@Override
public void onMessage(String channel, String message) {
log.fine("Sentinel " + host + ":" + port + " published: " + message + ".");
String[] switchMasterMsg = message.split(" ");
if (switchMasterMsg.length > 3) {
if (masterName.equals(switchMasterMsg[0])) {
initPool(toHostAndPort(Arrays.asList(switchMasterMsg[3], switchMasterMsg[4])));
} else {
log.fine("Ignoring message on +switch-master for master name "
+ switchMasterMsg[0] + ", our master name is " + masterName);
}
} else {
log.severe("Invalid message received on Sentinel " + host + ":" + port
+ " on channel +switch-master: " + message);
}
}
}, "+switch-master");
initPool 方法如下:如果发现新的master节点与当前的master不同,则重新初始化。
private void initPool(HostAndPort master) {
if (!master.equals(currentHostMaster)) {
currentHostMaster = master;
if (factory == null) {
factory = new JedisFactory(master.getHost(), master.getPort(), connectionTimeout,
soTimeout, password, database, clientName, false, null, null, null);
initPool(poolConfig, factory);
} else {
factory.setHostAndPort(currentHostMaster);
// although we clear the pool, we still have to check the
// returned object
// in getResource, this call only clears idle instances, not
// borrowed instances
internalPool.clear();
}
log.info("Created JedisPool to master at " + master);
}
}
通过以上两步,Jedis客户端在只知道哨兵地址的情况下便能获得master节点的地址信息,并且当发生故障转移时能自动切换到新的master节点地址。
Spring Boot 2 整合Redis Cluster模式
Spring Boot 2 整合Redis Cluster模式除了配置稍有差异,其它与整合单实例模式也类似,配置示例为
spring:
redis:
password: passw0rd
timeout: 5000
database: 0
cluster:
nodes: 192.168.40.201:7100,192.168.40.201:7200,192.168.40.201:7300,192.168.40.201:7400,192.168.40.201:7500,192.168.40.201:7600
max-redirects: 3 # 重定向的最大次数
jedis:
pool:
max-active: 8
max-wait: -1
max-idle: 8
min-idle: 0
在 一文掌握Redis的三种集群方案 中已经介绍了Cluster模式访问的基本原理,可以通过任意节点跳转到目标节点执行命令,上面配置中 max-redirects 控制在集群中跳转的最大次数。
查看JedisClusterConnection的execute方法,
public Object execute(String command, byte[]... args) {
Assert.notNull(command, "Command must not be null!");
Assert.notNull(args, "Args must not be null!");
return clusterCommandExecutor
.executeCommandOnArbitraryNode((JedisClusterCommandCallback<Object>) client -> JedisClientUtils.execute(command,
EMPTY_2D_BYTE_ARRAY, args, () -> client))
.getValue();
}
集群命令的执行是通过ClusterCommandExecutor.executeCommandOnArbitraryNode
来实现的,
public <T> NodeResult<T> executeCommandOnArbitraryNode(ClusterCommandCallback<?, T> cmd) {
Assert.notNull(cmd, "ClusterCommandCallback must not be null!");
List<RedisClusterNode> nodes = new ArrayList<>(getClusterTopology().getActiveNodes());
return executeCommandOnSingleNode(cmd, nodes.get(new Random().nextInt(nodes.size())));
}
private <S, T> NodeResult<T> executeCommandOnSingleNode(ClusterCommandCallback<S, T> cmd, RedisClusterNode node,
int redirectCount) {
Assert.notNull(cmd, "ClusterCommandCallback must not be null!");
Assert.notNull(node, "RedisClusterNode must not be null!");
if (redirectCount > maxRedirects) {
throw new TooManyClusterRedirectionsException(String.format(
"Cannot follow Cluster Redirects over more than %s legs. Please consider increasing the number of redirects to follow. Current value is: %s.",
redirectCount, maxRedirects));
}
RedisClusterNode nodeToUse = lookupNode(node);
S client = this.resourceProvider.getResourceForSpecificNode(nodeToUse);
Assert.notNull(client, "Could not acquire resource for node. Is your cluster info up to date?");
try {
return new NodeResult<>(node, cmd.doInCluster(client));
} catch (RuntimeException ex) {
RuntimeException translatedException = convertToDataAccessException(ex);
if (translatedException instanceof ClusterRedirectException) {
ClusterRedirectException cre = (ClusterRedirectException) translatedException;
return executeCommandOnSingleNode(cmd,
topologyProvider.getTopology().lookup(cre.getTargetHost(), cre.getTargetPort()), redirectCount + 1);
} else {
throw translatedException != null ? translatedException : ex;
}
} finally {
this.resourceProvider.returnResourceForSpecificNode(nodeToUse, client);
}
}
上述代码逻辑如下
- 从集群节点列表中随机选择一个节点
- 从该节点获取一个客户端连接(如果配置了连接池,从连接池中获取),执行命令
- 如果抛出ClusterRedirectException异常,则跳转到返回的目标节点上执行
- 如果跳转次数大于配置的值 max-redirects, 则抛出TooManyClusterRedirectionsException异常
可能遇到的问题
- Redis连接超时
检查服务是否正常启动(比如 ps -ef|grep redis
查看进程,netstat -ano|grep 6379
查看端口是否起来,以及日志文件),如果正常启动,则查看Redis服务器是否开启防火墙,关闭防火墙或配置通行端口。
- Cluster模式下,报连接到127.0.0.1被拒绝错误,如
Connection refused: no further information: /127.0.0.1:7600
这是因为在redis.conf中配置 bind 0.0.0.0
或 bind 127.0.0.1
导致,需要改为具体在外部可访问的IP,如 bind 192.168.40.201
。如果之前已经起了集群,并产生了数据,则修改redis.conf文件后,还需要修改cluster-config-file文件,将127.0.0.1替换为bind 的具体IP,然后重启。
- master挂了,slave升级成为master,重启master,不能正常同步新的master数据
如果设置了密码,需要在master, slave的配置文件中都配置masterauth password
相关阅读:
作者:空山新雨,一枚仍在学习路上的IT老兵
近期作者写了几十篇技术博客,内容包括Java、Spring Boot、Spring Cloud、Docker,技术管理心得等
欢迎关注作者微信公众号:空山新雨的技术空间,一起学习成长