一、新MQ组件的打开方式
我们来举一个例子：商城服务在创建店铺后，需要发MQ事件，由IOT服务和租赁服务处理后续的逻辑，那么MQ的生产者是商城，MQ的消费者分别是IOT、租赁。
首先定义一个店铺同步事件ShopSyncEvent：
@Data
@Builder
@AllArgsConstructor
@NoArgsConstructor
public class ShopSyncEvent extends MQEvent {
// 店铺ID
private String shopId;
// 店铺名称
private String shopName;
// 店铺详情
private JSONObject shopInfo;
@Override
public String getTopic() {
return "mall";
}
@Override
public String getTag() {
return "shop_sync";
}
}MQ生产者（商城服务）发布MQ事件：
// 在创建店铺成功后，发布店铺同步MQ事件
ShopSyncEvent.builder().shopId(shopInfo.getId()).shopName(shopInfo.getName())
.shopInfo(new JSONObject(shopInfo)).build().tenantId(shopInfo.getTenantId())
.publish();MQ消费者（IOT服务、租赁服务）分别监听MQ事件：
@Slf4j
@Component
public class MallMQListener {
/**
* 店铺同步
*/
@MQEventListener(topic = "mall", tags = "shop_sync")
public void onShopSync(ShopSyncEvent event) {
log.info("店铺同步至租赁/IOT：{}", event.getShopInfo());
// TODO 具体的消费逻辑
}
}那么问题来了，具体的MQ在框架内是如何实现的？好学的你此时选择了继续往下看~
二、定义核心组件
首先我们使用门面模式去设计一个自定义注解@MQEventListener，该注解与具体的MQ实现无关，是最轻量的存在：
/**
* MQ事件监听器，配合MQEvent使用
*/
@Documented
@Target(ElementType.METHOD)
@Retention(RetentionPolicy.RUNTIME)
public @interface MQEventListener {
// 消费者组，不设置默认为当前启动的应用名${spring.application.name}，一般多实例同一个消费者需要保持一致，避免同一业务重复消费
String group() default "";
// 主题，大分类，消费线程隔离，用于区分不同业务。也可作为小分类使用
String topic() default "DEFAULT";
// 标签列表，小分类，同一主题下共享消费线程，支持：通配符*、或||、非-，支持复合表达式如：A || B -C
String tags() default "*";
}再设计一个基类MQEvent，发送与接收的MQ事件统一继承该基类：
/**
* MQ事件基类
*/
@Data
public class MQEvent implements Serializable {
// 消息ID，默认当前时间戳
protected String msgId;
// 主题，配置base-mq.default-topic后无须每次指定
protected String topic;
// 标签，只支持单个标签，多标签需要分开发送
protected String tag;
// 租户ID，默认从线程上下文获取
protected String tenantId;
// 发布MQ事件
public void publish() {
this.publish(getTopic(), getTag(), getTenantId());
}
// 发布MQ事件
public void publish(String topic) {
this.publish(topic, getTag(), getTenantId());
}
// 发布MQ事件
public void publish(String topic, String tag) {
this.publish(topic, tag, getTenantId());
}
// 发布MQ事件
public void publish(String topic, String tag, String tenantId) {
setTopic(topic);
if (getTopic() == null) {
setTopic(SpringContext.getEnv().getProperty("base-mq.default-topic", "DEFAULT"));
}
setTag(tag);
setTenantId(tenantId != null ? tenantId : ThreadContext.get(ContextConstants.TENANT_ID));
setMsgId(getMsgId() == null ? String.valueOf(System.currentTimeMillis()) : getMsgId());
// 这里的BaseContext是在基础框架全局使用的上下文，用于解耦各个组件具体的技术代码
if (BaseContext.contains("MQEventPublisher")) {
BaseContext.<String, Consumer<MQEvent>>get("MQEventPublisher").accept(this);
}
}
public <T extends MQEvent> T tenantId(String tenantId) {
this.tenantId = tenantId;
return (T) this;
}
}接下来设计一个用于抽象不同MQ实现的配置文件BaseMQProperties：
@Data
@ConfigurationProperties(prefix = "base-mq")
public class BaseMQProperties {
// 是否启用
private boolean enable = false;
// MQ实现：目前支持kafka|rocket|redis|rabbit，一个应用只用一套MQ发布和订阅
private String impl = "none";
// 服务地址
private String server = "";
// 命名空间，或作为所有主题的前缀，用于环境隔离等场景，如UAT/生产/租户环境共用一个MQ
private String namespace = "";
// 是否持久化到本地，需要定义实现了MQEventStorer的Bean
private boolean persist = false;
// 序列化器，定义实现了MQEventSerialization的Bean
private String serialization = "JsonMQEventSerialization";
// 是否自动提交
private boolean autoAck = false;
// 发送失败重试次数
private int retries = 0;
// 用户
private String username = "";
// 密码
private String password = "";
// 生产者组
private String producerGroup = "DEFAULT";
// 默认主题
private String defaultTopic = "DEFAULT";
// 交换机
private String exchange = "";
public String namespace(String concat) {
if (this.namespace != null && !this.namespace.isEmpty()) {
return namespace + concat;
}
return "";
}
}最最重要的MQ组件MQClient，设计为一个接口，并在接口内定义default方法：
/**
* MQClient接口，MQ实现类实现该接口做差异化实现
*/
public interface MQClient {
// MQ实现
String impl();
// MQ配置
default BaseMQProperties config() {
return SpringContext.getBean(BaseMQProperties.class);
}
// 整体初始化
default void init() {
if (!config().isEnable() || !Objects.equals(config().getImpl(), impl())) return;
Logger log = LoggerFactory.getLogger("### BASE-MQ : " + impl() + "Client ###");
// 初始化MQ事件发布器
if (Objects.equals(config().getImpl(), impl()) && !BaseContext.contains("MQEventPublisher")) {
log.info("Initializing MQEventPublisher");
Consumer<MQEvent> producer = initProducer();
if (producer != null) {
BaseContext.inject("MQEventPublisher", producer);
}
}
// 初始化MQ事件监听器
log.info("Initializing MQEventListener");
String[] beanNames = SpringContext.getBeanDefinitionNames();
List<MQListener> mqListeners = new ArrayList<>();
for (String name : beanNames) {
try {
// 找出所有Bean下注解了@MQEventListener的方法
Class<?> targetClass = AopUtils.getTargetClass(SpringContext.getBean(name));
Method[] declaredMethods = targetClass.getDeclaredMethods();
for (Method listenerMethod : declaredMethods) {
MQEventListener mqEventListener = AnnotationUtils.findAnnotation(listenerMethod, MQEventListener.class);
if (mqEventListener != null) {
// 消费者组，不设置默认按Spring工程的应用名隔离，当然也可以自定义
String group = mqEventListener.group();
if (group.isEmpty()) {
group = SpringContext.getEnv().getProperty("spring.application.name");
}
// 取出首个参数，用于接收MQ事件后自动解析到该类型
Class<MQEvent> eventClass = (Class<MQEvent>) listenerMethod.getParameterTypes()[0];
mqListeners.add(MQListener.builder()
.listenerMethod(listenerMethod)
.eventClass(eventClass)
.consumer(mqEvent -> {
try {
// 接收并解析为MQEvent后，具体调用该方法的位置
listenerMethod.invoke(SpringContext.getBean(listenerMethod.getDeclaringClass()), mqEvent);
} catch (Exception e) {
throw new RuntimeException(e);
}
})
.group(group)
.topic(mqEventListener.topic())
.tags(mqEventListener.tags())
.build());
}
}
} catch (Exception ignore) {
}
}
for (MQListener listener : mqListeners) {
try {
// 具体创建消费者的逻辑，不同MQ实现不一样
initConsumer(listener);
listener.setSuccess(true);
} catch (Exception e) {
log.error("Listen MQ [{}] failed!", listener.topicAndTags(), e);
}
}
log.info("Listening MQ: {}", mqListeners.stream().filter(MQListener::isSuccess).map(MQListener::topicAndTags).collect(Collectors.joining(",")));
}
// 初始化生产者
Consumer<MQEvent> initProducer();
// 初始化消费者
void initConsumer(MQListener mqListener) throws Exception;
// 启动
void start();
// 序列化器，默认取配置了的实现了MQEventSerialization的Bean
default MQEventSerialization serialization() {
return SpringContext.getBean(config().getSerialization());
}
// 消费MQ事件
default void consume(MQListener listener, MQEvent mqEvent) {
Logger log = LoggerFactory.getLogger("### BASE-MQ : " + impl() + "Client ###");
log.info("Consume MQ: {}", serialization().<String>serialize(mqEvent));
// MQ事件持久化
if (config().isPersist()) {
try {
MQEventStorer mqEventStorer = SpringContext.getBean(MQEventStorer.class);
mqEventStorer.store(mqEvent.getTopic(), mqEvent);
} catch (Exception e) {
log.error("Persist MQ failed: {}", serialization().serialize(mqEvent), e);
}
}
// 再调用方法执行业务
listener.getConsumer().accept(mqEvent);
}
}最后咱们定义好配置类，统一加载那些实现了MQClient的Bean：
/**
* 基础MQ配置
*/
@Order(PriorityOrdered.HIGHEST_PRECEDENCE + 10)
@Configuration
@ConditionalOnProperty(prefix = "base-mq", name = "enable", havingValue = "true")
public class BaseMQConfig {
@Autowired
private List<MQClient> mqClients;
@PostConstruct
public void init() {
if (mqClients != null && !mqClients.isEmpty()) {
// 获取所有实现了MQClient的MQ客户端实现类
for (MQClient mqClient : mqClients) {
// 先初始化
mqClient.init();
// 再启动
mqClient.start();
}
}
}
}核心组件中的其它类比较简单，这里先略过。
三、定义不同的MQClient实现
各个MQ之间的特性在这里就不一一介绍了，其中70%左右的代码由阿里的通义千问AI实现（推荐大家平时多用用），直接看实现代码吧，相关作用在注释里有介绍：
1.RocketMQ实现
/**
* Rocket客户端实现
*/
@Slf4j(topic = "### BASE-MQ : rocketClient ###")
public final class RocketClient implements MQClient {
private static Map<String, Supplier> LISTENERS = new ConcurrentHashMap<>();
@Override
public String impl() {
return "rocket";
}
@Override
public Consumer<MQEvent> initProducer() {
try {
// 创建 DefaultMQProducer 实例
DefaultMQProducer rocketProducer;
if (!config().getUsername().isEmpty() && !config().getPassword().isEmpty()) {
// 需要鉴权
rocketProducer = new DefaultMQProducer(config().getNamespace(), config().getProducerGroup(),
new AclClientRPCHook(new SessionCredentials(config().getUsername(),
config().getPassword())));
} else {
rocketProducer = new DefaultMQProducer(config().getNamespace(), config().getProducerGroup());
}
rocketProducer.setNamesrvAddr(config().getServer());
// 启动rocketProducer实例
rocketProducer.start();
DefaultMQProducer finalRocketProducer = rocketProducer;
return mqEvent -> {
// 定义具体的MQ事件发布逻辑
String message = serialization().serialize(mqEvent);
log.info("Publish MQ: {}", message);
String tags = mqEvent.getTag() == null ? "" : mqEvent.getTag();
try {
finalRocketProducer.send(new Message(mqEvent.getTopic(), tags, message.getBytes(RemotingHelper.DEFAULT_CHARSET)));
} catch (Exception e) {
log.error("Publish MQ: {} failed!", message, e);
}
};
} catch (MQClientException e) {
log.error("创建Rocket生产者失败", e);
}
return null;
}
@Override
public void initConsumer(MQListener mqListener) throws Exception {
DefaultMQPushConsumer defaultMQPushConsumer;
if (!config().getUsername().isEmpty() && !config().getPassword().isEmpty()) {
// 需要鉴权的情况
defaultMQPushConsumer = new DefaultMQPushConsumer(config().getNamespace(), mqListener.getGroup(),
new AclClientRPCHook(new SessionCredentials(config().getUsername(), config().getPassword())));
} else {
defaultMQPushConsumer = new DefaultMQPushConsumer(config().getNamespace(), mqListener.getGroup());
}
defaultMQPushConsumer.setNamesrvAddr(config().getServer());
defaultMQPushConsumer.subscribe(mqListener.getTopic(), mqListener.getTags());
// 创建MessageListenerConcurrently实例
defaultMQPushConsumer.registerMessageListener((MessageListenerConcurrently) (messageExts, context) -> {
// 获取方法参数
for (MessageExt messageExt : messageExts) {
MQEvent mqEvent = null;
try {
// 反序列化为MQEvent对象
mqEvent = serialization().deserialize(new String(messageExt.getBody(), StandardCharsets.UTF_8), mqListener.getEventClass());
consume(mqListener, mqEvent);
} catch (Exception e) {
log.error("Consume MQ failed: {}", mqEvent, e);
if (!(e instanceof ServiceException)) {
return ConsumeConcurrentlyStatus.RECONSUME_LATER;
}
}
}
return ConsumeConcurrentlyStatus.CONSUME_SUCCESS;
});
LISTENERS.put(mqListener.getTopic(), () -> defaultMQPushConsumer);
}
@Override
@SneakyThrows
public void start() {
if (!config().isEnable() || !Objects.equals(config().getImpl(), impl())) return;
// 在此统一启动所有消费者
for (Supplier<DefaultMQPushConsumer> listener : LISTENERS.values()) {
listener.get().start();
}
}
}2.Kafka实现
之前底层依赖使用spring-data-kafka，用了两年，回过头来发现Spring封装得太死了，后来换成了kafka-clients：
/**
* Kafka客户端实现
*/
@Slf4j(topic = "### BASE-MQ : kafkaClient ###")
public final class KafkaClient implements MQClient, DisposableBean {
private static List<Supplier> LISTENERS = new ArrayList<>();
@Override
public String impl() {
return "kafka";
}
@Override
public Consumer<MQEvent> initProducer() {
// 创建 Producer 配置
Properties props = new Properties();
props.put("bootstrap.servers", config().getServer()); // Kafka broker 地址
props.put("acks", "all");
props.put("retries", config().getRetries());// 失败重试次数
props.put("batch.size", 16384);
props.put("linger.ms", 1);
props.put("buffer.memory", 33554432);
props.put("key.serializer", "org.apache.kafka.common.serialization.StringSerializer");
props.put("value.serializer", "org.apache.kafka.common.serialization.StringSerializer");
// 创建KafkaProducer实例
Producer<String, String> producer = new KafkaProducer<>(props);
return mqEvent -> {
// 定义具体的MQ事件发布逻辑
String message = serialization().serialize(mqEvent);
log.info("Publish MQ: {}", message);
try {
// topic使用下划线的方式拼接
producer.send(new ProducerRecord<>(config().namespace("_") + mqEvent.getTopic(), message));
} catch (Exception e) {
log.error("Publish MQ: {} failed!", message, e);
}
};
}
@Override
public void initConsumer(MQListener mqListener) throws Exception {
Properties props = new Properties();
props.put("bootstrap.servers", config().getServer());
props.put("group.id", mqListener.getGroup()); // 消费者组
props.put("enable.auto.commit", config().isAutoAck());
props.put("auto.offset.reset", "earliest"); // 从最早的偏移量开始消费
props.put("key.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
props.put("value.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
KafkaConsumer<String, String> consumer = new KafkaConsumer<>(props);
consumer.subscribe(Collections.singletonList(config().namespace("_") + mqListener.getTopic()));
LISTENERS.add(() -> {
GlobalThreadPool.submit(() -> {
while (true) {
ConsumerRecords<String, String> records = consumer.poll(Duration.ofMillis(1000));
for (ConsumerRecord<String, String> record : records) {
// 处理消息
// Kafka没有标签过滤机制，所以自己实现一遍
MQEvent mqEvent = serialization().deserialize(record.value(), mqListener.getEventClass());
if (!MQFilter.matchExps(mqEvent.getTag(), mqListener.getTags())) {
continue;
}
try {
consume(mqListener, mqEvent);
if (!config().isAutoAck()) {
consumer.commitSync(); // 手动提交偏移量
}
} catch (Exception e) {
log.error("Consume MQ failed: {}", mqEvent, e);
if (e instanceof ServiceException) {
if (!config().isAutoAck()) {
consumer.commitSync(); // 手动提交偏移量
}
}
}
}
}
});
return consumer;
});
}
@Override
public void start() {
if (!config().isEnable() || !Objects.equals(config().getImpl(), impl())) return;
LISTENERS.forEach(Supplier::get);
}
@Override
public void destroy() throws Exception {
for (Supplier<KafkaConsumer> listener : LISTENERS) {
listener.get().close();
}
}
}3.Redis实现
Redis严格来说不算MQ，但因为它是一个独立于应用之外的带发布/订阅功能的中间件，且算它半个MQ吧，一般缓存服务用得多，如果要考虑减少服务器资源开销，可以考虑资源重复利用：
/**
* Redis客户端实现
*/
@Slf4j(topic = "### BASE-MQ : redisClient ###")
@Component
public final class RedisClient implements MQClient, DisposableBean {
private static JedisPool JEDIS_POOL;
private static List<Supplier> LISTENERS = new ArrayList<>();
@Override
public String impl() {
return "redis";
}
@Override
public Consumer<MQEvent> initProducer() {
return mqEvent -> {
String message = serialization().serialize(mqEvent);
log.info("Publish MQ: {}", message);
// channel=namespace:topic:tag或namespace:topic或topic
String channel = String.format("%s%s%s", (config().getNamespace().isEmpty() ? "" : config().getNamespace() + ":"),
mqEvent.getTopic(),
(mqEvent.getTag() == null || mqEvent.getTag().isEmpty()) ? "" : (":" + mqEvent.getTag()));
GlobalThreadPool.submit(() -> {
try {
jedis().publish(channel, message);
} catch (Throwable e) {
log.error("Publish MQ to [{}] failed: {}", channel, message, e);
}
});
};
}
@Override
public void initConsumer(MQListener mqListener) throws Exception {
// channel=namespace:topic:tag或namespace:topic或topic，TODO 暂不支持通配符*和非-
List<String> channels = new ArrayList<>();
if (mqListener.getTags() != null && !mqListener.getTags().isEmpty()) {
Set<String> tags = MQFilter.findIncludes(mqListener.getTags());
if (!tags.isEmpty()) {
for (String tag : tags) {
channels.add((config().getNamespace().isEmpty() ? "" : config().getNamespace() + ":")
+ mqListener.getTopic() + (":" + tag));
}
}
} else {
channels.add((config().getNamespace().isEmpty() ? "" : config().getNamespace() + ":") + mqListener.getTopic());
}
// Redis原子加锁脚本
String lockScript = "if redis.call('get', KEYS[1]) == false then redis.call('setex', KEYS[1], tonumber(ARGV[1]), KEYS[1]) return 1 else return 0 end";
LISTENERS.add(() -> {
GlobalThreadPool.submit(() -> {
try (Jedis jedis = jedis()) {
jedis.subscribe(new JedisPubSub() {
@Override
public void onMessage(String channel, String message) {
MQEvent event = null;
try {
event = serialization().deserialize(message, mqListener.getEventClass());
//按组消费：Redis本身没有消费组的概念，这里我们锁住某条消息的消费资格10秒钟，以实现同消费组（如多实例）只能由一个消费者消费
String lockKey = String.format("ChannelGroupLock:%s:%s:%s", channel, mqListener.getGroup(), event.getMsgId());
if ((long) jedis().eval(lockScript, 1, lockKey, "10000") == 1) {
//本条消息在该组未消费
consume(mqListener, event);
}
} catch (Exception e) {
log.error("Consume MQ failed: {}", event, e);
}
}
@Override
public void onSubscribe(String channel, int subscribedChannels) {
log.info("Subscribed channel: {}", channel);
}
@Override
public void onUnsubscribe(String channel, int subscribedChannels) {
log.info("Unsubscribed channel: {}", channel);
}
}, channels.toArray(new String[0]));
}
});
return null;
});
}
@Override
public void start() {
if (!config().isEnable() || !Objects.equals(config().getImpl(), impl())) return;
LISTENERS.forEach(Supplier::get);
}
@Override
public void destroy() throws Exception {
if (JEDIS_POOL != null) {
JEDIS_POOL.close();
}
}
private Jedis jedis() {
if (JEDIS_POOL == null) {
String[] hostAndPort = config().getServer().split(":");
JedisPoolConfig poolConfig = new JedisPoolConfig();
poolConfig.setMaxTotal(100);
poolConfig.setMaxIdle(50);
poolConfig.setMinIdle(10);
poolConfig.setTestOnBorrow(true);
if (!config().getPassword().isEmpty()) {
JEDIS_POOL = new JedisPool(poolConfig, hostAndPort[0], Integer.parseInt(hostAndPort[1]), 5000, config().getPassword());
} else {
JEDIS_POOL = new JedisPool(poolConfig, hostAndPort[0], Integer.parseInt(hostAndPort[1]));
}
}
return JEDIS_POOL.getResource();
}
}之前的底层依赖使用了spring-data-redis，但我发现很多工程中RedisTemplate版本冲突太多了，一气之下我换成更轻量的jedis（spring-data-redis底层也是用jedis）。
4.RabbitMQ实现
RabbitMQ也是常用的MQ之一，我们需要考虑兼容它的AMQP协议，以及标签Tags的实现，这里用了它默认的Direct交换机：
@Component
@Slf4j(topic = "### BASE-MQ : rabbitClient ###")
public final class RabbitClient implements MQClient, DisposableBean {
private static Connection CONNECTION;
private static List<Supplier> LISTENERS = new ArrayList<>();
@Override
public String impl() {
return "rabbit";
}
@Override
public Consumer<MQEvent> initProducer() {
try {
com.rabbitmq.client.Channel channel = connection().createChannel();
return mqEvent -> {
// 定义具体的MQ事件发布逻辑
String message = serialization().serialize(mqEvent);
log.info("Publish MQ: {}", message);
try {
// 路由键=namespace.topic或namespace.topic.tag
String routingKey = config().namespace(".") + mqEvent.getTopic() + (mqEvent.getTag() == null || mqEvent.getTag().isEmpty() ? "" : ("." + mqEvent.getTag()));
channel.basicPublish(config().getExchange(), routingKey, null, message.getBytes());
} catch (Exception e) {
log.error("Publish MQ: {} failed!", message, e);
}
};
} catch (IOException e) {
throw new RuntimeException(e);
}
}
@Override
public void initConsumer(MQListener mqListener) throws Exception {
// 队列名=group.namespace.topic.className.methodName
String queue = mqListener.group(".") + config().namespace(".") + mqListener.getListenerMethod().getDeclaringClass().getSimpleName() + "." + mqListener.getListenerMethod().getName();
List<String> routingKeys = new ArrayList<>();
if (mqListener.getTags() != null && !mqListener.getTags().isEmpty()) {
// 通过多个tags绑定到对应的路由键
Set<String> tags = MQFilter.findIncludes(mqListener.getTags());
if (!tags.isEmpty()) {
// 路由键=namespace.topic.tag1、namespace.topic.tag2，TODO 这里只能用或||的关系，通配符*和非-是不能生效的
for (String tag : tags) {
String routingKey = config().namespace(".") + mqListener.getTopic() + "." + tag;
routingKeys.add(routingKey);
}
}
} else {
// 路由键=namespace.topic
String routingKey = config().namespace(".") + mqListener.getTopic();
routingKeys.add(routingKey);
}
try (com.rabbitmq.client.Channel channel = connection().createChannel()) {
channel.queueDeclare(queue, true, false, false, null);
// 队列绑定到多个路由器
for (String routingKey : routingKeys) {
channel.queueBind(queue, config().getExchange(), routingKey);
}
DeliverCallback deliverCallback = (consumerTag, delivery) -> {
String message = new String(delivery.getBody(), StandardCharsets.UTF_8);
MQEvent event = serialization().deserialize(message, mqListener.getEventClass());
if (!MQFilter.matchExps(event.getTag(), mqListener.getTags())) {
return;
}
try {
consume(mqListener, event);
if (!config().isAutoAck()) {
channel.basicAck(delivery.getEnvelope().getDeliveryTag(), false);
}
} catch (Exception e) {
log.error("Consume MQ failed: {}", event, e);
if (e instanceof ServiceException) {
if (!config().isAutoAck()) {
channel.basicAck(delivery.getEnvelope().getDeliveryTag(), false);
}
}
}
};
LISTENERS.add(() -> GlobalThreadPool.submit(() -> {
try {
channel.basicConsume(queue, config().isAutoAck(), deliverCallback, consumerTag -> {
});
} catch (Exception e) {
}
return channel;
})
);
}
}
@Override
public void start() {
if (!config().isEnable() || !Objects.equals(config().getImpl(), impl())) return;
LISTENERS.forEach(Supplier::get);
}
@Override
public void destroy() throws Exception {
try {
for (Supplier<com.rabbitmq.client.Channel> listener : LISTENERS) {
listener.get().close();
}
} catch (Exception e) {
log.error("Error closing RabbitMQ channel", e);
}
try {
connection().close();
} catch (Exception e) {
log.error("Error closing RabbitMQ connection", e);
}
}
private Connection connection() {
if (CONNECTION == null) {
ConnectionFactory factory = new ConnectionFactory();
String[] hostAndPort = config().getServer().split(":");
factory.setHost(hostAndPort[0]);
factory.setPort(Integer.parseInt(hostAndPort[1]));
factory.setUsername(config().getUsername());
factory.setPassword(config().getPassword());
try {
CONNECTION = factory.newConnection();
} catch (IOException | TimeoutException e) {
throw new RuntimeException(e);
}
}
return CONNECTION;
}
}
扫一扫
1818
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
