JmsOutboundGateway连接MQ超时设置及无限重试仍超时问题咨询

场景说明 #

当MQ管理器宕机时，JmsOutboundGateway会尝试连接MQ一段时间后释放线程/请求，当前存在两个核心疑问：

1. 这个连接超时时间在哪里设置或查看？是否由ConnectionFactory控制？
2. 已将FixedBackOff设置为maxAttempts=unlimited，但请求仍会触发超时，原因是什么？

错误日志 #

org.springframework.integration.jms.JmsOutboundGateway$GatewayReplyListenerContainer handleListenerSetupFailure.940  ] - Setup of JMS message listener invoker failed for destination '' - trying to recover. Cause: null
org.springframework.integration.jms.JmsOutboundGateway$GatewayReplyListenerContainer refreshConnectionUntilSuccessful.1007  ] - Could not refresh JMS Connection for destination '' - retrying using FixedBackOff{interval=5000, currentAttempts=2, maxAttempts=unlimited}. Cause: JMSWMQ0018: Failed to connect to queue manager 'LOCAL' with connection mode 'Client' and host name 'localhost(1415)'.; nested exception is com.ibm.mq.MQException: JMSCMQ0001: WebSphere MQ call failed with compcode '2' ('MQCC_FAILED') reason '2538' ('MQRC_HOST_NOT_AVAILABLE').
org.springframework.integration.jms.JmsOutboundGateway$GatewayReplyListenerContainer refreshConnectionUntilSuccessful.1007  ] - Could not refresh JMS Connection for destination '' - retrying using FixedBackOff{interval=5000, currentAttempts=3, maxAttempts=unlimited}. Cause: JMSWMQ0018: Failed to connect to queue manager 'LOCAL' with connection mode 'Client' and host name 'localhost(1415)'.; nested exception is com.ibm.mq.MQException: JMSCMQ0001: WebSphere MQ call failed with compcode '2' ('MQCC_FAILED') reason '2538' ('MQRC_HOST_NOT_AVAILABLE').
org.springframework.integration.dispatcher.UnicastingDispatcher logExceptionBeforeFailOver.185  ] - An exception was thrown by 'bean 'jmsOutbound...Gateway'; defined in: 'class path resource [/endpoints/GatewayConfig.class]'; from source: '.endpoints....jmsOutbound...Gateway(org.springframework.jms.connection.CachingConnectionFactory,org.springframework.integration.jms.JmsOutboundGateway$ReplyContainerProperties)'' while handling 'GenericMessage [payload= dummy message, headers={Persistence=0, replyChannel=org.springframework.messaging.core.GenericMessagingTemplate$TemporaryReplyChannel@20e63967, Format=MQSTR   , errorChannel=org.springframework.messaging.core.GenericMessagingTemplate$TemporaryReplyChannel@20e63967, Priority=4, messageId=[B@4569ec95, id=016d2bb2-9b03-aad8-2dd1-13cb92e6c42e, uuid=bf90a819-5430-49f8-8dd3-a23eb87e515f, timestamp=1710398153261}]': failed to handle a message in the [bean 'jmsOutboundGateway'; defined in: 'class path resource [.../endpoints/GatewayConfig.class]'; from source: 'com....configuration.endpoints.GatewayConfig.jmsOutbound...Gateway(org.springframework.jms.connection.CachingConnectionFactory,org.springframework.integration.jms.JmsOutboundGateway$ReplyContainerProperties)']; nested exception is com.ibm.msg.client.jms.DetailedIllegalStateException: JMSWMQ0018: Failed to connect to queue manager 'LOCAL' with connection mode 'Client' and host name 'localhost(1415)'.
Check the queue manager is started and if running in client mode, check there is a listener running. Please see the linked exception for more information.. Failing over to the next subscriber.

相关代码 #

消息网关定义 #

@MessagingGateway(name = "LocalGateway")
public interface LocalGateway {

    @Gateway(requestChannel = "inputLocalChannel")
    ListenableFuture<Message<String>> sendMsg(Message<String> request);

}

通道与网关配置 #

@Bean
public MessageChannel inputLocalChannel(AsyncTaskExecutor gatewayReqExecutor) {
    return new ExecutorChannel(gatewayReqExecutor);
}

@Bean
@ServiceActivator(inputChannel = "inputLocalChannel")
public JmsOutboundGateway firstHandler(){
    // 自定义配置代码
}

@Bean
@ServiceActivator(inputChannel = "inputLocalChannel")
public JmsOutboundGateway secondHandler(){
    // 自定义配置代码
}

问题解答 #

1. 连接超时的配置位置 #

这个超时不由JmsOutboundGateway本身控制，而是由底层MQ ConnectionFactory主导：

• 对于IBM MQ的MQConnectionFactory，可通过setConnectionTimeout(int timeout)方法设置单次连接MQ的超时时间（单位毫秒）。
• 如果使用Spring的CachingConnectionFactory包装MQ ConnectionFactory，超时配置需设置在被包装的原始MQ ConnectionFactory上，CachingConnectionFactory本身不处理连接超时逻辑。

你可以通过查看ConnectionFactory的初始化代码，或配置文件中对应的连接超时参数（如IBM MQ的connectionTimeout属性）确认当前设置。

2. FixedBackOff无限重试但请求仍超时的原因 #

日志中显示的FixedBackOff是GatewayReplyListenerContainer用于后台恢复连接的重试策略，和当前请求的处理生命周期无关：

• MQ宕机时，JmsOutboundGateway发送请求阶段会先尝试建立连接，这个阶段的超时由ConnectionFactory的连接超时控制，一旦单次连接超时，请求会直接抛出异常触发失败转移（日志中的Failing over to the next subscriber可佐证）。
• FixedBackOff的无限重试是监听容器在后台尝试恢复连接的逻辑，不会阻止当前请求因连接超时失败。

若需要让请求发送阶段也进行重试，需在JmsOutboundGateway上单独配置请求重试策略，比如通过setRetryTemplate(RetryTemplate retryTemplate)方法，结合FixedBackOff实现请求发送的无限重试。

内容的提问来源于stack exchange，提问作者Luke