客户端使用 STOMP 版本 1.2 而不是 RabbitMQ 版本 3.9.3。
客户端用于聊天应用程序作为传入消息的订阅者。
a["CONNECTED\nserver:RabbitMQ/3.9.3\nsession:session-paGrRWWiEsYECaWdVdqkXQ\nheart-beat:4000,4000\nversion:1.2\nuser-name:6208370595c29c4357f9b81c\n\n\u0000"]
订阅的 ack 标头是client-individual。
客户端能够从 RabbitMQ 中确认消息而不会出现任何异常。
问题是，当客户端不确认时，消息不会重新传递给客户端。
如果订阅者没有确认消息，我希望将消息重新传递给订阅者。
是否可以使用 STOMP 和 RabbitMQ 内置解决方案创建这样的功能，而不进行任何自定义确认？
我尝试过的另一种更糟糕的解决方案是，我在rabbitmq.conf中将consumer_timeout设置为10000，这应该断开10秒内没有确认消息的订阅者，然后客户端将重新连接到客户端并获取丢失的消息。
该解决方案的问题在于，在未发送 ack 后断开订阅者需要 1 分钟，而不是 10 秒。
The client uses STOMP version 1.2 over RabbitMQ version 3.9.3.
The client is used for a chat app as a subscriber for incoming messages.
a["CONNECTED\nserver:RabbitMQ/3.9.3\nsession:session-paGrRWWiEsYECaWdVdqkXQ\nheart-beat:4000,4000\nversion:1.2\nuser-name:6208370595c29c4357f9b81c\n\n\u0000"]
The ack header on subscription is client-individual.
The client is able to acknowledge a message without any exceptions from RabbitMQ.
The problem is, when the client doesn't acknowledge, the message isn't redelivered to the client.
I want to have the message redelivered to the subscriber if the subscriber didn't acknowledged the message.
Is it possible to create such a feature using STOMP and RabbitMQ built-in solutions, without doing any custom acknowledges?
The alternative worse solution I tried was that I set up consumer_timeout to 10000 in rabbitmq.conf, which should disconnect a subscriber which doesn't ack a message for 10s, then the client would reconnect to the client and fetch the missed messaged.
The problem with this solution is that it takes 1 minute, not 10 seconds to disconnect a subscriber after not sending ack.
发布评论
评论（2）
nack 消息的处理方式由消息代理决定，在您的例子中是 RabbitMq。它们可以被丢弃、发送到死信队列或重新排队。
这些文档将为您提供帮助：https://www.rabbitmq.com/nack.html
为了使用重新排队机制，您必须在批量情况下使用
basic.reject或basic.nack。如何设置取决于您使用的客户端库 api。
另请参阅：https://www.rabbitmq.com/confirms.html#消费者 nacks 请求
How nack messages are handled is decided by the message broker, in your case RabbitMq. They can either be dropped, sent to a dead letter queue or re-enqueued.
The docs will help you out: https://www.rabbitmq.com/nack.html
In order to use requeuing mechanism you have to use
basic.rejectorbasic.nackin case of bulk.How you set it depends on the client library api you use.
Also refer to: https://www.rabbitmq.com/confirms.html#consumer-nacks-requeue
嗯...关于使用rabbitmq 时选择哪种配置有很多讨论。我将推荐以下设置（首先）
使用
fanout模型（如 文档）。这将允许您：简而言之：是的，有一个适合您需求的配置
深入：您应该仔细设计您的设置以满足您的所有需求（包括高可用性），以确保您的代码和部署能够随着项目的进展而保持不变
Well...there's lots to discuss about which configuration to choose when using
rabbitmq. I will recommend the following setup (to begin with)Use
fanoutmodel (as described in the docs). This will allow you to:In short: yes, there is a configuration which suits your needs
In depth: you should carefully craft your setup to suit all your needs (including high availability) in order to ensure that your code and deployment will hold as your project progress