Asynchronous tasks are everywhere ion web applications, to manage long runnig tasks or to ease failed and retry of other tasks. Since I discovered RabbitMQ, I'm using it on many project, some simple use but other with tedious setup. I also manage the server. On this page, I share my experience with these blog posts.
Posting a message in RabbitMQ is very easy, unfortunately RabbitMQ does not provide any feedback on the success or failure of the delivery in a queue. A simple typo in the routing key and the message will be lost forever without being informed. The Alternate Exchange option allows you to retrieves these "lost" messages in a specific queue.
Sometimes the processing of a message causes an error. In this case, it is common to move this message to another queue to try to process it later. All this, can be automated in RabbitMQ with the DLX / DLK option and the application of a TTL. But be careful not to create an infinite loop of attempted treatment, in this article I will introduce x-death header to control the number of retries.
RabbitMQ is fast, sometimes a bit too much compared to the rest of the infrastructure. A database replication delay can crash the processing of a message. I show you how I handle this problem and use the dead letter exchange and dead letter routing key options to delay the processing of a message.