Hi all,
In our current production setup, we are managing over 100,000 delayed messages with varying delay intervals ranging from 2 weeks to 6 or 7 months. Unfortunately, we have observed that certain messages, which should be processed in the future (e.g., 6 months from now), are being executed prematurely. This behavior is unexpected and inconsistent with the intended delay interval.
We would like to inquire whether the tool has any known limitations regarding the handling of a higher number of delayed messages. We came across a related issue in RabbitMQ (Delay interval predictability · Issue #72 · rabbitmq/rabbitmq-delayed-message-exchange · GitHub), which raised concerns about performance and scalability under similar circumstances.
It is worth noting that we started encountering this issue a couple of months after transitioning our infrastructure and migrating to a RabbitMQ Cluster with three nodes from a single RabbitMQ server without cluster configuration. Additionally, the number of delayed messages we are now managing has significantly increased since the migration.
Our primary question is whether this issue is known or if other users have encountered a similar problem. We would greatly appreciate any insights, information, or guidance you can provide to help us address this challenge.
Here are some additional details about our environment:
- NServiceBus version: 6.1.1
- RabbitMQ version: 3.8.14
- Erlang version: 22.2.7