Release Notes
Announcements


Feature | 4.x Cluster | 5.x Cluster |
Retry queue granularity | Retry queues are created by group. A group only has one retry queue, and all failed messages from all topics it subscribes to are sent to the retry queue. | Retry queues are created by group and topic. A group has a dedicated retry queue for each topic it subscribes to. |
Impact of resetting classic queues | Resetting the consumer offset of a topic does not automatically reset the retry queue. | Resetting the consumer offset of a topic will automatically reset the offset of the dedicated retry queue. |
Reasons for this design | This design avoids mutual impact. If a group subscribes to Topic A and Topic B, and the shared retry queue is reset when the offset for Topic A is reset, the failed messages of Topic B are affected. | Retry queues of different topics are isolated from each other. Each topic has its own independent retry queue. When Topic A is reset, only the retry queue of Topic A is affected. The retry queue of Topic B is not affected. Therefore, the consumption process can be reset automatically. |
Additional feature | An independent feature for resetting the retry queue offset is provided. To clear all failed messages, you can perform this operation separately. | No additional features are needed because the reset queue offset is automatically reset when a topic is reset. |

피드백