The queue hcpcs.s3.mirrorTransfer accepts messages from an external source, such as an AWS SQS queue. When a message is accepted, it becomes invisible in the external queue.
The maximum length of the hcpcs.s3.mirrorTransfer queue is 1000000 messages. If the queue overflows (that is, if this limit is exceeded), the oldest messages are rejected as failed but not removed from the external queue and become visible again for other consumers to process. A typical default value for the visibility timeout is 30 seconds.
Failed message publishing events are logged at the WARNING level. You can also monitor rejected messages using the Prometheus metric mirror_publish_failures.
If rejected messages become an issue, you have a number of options to reconfigure the external source:
- Decrease the external queue's visibility timeout period.
- Increase the external queue's retention period before a message is discarded.
- Establish an external dead-letter process for discarded messages.