Use queue attempts to govern known retries and retain exhaustion

Context

Unavailable applications and lost acknowledgement need a visible stopping condition.

Decision

Use queue receive counts for known transient failures and record exhaustion; use one redispatch before abandoning an overdue acknowledgement.

Alternatives Considered

Use a separate consumer to close dead-lettered work. A separate dead-letter consumer needs its own redrive and acknowledgement policy.

Track retry attempts in database records and a schedule. A database retry budget would duplicate the queue receive budget.

Consequences

Known exhaustion becomes staff attention, while arbitrary throws may bypass that record. Physical warehouse delays are subject to the same acknowledgement rule.