Phase 5 — Retry, Dead Letter & Failure Recovery

[aaqib-hafeez-khan-in]

Parent epic: #3

Objective

Make asynchronous processing resilient to transient and permanent failures.

Scope

• Add bounded retries with backoff.
• Track retry attempts and failure reasons.
• Introduce dead-letter handling for permanently failed work.
• Make retry and recovery operations safe and idempotent.
• Provide reproducible transient-failure and permanent-failure scenarios.

Acceptance criteria

• Transient failures retry within a defined bounded policy.
• Retry attempts and failure reasons are persisted/observable.
• Permanently failed work is routed to a dead-letter path.
• Recovery does not duplicate financial effects.
• Tests cover retry exhaustion and successful recovery.

Out of scope

• Reconciliation
• Settlement
• Fraud rules

Delivery

Implement as a focused PR linked back to #3.