Problem It Solves

In many real-world scenarios, tasks may fail due to transient issues like network instability or temporary unavailability of external services. Without a retry mechanism, such failures require manual intervention. This middleware provides automatic retry logic, reducing the need for manual restarts and increasing overall system resilience.

Key Features
• ✅ Retry support with configurable maximum number of attempts (max_retries)
• ⏱️ Customizable delay between retries (delay)
• 📈 Support for exponential backoff (use_delay_exponent)
• 🎲 Optional jitter to randomize delay and reduce retry bursts (use_jitter)
• 🏷️ Label-based control at the task level:
• retry_on_error
• max_retries
• delay
• ⏳ Ability to use a custom ScheduleSource for flexible rescheduling
• ❌ Optionally replaces the task result with NoResultError until retries are exhausted (no_result_on_retry)

How It Works

When a task raises an exception:
1. The middleware checks if retries are enabled via the retry_on_error label.
2. If the retry count hasn’t reached the maximum, the task is rescheduled with a computed delay.
3. If no_result_on_retry is enabled, the task result is temporarily replaced with NoResultError, allowing for deferred handling.