Space retries so failure can recover#
1s → 2s → 4s → 8s → … + jitterImmediate retries amplify a slow dependency and can reorder or duplicate work. Bound attempts, add randomized delay, preserve attempt metadata, and move permanently failing messages to an owned remediation path.
| Often retryable | Usually requires correction |
|---|---|
| Timeout, temporary unavailable, rate limit | Malformed payload, invalid schema |
| Transient network failure | Permanent missing resource or denied authorization |
Classification is contextual. Cross-link to Retries, Exponential Backoff and Jitter for deeper budgets and storm prevention.
Related practice problems
Use the manifest-backed prompts below to apply this lesson in a complete design.
- Amazon SQS visibility timeoutAWS Documentation