Retry and Backoff Policy¶
Stable ID: component.retry
Kind: shared-component
Status: stable
Source: skill/references/shared-components/retry.md
Purpose¶
Centralise retry classification, bounded backoff, jitter, cancellation, and observability for transient failures.
Use when¶
- A remote or concurrent operation has documented transient failure modes.
- Several callers need the same retry budget and telemetry.
- Idempotency and cancellation semantics are understood.
Avoid when¶
- The operation is not idempotent and cannot be safely deduplicated.
- The failure is deterministic or indicates invalid input.
- Retries would amplify overload or hide a durable outage.
Applicability¶
- Languages:
any,python,q-kdbplus,cpp,rust,dart-flutter,typescript - Related entries:
example.retry-policy,pattern.strategy