Skip to content

Evaluate endpoint-specific POST 5xx retries for idempotent creates #71

@justin-layerv

Description

@justin-layerv

Summary

PR #69 keeps POST status-code retries conservative (429 only) because /v1/resolve can consume one-time tokens after a server-side knock failure. Now that mutating POST requests get a stable per-call Idempotency-Key, ordinary create-style POST endpoints may be safe to retry on transient 502/503/504 once the service contract and failure modes are reviewed per endpoint.

Follow-up Scope

  • Inventory POST endpoints that are replay-safe under qurl-service idempotency caching.
  • Consider a per-endpoint retry policy instead of the current global POST {429} policy.
  • Keep /v1/resolve conservative unless service behavior proves status-code 5xx replay is safe for one-time-token knock failures.
  • Add sync/async regression tests for any widened retry set.

Context

This was intentionally deferred from PR #69 to avoid broadening runtime retry behavior during a large SDK contract sync.

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions