Skip to main content
  1. Generate and persist a stable event_id.
  2. Retry only on transient failures (429, 5xx, timeouts).
  3. Reuse the same event_id across retries.

​
Retry Backoff

  • Attempt 1: immediate
  • Attempt 2: 1s
  • Attempt 3: 2s
  • Attempt 4: 4s
  • Attempt 5: 8s
Stop retrying on 400/401/403/422 and fix payload/auth first.