Guides

Retry Strategy

Build fault-tolerant job orchestration.

When to Retry

Error TypeRetry?Notes
429 rate_limitedYesBackoff with jitter
5xx server errorYesRetry 3-5 times
4xx invalid_requestNoFix payload first
insufficient_creditsNoTop up credits

Idempotency and Deduplication

  • Assign your own external_job_id for each source asset.
  • Before create-job retry, query internal DB to avoid duplicates.
  • Store all attempts and final status for observability.

Quality Guardrails

  • Prefer representative source clips when tuning model defaults.
  • Keep input compression moderate to preserve edge details.
  • Use explicit output format policy for transparent vs non-transparent workflows.
  • Run A/B validation with the same clip set before changing defaults.

Implementation Checklist

  1. Define payload schema validation in backend before forwarding requests.
  2. Store model/output_format/background settings with each job record.
  3. Add internal quality review for difficult scenes (hair, glass, motion blur).
  4. Create runbook entries for model-specific failure cases.

Retry Budget Policy

Request TypeMax RetriesBackoff
Create/Start job3Exponential + jitter
Status pollingUnlimited until timeout windowAdaptive interval
Download/output fetch2-3Short linear or exponential