Editorial draft available to read. Factual and language review is pending. Examples are illustrative.
Decision and method
Record the service's documented limits and distinguish restrictions by account, operation or time interval. Define priority, queuing, pausing and recovery capacity. When the provider indicates when to retry, incorporate that signal into scheduling. Avoid several workers restarting a burst simultaneously and immediately exhausting the same allowance again.
Illustrative example
Illustrative example: a historical import shares a connector with new orders. The design reserves capacity for current work and slows the import when a limit is reached. The dashboard distinguishes queued records from permanent errors so an operator does not manually restart the entire import during an expected pause.
Acceptance check
Simulate repeated limits, changing capacity and several concurrent processes. Check that queue growth is visible and older tasks are not forgotten. Accept when retries are bounded, priorities are explainable and processing recovers without duplicate effects or invented success states. Record the assumptions that determine whether accumulated work can eventually catch up.
Key takeaways
- Treat limits as shared capacity with tested queuing, priorities and recovery behavior.