Exponential backoff
ļ»æ uses exponential backoff to automatically schedule reruns with increasing time delays. The reruns trigger when a module in the outputs:
- RateLimitError
- ConnectionError
- ModuleTimeoutError
The time delays between the scenario reruns depend on whether incomplete executionsļ»æ are enabled or not:
Rerun number | Incomplete executions enabled | Incomplete executions disabled |
|---|---|---|
1 | 1 minute | 1 minute |
2 | 10 minutes | 2 minutes |
3 | 10 minutes | 5 minutes |
4 | 30 minutes | 10 minutes |
5 | 30 minutes | 1 hour |
6 | 30 minutes | 3 hours |
7 | 3 hours | 12 hours |
8 | 3 hours | 24 hours |
This means that when a module in your outputs the ConnectionError because the third-party service is unavailable, schedules the rerunning of the one minute after the error occurs.
If a module during the rerun outputs the ConnectionError again, schedules another rerun after 2 minutes and so on. If the 8th attempt fails, disables scheduling of the .
Until disables scheduling the , the follows the run schedule in addition to the exponential backoff reruns (though there are exceptionsļ»æ).
If you want to wait with scheduling the next runs until the reruns finish, enable sequential processingļ»æ in settings.
If you enable the option to store incomplete executions in the settings, stores the run in incomplete executions.
ļ»æ then schedules reruns from the incomplete execution data. If all rerun attempts fail, keeps the incomplete execution for you to resolve manually.