What are the steps to reproduce your problem?
1. Create BPMN process with asyncBefore="true" on start event (see attachment)
2. Define job retry time cycle on start event (see attachment) or globally in the engine, value: R3/PT30S.
3. Let the start event be followed by a service task with always failing delegate

What is the problem?
Retry time cycle is not applied on start event - it will be retried 3 time one after another.

What would be the expected behavior:
Start event is retried in correspondence with defined retry time cycle: 3 times with the delay 30 sec each.