RAN_EXECUTE_DUE triggered before onFailureHandler executes

[goulashify]

Hi!

Love db-scheduler, thanks for keeping it tip-top!

This is more of an FYI, not sure if this is a bug or not, but I've built my own TestScheduler, basically ManualScheduler w/ await utils, using a SchedulerListener to listen to events and noticed that when a long failureHandler is used (eg Thread.sleep(500) or so), RAN_EXECUTE_DUE is received before the failure handler completes. An ExecutionComplete event is received after the actual failure handler is executed, so:

1. schedule(taskWithLongFailureHandler)
2. await EXECUTE_DUE
3. EXECUTE_DUE triggered
   1. the task handler completed erroneously
   2. the failure handler is still in-flight
4. failure handler executes
5. ExecutionComplete triggered

Is this intended behavior?

Cheers,
– Dani

[kagkarlsson]

Good to hear 🙏

This sounds like expected behavior. EXECUTE_DUE is just the process that fetches executions that are due and sends them (async) to an ExecutorService for execution. You probably need to wait for the execution-events

[goulashify]

Thanks for the quick reply Gustav!

Ended up doing a combination of:

1. runAnyDueExecutions()
2. EXECUTE_DUE received
3. async poll until executor.getNumberInQueueOrProcessing() == 0

So that way, I can have a scheduler.runAndAwaitAnyDueExecutions() that blocks until all due is executed, incl. failure handlers (using a virtual thread executor for testing instead of the same thread executor on the ManualScheduler).