Github user jose-torres commented on the issue:
https://github.com/apache/spark/pull/20622
The difference in ContinuousExecution is that the thread isn't doing any
metadata work like looking for new batches - it's either running the Spark job
or cleaning up after finishing it.
In any case, this sounds like something we aren't plausibly going to solve
here, so I'll update the PR to just resolve the flakiness in the way you
suggest.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
