dongjoon-hyun commented on PR #58054:
URL: https://github.com/apache/spark/pull/58054#issuecomment-5333643460

   Thank you for the careful trace, @viirya . Fixed in 7dcc470. The failed 
static resume path now restores the hold invariant symmetrically with 
`holdExecutors()`: alongside re-arming the guard, it pushes the zero 
requirement again and decommissions the current `getExecutorIds()` (the 
registration guard remains the backstop if that push also fails), so the state 
goes back to "requirement 0 + existing executors draining" instead of "held 
with a non-zero requirement". The shared logic is extracted into 
`zeroExecutorRequirementAndDrain()`, and, as you suggested, the test asserts 
the post-condition deterministically: it drives that method with a mocked 
`ExecutorAllocationClient` and verifies the zero-requirement push and the 
decommission of the existing executors, rather than trying to reproduce the 
sub-millisecond registration window.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to