Github user florianschmidt1994 commented on the issue:
https://github.com/apache/flink/pull/6216
Thanks @zentol, I have one remark (see above), besides that looks good to
me!
Additionally I had some ideas that came to mind that I think we could
discuss:
- We have a common pattern of `wait_for_sth` functions, that either
- get stuck in a loop for ever if the desired event doesn't happen (I
think `wait_for_job_state_transition` also behaves like that, right?)
- or iterate a fixed number of times and then continue execution, whereas
instead they should fail.
I think we should add an issue for that to refactor that over all the tests
to have consistent and useful behaviour
- Also I think that we could have the backup config and revert config as
part of the test runner and always do that, so we avoid running into a
corrupted flink-dist if tests don't behave correctly?
What do you think!
---
