Github user pnowojski commented on the issue:
https://github.com/apache/flink/pull/5718
I think that timeouts in IDE are only making things worse:
1. you can not debug, since longer debugging session will fail
2. in case of deadlocks, locally you can always manually dump/view stack
traces but you also can do full thread/memory dump and/or attach a debugger.
On travis you can not debug it in any way, thus automated actions make
sense (and they preserve some travis resources for other builds).
Furthermore, such rule would have to be implemented in all of our tests,
not only here. Also IMO if you want to replace/remove `travis_mvn_watchdog`
with such rule, this is out of scope of this ticket and it should be discussed
separately (I would be against it), so that it doesn't hinder this ticket.
When I originally created those tests (with `@Test(timeout = xxxx)`) I did
it by copying without thinking from a test, that was printing something to logs
once every checkpoint, thus it was interrupting travis watchdog and one could
argue that it required this special case timeout. This commit makes it more
consistent with rest of the tests.
---
