zentol opened a new pull request #12021:
URL: https://github.com/apache/flink/pull/12021


   There are 2 problems in this test:
       The Job to be canceled also uses the FailingBlockingInvokable, but we 
never actually want the Invokable to throw an exception. It should just use a 
BlockingNoOpInvokable.
       The FailingBlockingInvokable uses static state to control the behavior; 
if you don't fork the JVM this can result in unexpected behavior in subsequent 
tests. I believe this is what is happening here; the failing test runs first, 
sets the blocking flag to false, then the cancel test runs and goes straight to 
throwing the exception instead of blocking.
   
   This PR is a quick fix, but long-term we should find a safe pattern for such 
stateful Invokables.
   


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


Reply via email to