[jira] [Created] (MESOS-623) Cluster::slaves::shutdown() doesn't call shutdown the slaves

["Vinod Kone (JIRA)" <ji...@apache.org>]

Vinod Kone created MESOS-623:
--------------------------------

             Summary: Cluster::slaves::shutdown()  doesn't call shutdown the slaves
                 Key: MESOS-623
                 URL: https://issues.apache.org/jira/browse/MESOS-623
             Project: Mesos
          Issue Type: Bug
            Reporter: Vinod Kone
             Fix For: 0.14.0


Currently, it simply terminates the slaves without calling shutdown. While the former doesn't wait for executors to shutdown, the latter waits for the executors to be terminated.

While fixing this in 'Cluster' is simple, there are quite a few tests that are going to break with this change.

--> MockExecutor's shutdown is a no-op, so the slave never gets an executorTerminated() until it times out and sends a kill via the isolator. This will increase the test duration by 5s!

--> Some tests resume the Clock after Cluster::shutdown() which means the slave never times out and sends a kill via the isolator.

--> At least one test (AllocatorTest.FrameworkExited) uses the same executor for 2 different frameworks, which breaks the isolator when killing the executors twice.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira