> How did you kill the TaskManagers? I assume you didn't kill the JVM
process because otherwise you wouldn't see the finalizer objects piling up.
Till, I configure Chao Monkey to always kill the newest/same TaskManager.
So other N-1 TaskManagers stayed up during the whole process. Each of them
exp
Stephan, agree that it is not a real memory leak. I haven't found it
affecting the system. so it is sth odd for now.
but if it is not really necessary, why do we want to defer memory release
with unpredictable behavior? can StreamTask stop() method take care of the
cleanup work and don't need to r
Hi!
>From my understanding, overriding finalize() still has some use cases and
is valid if done correctly, (although PhantomReference has more control
over the cleanup process). finalize() is still used in JDK classes as well.
Whenever one overrides finalize(), the object cannot be immediately ga
Hi Steven,
the finalize method in StreamTask acts as a safety net in case the services
of the StreamTask haven't been properly shut down. In the code, however, it
looks as if the TimerService, for example, is always being stopped in the
finally block of the invoke method. Thus, it might not be nec
Hi Steven,
thanks for reporting this issue.
Looping in Till who's more familiar with the task lifecycles.
Thanks, Fabian
2017-09-12 7:08 GMT+02:00 Steven Wu :
> Hi ,
>
> I was using Chaos Monkey to test Flink's behavior against frequent killing
> of task manager nodes. I found that stopped/disp
