Hi there,
I have been working on PersistentTTLNode recipe lately to fix
https://github.com/apache/curator/issues/1258. I have realized that the recipe
is leaking threads in a certain scenario.
The recipe allows to be constructed using a provided ScheduledExecutorService
or using an internally created Executor. The code tries to be generic and when
closing the recipe is only cancelling recipe touch task, BUT the Executor is
never shutdown. That means that when the Executor is internally created, the
Executor Thread will remain there forever. In my view that is leak that should
solved by distinguishing the case in which Executor is created internally or
provided externally. In the first the case the Executor MUST be shutdown when
the recipe is closed.
Let me know if you agree with analysis and I will open a new ticket.
Regards,
/Cheva
