[
https://issues.apache.org/jira/browse/SAMZA-2039?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Osaid Rehman Nasir updated SAMZA-2039:
--------------------------------------
Comment: was deleted
(was: [~cameronlee] Hey I'd like to start working on this issue. Any pointers
on where I should start with ?)
> When TaskCallbackImpl.failure is called due to timeout, interrupt is called
> on the current thread
> -------------------------------------------------------------------------------------------------
>
> Key: SAMZA-2039
> URL: https://issues.apache.org/jira/browse/SAMZA-2039
> Project: Samza
> Issue Type: Improvement
> Reporter: Cameron Lee
> Priority: Major
> Labels: newbie
>
> When TaskCallbackImpl.failure is called, the scheduledFuture in the
> TaskCallbackImpl object gets cancelled. However, in the failure case of
> hitting the task callback timeout, then it is the scheduledFuture that is
> calling TaskCallbackImpl.failure. Therefore, the scheduledFuture is calling
> cancel on itself. This may impact the execution of the thread if the
> interrupted status is ever checked. For example, StreamAppender uses a
> BlockingQueue to store messages, but that will check if the thread is
> interrupted and not handle the "append" correctly. Once a thread interrupted
> state is checked, it gets cleared, so further execution is not impacted, but
> it still causes some odd and hard-to-track behavior.
> Since we want the scheduledFuture to run in the timeout case, it should not
> cancel itself.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
