Github user revans2 commented on the issue:
https://github.com/apache/storm/pull/2341
@Ethanlm I think we want to combine STORM-2753 with this into a single pull
request.
@HeartSaVioR
I think I need to give a little background on this so it makes since, and I
am not sure this is the best solution which is why I would like to see the two
combined into a single pull request.
We ran into some issues where pacemaker servers were being killed because
of a client being shot at the wrong time, or security scans, etc. The issue
was that the uncaught exception handler was not what we wanted for the
pacemaker server. We then adjusted some things and ended up with an end user
that had their topology crashing periodically because an exception that was
being thrown on a background thread, and ignored previously, was now fatal.
So the goal with these was to find a balance. We want the default uncaught
exception handler for all threads to be a little bit more lenient. But for
daemon threads we already handle exceptions very strictly so the default
UncaughtExceptionHandler is not needed here.
But thinking about it now perhaps what we want is something that is a bit
more explicit about what is happening. For example we can perhaps have a
separate default exception handler for the worker from everything else.
---
