When there is a sharp checkpoint, log flusher thread is stopped. When the
checkpoint is over, new log flusher is created and works. There could be a case
that the old log flusher thread may not be stopped and wait forever. This
doesn't cause any correctness issue but the resource of the thread itself is
leaked in the case.
Original issue reported on code.google.com by kiss...@gmail.com on 23 Aug 2013 at 10:09
Original issue reported on code.google.com by
kiss...@gmail.com
on 23 Aug 2013 at 10:09