Github user revans2 commented on the issue:
https://github.com/apache/storm/pull/2382
@srdo I agree surprising. But I was looking into it. Each tuple that is
leaked is only about 120 bytes in 1.x and about 100 in 0.10 and below, in 2.x
it is 40 bytes. So in order to leak 1 MB in 1.x it would take 9,000 tuples
that are truly lost, not just timed out but truly lost. So my guess is that it
is just such a small leak that most of the time it is not a big deal. And when
it is a big deal so much else that is bad is happening that no one debugs the
acker because they need to fix the actual cause of the missing tuples.
---
