Github user srdo commented on the issue:
https://github.com/apache/storm/pull/2454
@hmcl I don't think the drawback scenario can happen "normally", since the
spout never has OffsetManagers for partitions it isn't assigned except when
reactivating after deactivation. It's mainly a drawback for debugging in case
we introduce a bug where we don't clean up the OffsetManagers correctly, or
create too many OffsetManagers for some reason. Previously we'd be getting an
error when trying to commit on a partition the consumer doesn't own. Now it
will quietly skip that partition in commitOffsets, so we won't necessarily be
told if there's superfluous OffsetManagers floating around.
I think it's a fairly minor downside.
---
