We do have some race conditions there. We don't see them fail in the unit tests, because our #wait are bounded. But from a performance point of view, they do occur. I've reviewed them and fix all the issue I found excepted in the AM (haven't reviewed this one, may be it's fine).
On a perf test, this seems to improve the max latency.
|Transition||Time In Source Status||Execution Times||Last Executer||Last Execution Date|
|3m 52s||1||Nicolas Liochon||23/Jun/14 14:41|
|20h 51m||1||Nicolas Liochon||24/Jun/14 11:33|
|242d 12h 1m||1||Enis Soztutar||21/Feb/15 23:34|
|Status||Resolved [ 5 ]||Closed [ 6 ]|
|Status||Patch Available [ 10002 ]||Resolved [ 5 ]|
|Hadoop Flags||Reviewed [ 10343 ]|
|Resolution||Fixed [ 1 ]|
|Status||Open [ 1 ]||Patch Available [ 10002 ]|