Details
-
Bug
-
Status: Closed
-
Minor
-
Resolution: Not A Problem
-
None
-
None
-
None
Description
When running the test repeatedly on the same cluster one can sometimes see unexpected reference count, where the number found is (in the observed case) a multiple of the number expected, so instead ok 2.5m nodes it finds 12.5m, for example. It looks like it's reading the data from the old run. Setup should delete that (not cleanup, as the data may be used for debugging after the test)