A single failure was observed in BasicClusteredCacheOpsReplicationAndRestartIT.testClear() where after the failover occurs the one (or more) of the keys are still found inside the cleared cache. The failure is extremely rare, with only one known occurrence and must therefore be a fairly narrow race condition, presumably precipitated by the failover.