On system with slower reactions, there could be a NullPointerException in log.
Problem happens during shutdown of backup server during activation (using SharedStoreBackupActivation)
Scenario: live/backup pair with shared store. Live is stopped and then backup is stopped before activation is complete. On slower machines, there is a possibility, that FailbackChecker is initialized even if cluster connection is cleared to null (because of shutting down). This ends with NPE during backup shutdown.
It is possible to simulate it with SharedStoreBackupTest with addition of Thread.sleep(500) into begining of SharedStoreBackupActivation.startFailbackChecker. Exception can be then seen in test output.