No failover: Just one node ID specified - Only the specified node observes the local/remote filesystem; however, the node status must be "ready". When the node isn’t ready, the file system isn’t checked at all.

Failover with node concurrency:: No node ID specified (empty input) - All Cluster nodes with the status "ready" concurrently check the local/remote filesystem according to file event listener attributes settings. In this mode, when the listener is configured to observe the local filesystem, each Cluster node observes its own local file system. So it’s useful only when the observed path is properly shared among the Cluster nodes. It may behave unpredictably otherwise. On the other hand, when the listener is configured to observe the remote filesystem, listeners running on different Cluster nodes may connect to the same remote resource. The nodes use a locking mechanism when accessing the local or remote filesystem, so no conflict between listeners running concurrently on different nodes can occur.

Failover with node reservation: More node IDs specified (separated by comma) - Just one of the specified nodes checks its filesystem. If the node fails for any reason (or its status isn’t "ready"), any other "ready" node from the list continues with checking. Please note, that when file event listener is re-initialized on another Cluster node, it compares the last directory content detected by the failed node with the current directory content.