When the Retrospect backup server is not able to back up a client on the network, the event is stored in backup and event logs. These events can be configured to notify the backup administrator. Once the administrator is aware of the problem, a simple reboot of the networked client machine or restart of the client should restore normal functionality.
The issue itself is due to a design choice to terminate the application when exceptional conditions are encountered instead of trying to recover. While this does result in an unexpected termination of the client, it also helps protect the client from other types of attacks, such as buffer overflows, that may result in arbitrary code execution.