A distributed system may have a number of separate interfaces called ports and in testing it may be necessary to have a separate tester at each port. This introduces a number of issues, including the necessity to use synchronised test sequences and the possibility that output-shifting faults go undetected. This paper considers the problem of generating a minimal synchronised test sequence that detects output-shifting faults when the system is specified using a finite state machine with multiple ports. The set of synchronised test sequences that detect output-shifting faults is represented by a directed graph G and test generation involves finding appropriate tours of G. This approach is illustrated using the test criterion that the test sequence contains a test segment for each transition.