Methods and apparatus for performing fault isolation and failure

Abstract

The invention relates to methods and apparatus for isolating faults in link-connected systems utilizing fault reports generated from within the system itself. The reports are transmitted to a central location, preferably during a predetermined time period, and are used to create a single error message identifying the probable nature and location of the fault. A preferred embodiment of the invention does not require either the construction or maintenance of systemwide configuration tables, commonly used performing fault location and analysis. Instead, each unit of a pair of link coupled units, initially or on reconnection, interrogates a link adapter at the other end of the link for an identifier that identifies both the remote unit and the remote link adapter. This 'nearest neighbor' information is stored locally at each unit, and is transmitted to the central location when an error is detected. The nearest neighbor information, rather than information from a configuration table, may be used to combine multiple records relating to a fault event, locate the fault and diagnose its cause. Additionally, a preferred embodiment of the invention provides a plurality of reporting paths for each unit in the system, so that the failure of a given unit or link does not prevent error information from being communicated to the central location.