Browse Prior Art Database

Using Software to Buffer And Transmit Hardware Error Data

IP.com Disclosure Number: IPCOM000120323D
Original Publication Date: 1991-Apr-01
Included in the Prior Art Database: 2005-Apr-02
Document File: 3 page(s) / 95K

Publishing Venue

IBM

Related People

Daly, JC: AUTHOR [+5]

Abstract

Disclosed is a process to ensure that notification of a failure in a remote data link (such as an I/O link) is passed to a central control point by ALL affected Central Processing Complexes (CPCs) even when normal communication paths for such notification are inoperative.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Using Software to Buffer And Transmit Hardware Error Data

      Disclosed is a process to ensure that notification of a
failure in a remote data link (such as an I/O link) is passed to a
central control point by ALL affected Central Processing Complexes
(CPCs) even when normal communication paths for such notification are
inoperative.

      Certain classes of input/output (I/O) devices can be connected
to multiple CPCs that are often large distances from both the I/O
device and from each other.  When a malfunction or break occurs in
the link between this device and one or more of the CPCs, the speed
with which repair can be made depends on the rapid notification to a
central control point (rather than just the notification of the
individual affected CPCs).

      Similarly, accurate analysis of the scope and nature of the
problem often requires that this control point know of all the CPCs
that were notified of the failure and how that failure was manifested
to each one.  For example, if a malfunction occurred in a link to
four separate CPCs, accurate analysis of the malfunction at the
central control point requires that this control point receive data
about the failure from all four CPCs.

      The communication path for such notification has several
distinct parts (see Fig. 1):
      1.   the path from the link to its interface in each
           CPC
      2.   the path within a given CPC from the link
           interface to the interface to the central control
           point
      3.   the path from the CPC to the central control
           point.

      A failure in any of these paths on a given CPC would result in
the control point receiving data from only some of the affected CPCs
and this, in turn, could completely obviate the ability of the
program(s) on that control point to accurately diagnose the problem.

      This article describes a method for ensuring the delivery of
link-failure information to the central control poin...