US 6,665,811 B1Granted

Method and apparatus for checking communicative connectivity between processor units of a distributed system

AssigneeHewlett-Packard Development Company, L.P.

InventorsMarcelo Moraes de Azevedo, Robert L. Jardine

Life of this patent

Estimating the term…

Loading patent…

Abstract

An apparatus and protocol to determine the group of processors that will survive communications faults and/or timed-event failures in a distributed multiprocessor or clustered system. The processors are coupled to one another by an inter-processor communication network that is used, inter alia, by each processor to send path probe messages to each of the other processor units on all available network paths. A processor is suspected of being communicatively isolated, or having ceased operations, when one or more of the other processors detects the absence of an acknowledgment response from the processor. When this happens, all of the processors are subjected to a series of stages in which they repeatedly broadcast their status and connectivity to each other.

Claims39

Dependent claims

In a multiprocessor system having a plurality of processor units, a method for distributively agreeing on processor unit membership in said multi-processor system, the method including the steps:

communicatively coupling said plurality of processors units to form, between each of the plurality of processor units and each other of the processor units, at least one communication path;
each of the plurality of processor units periodically sending to each of the other of the plurality of processor units a path probe message on every available communication path communicatively coupling such processor units;
the other of the plurality of processors responding to receipt of the path probe message by returning an acknowledgment;
a first one of the plurality of processors detecting a communications failure by failing to receive the acknowledgment from a second one another of the plurality of processors within a predetermined period of time to initiate a Regroup operation to attempt to determine on each of said plurality of processors still operating which of said plurality of processors are still operating and still communicatively coupled;
the plurality of processors holding input/output operations substantially upon the initiation of the Regroup operation;
secondly determining on said each processor still operating whether said each processor still operating is to continue operations or to halt operations;
maintaining the hold of input/output operations on each said processor to halt operations and releasing the hold of input/output operations on each said processor to continue operations.

4 more elements

A computer system comprising:

a communications network;
a plurality of processors, communicatively connected by means of said network to provide at least two communication paths between each of the plurality of processors, each of the plurality of processors having a respective memory wherein is located a computer program for causing the computer system to tolerate inter-processor communications faults by detecting a communications failure by
(i) each one of the plurality of processors periodically sending a path probe message to each of the other ones of the plurality of processors on all communication paths;
(ii) waiting a predetermined period of time for receipt of an acknowledgment of the path probe message;
(iii) determining which processors are communicably coupled to one another if a predetermined acknowledgment to the path probe is not received back within a predetermined time.

2 more elements

A method for checking communicative connectivity between multiple processors having an interrupt priority hierarchy, comprising:

each processor:
receiving a first interrupt having a first priority;
sending an I'mAlive message to each other processor in response to the first interrupt;
determining whether I'mAlive messages have been received from each other processor within a first predetermined time period;
receiving a second interrupt having a second priority, the second priority being higher than the first priority in the interrupt priority hierarchy;
sending a path probe message to each other processor in response to the second interrupt;
responding to path probe messages received from each other processor,
determining whether a predetermined response has been received from each other processor within a second predetermined time period;
and
initiating a Regroup operation upon determining that l'mAlive messages have not been received from each other processor within the first predetermined time period or determining that the predetermined response has not been received from each other processor within the second predetermined time period.

7 more elements

Dependencies are inferred from the claim text.