Method of preventing false or unnecessary failovers in a high availability cluster by using a quorum service

• US 6,389,551 B1
• Filed: 12/17/1998
• Issued: 05/14/2002
• Est. Priority Date: 12/17/1998
• Status: Expired due to Term

• Current Assignee: SIOS Technology Corp. (SIOS Corp.)
• Original Assignee: SIOS Technology Corp. (SIOS Corp.)
• Inventors: Yount, Nelson H.
• Primary Examiner(s): Beausoleil, Robert
• Assistant Examiner(s): Bonzo, Bryce P.
• Application Number:
  US09/213,206
• Time in Patent Office: 1,244 Days
• Field of Search: 714/1, 714/4, 714/25, 714/43, 714/55
• US Class Current: 714/4.11
• CPC Class Codes: G06F 11/2025 using centralised failover control functionality

Abstract

A system and method of providing a quorum service which each system of a cluster registers with prior to a potential failover to insure proper functionality of the cluster is provided.

Claims

1. A method of preventing false or necessary failovers due to network failures in a high availability cluster including a plurality of systems communicating via a network communication path, comprising the steps of:

providing an independent quorum service which each of said systems can communicate with independently of said network communication path;

after failure of a particular system in said cluster is suspected due to failure of a periodic heartbeat communication signal sent along said network communication path, sending a registration signal to said quorum service from other systems indicating that such other systems are operational;

initiating shutdown procedures at said particular system if said particular system is unable to send a registration signal to said quorum service; and

proceeding with failover activities by at least one of the systems other than said particular system that is unable to send a registration signal to said quorum service so as to proceed to run at least one application otherwise intended to be run on said particular system.

2. The method of claim 1 wherein the step of requesting registration status occurs after a configurable period of time.

3. The method of claim 1 wherein each of said plurality of systems includes at least one processor.

4. The method of claim 1 wherein said quorum service includes a standalone personal computer.

5. A method of preventing false or unnecessary failovers in a high availability cluster due to network failures, wherein said high availability cluster includes a plurality of systems, comprising the steps of:

providing a quorum service which each of said systems can independently communicate with;

sending a registration signal from each system indicating that the system is operational when the failure of any system in the cluster is suspected;

initiating shutdown procedures at a particular system if said particular system is unable to send a registration signal to said quorum service;

requesting registration status by one of the systems other than said particular system that is unable to send a registration signal to said quorum service;

proceeding with failover activities by at least one of the systems other than said particular system that is unable to send a registration signal to said quorum service; and

clearing all registrations from the quorum service.

6. A system for preventing false or unnecessary failovers due to network failures in a high availability cluster, comprising:

a plurality of systems running applications, each of said systems defining a respective node in said high availability cluster,

a quorum service distinct from said nodes in said high availability cluster;

means for connecting each of said plurality of systems to said quorum service such that said plurality of systems can communicate with said quorum service even in the event of said network failures;

wherein each of said plurality of systems sends a registration signal to said quorum service or initiates shut-down procedures if a particular system is unable to send a registration signal and wherein failover activities are initiated by at least one other system if said particular system is unable to send a registration signal to said quorum service.

7. A system for preventing false or unnecessary failovers due to network failures in a high availability cluster, wherein said high availability cluster includes a plurality of systems defining respective nodes thereof, comprising:

means for providing a quorum service distinct from said nodes in said high availability cluster such that each of said systems can independently communicate with said quorum service even in the event of said network failures;

means for sending a registration signal from each system indicating that the system is operational when the failure of any system in the cluster is suspected;

means for initiating shutdown procedures at a particular system if the particular system is unable to send a registration signal to said quorum service;

means for requesting registration status by one of the systems others than the particular system that is unable to send a registration signal to said quorum service; and

means for proceeding with failover activities by at least one of the systems other than the particular system that is unable to send a registration signal to said quorum service.