Generic fault tolerant platform

• US 5,978,933 A
• Filed: 01/11/1996
• Issued: 11/02/1999
• Est. Priority Date: 01/11/1996
• Status: Expired due to Term

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Company (HP Inc.)
• Inventors: Rosset, Jean-Michel, Pierrot, Denis, Wyld, Brian Christopher, Rondel, Pierre D. M., Sauvage, Pierre
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): ELISCA, PIERRE E
• Application Number:
  US08/584,830
• Time in Patent Office: 1,391 Days
• Field of Search: 395/182.08, 395/182.09, 395/182.1, 395/182.11, 395/181, 395/185.08, 395/182.04, 395/182.05, 395/291, 395/296, 395/728, 395/182.02, 711/147, 711/151
• US Class Current: 714/13
• CPC Class Codes: G06F 11/1482 by means of middleware or OS functionality
• CPC Class Codes: G06F 11/1492 by run-time replication performed by the application software
• CPC Class Codes: G06F 11/2023 Failover techniques
• CPC Class Codes: G06F 11/2033 switching over of hardware resources
• CPC Class Codes: G06F 11/2035 without idle spare hardware
• CPC Class Codes: G06F 11/2097 maintaining the standby controller/processing unit updated initialisation or re-synchronisation thereof G06F11/1658 and subgroups

Abstract

A fault tolerant platform is provided that comprises two systems running pairs of processes in the active and standby state, one process from each pair running on each system. Each system comprises a fault-tolerance controlling process, first communication channels provided between the fault-tolerance controlling process and the processes in the active or standby state running on its system, and second communication channels provided between the fault-tolerance controlling processes. Management of fault tolerance (that is, promoting a process in a standby state to the active state, and making a process in an active state exit from the active state) is handled by the fault-tolerance controlling processes. A generic management of fault tolerant processes is thus provided in which fault detection and switchover is carried out independently of the applications. The invention thus ensures efficient and coherent switchover between active and standby processes.

Claims

1. A method for preventing split brain syndrome in a fault tolerant platform having at least one process in an active state of execution and at least one process in a standby state, switchover capabilities for promoting a process in the standby state to an active state, and commonly available resources dedicated to the processes in the active and standby states, said method comprising the steps of:

limiting simultaneous access to the commonly available resources to a maximum number of processes in the active state;

giving priority of access to the commonly available resources to processes last promoted to the active state; and

terminating an active state of a process that is in an active state of healthy execution when said process in an active state of healthy execution is not allowed access to the commonly available resources.

2. A method according to claim 1, wherein said fault tolerant system has one process in the active state and one process in the standby state, and wherein access to the commonly available resources is limited to one process in the active state.

3. A method according to claim 1, comprising the further step of:

causing an active process that leaves its active state to inform other processes in a standby or active state.

4. A fault tolerant platform comprising:

at least two systems;

multiple processes in an active state running on said systems;

each process in an active state having at least one corresponding replicate process in a standby state concurrently running on a different one of said systems to the system running said process in an active state;

fault-tolerance manager means for monitoring said processes and responsive to an apparent fault in one said process for individually promoting a corresponding replicate process in a standby state to the active state; and

first communication channels between said processes in an active state and said fault-tolerance manager means and between said processes in a standby state and said fault-tolerance manager means;

the fault-tolerance manager means comprising a respective fault-tolerance controlling process running on each system, and second communication channels between said fault tolerance controlling processes for exchanging health monitoring and status messages, said first communication channels comprising a communication channel between a process in an active or standby state running on a system and the fault-tolerance controlling process running on a same said system.

5. A fault tolerant platform according to claim 4, further comprising a third communication channel between a process in an active state and a corresponding replicate process in the standby state, for exchanging health monitoring and replication data messages.

6. A fault tolerant platform according to claim 4, further comprising at least a user process, and fourth communication channels between said user process and said process in the active state, and between said user process and said process in the standby-state.

7. The fault tolerant platform as recited in claim 4, further comprising:

commonly available resources available to said process in the active state and said corresponding replicate process in the standby state; and

wherein the fault-tolerance manager means controls exit of said process in an active state to said standby state, in dependence upon sensing that said commonly available resource is being seized by said peer process.

8. The fault tolerant platform as recited in claim 4, further comprising:

commonly available resources available to said process in the active state and said corresponding replicate process in the standby state; and

wherein the fault-tolerance manager means controls exit of said first process in an active state to said standby state, in dependence upon sensing that said first process in an active state is not allowed to access to said commonly available resources.

9. The fault tolerant platform according to claim 8, wherein said fault tolerant platform has one process in the active state and one process in the standby state, and wherein said means for limiting limits access to the commonly available resources to one process in the active state.

11. An apparatus for managing fault tolerance according to claim 8, wherein a fault-tolerance controlling process running on a system informs each fault-tolerance controlling process running on another system of a state of processes running on its system, on said second communication channels.

12. An apparatus for managing fault tolerance according to claim 8, wherein a process running on one system informs the fault-tolerance controlling process running said one system of its state, on said first communication channels.

13. An apparatus for managing fault tolerance according to claim 8, wherein said platform comprises third communication channels between said first process in an active state running on one system and said corresponding replicate process in a standby state running on another system, said apparatus further comprising means for mutual health monitoring of said processes in an active and standby state on said third communication channels.

14. An apparatus for managing fault tolerance according to claim 8, wherein said platform comprises third communication channels between said first process in an active state running on one system and said corresponding replicate process in a standby state running on another system, and wherein a process running on a system informs its corresponding replicate process of its state using said third communication channels.

15. An apparatus for managing fault tolerance according to claim 14, wherein a process running on a system informs the fault-tolerance controlling process running on its system of the state of its corresponding replicate process using said first communication channels.

16. An apparatus for managing fault tolerance according to claim 8, wherein said platform comprises third communication channels between said first process in an active state running on one system and said corresponding replicate process in a standby state running on another system, and wherein a process running on a system exchanges replication data messages with its corresponding replicate process, using said third communication channels.

17. An apparatus for managing fault tolerance according to claim 8, wherein said platform further comprises at least a user process, and fourth communication channels between said user process and said first process in the active state, and between said user process and said corresponding replicate process in the standby state, and wherein said first process in the active state, and said corresponding replicate process in the standby state are grouped under a class name, and wherein said user process sends messages to the class name, said messages being directed to the process in the active state transparently for the user process.

18. An apparatus for managing fault tolerance according to claim 8, wherein said platform further comprises at least a user process, and fourth communication channels between said user process and said first process in the active state, and between said user process and said corresponding replicate process in the standby state, and wherein replication messages are sent by the user process to the process in the standby state.

10. An apparatus for managing fault tolerance in a platform comprising:

at least two systems;

at least a first process in an active state running on one system;

at least one corresponding replicate process in a standby state running on another system;

fault-tolerance controlling means for monitoring said processes and responsive to an apparent fault in one said process for individually promoting a corresponding replicate process in said standby state to the active state;

first communication channels comprising a communication channel between a process in an active or standby state running on a system and a fault-tolerance controlling process running on a same said system; and

wherein the fault-tolerance controlling means comprises a respective fault-tolerance controlling process running on each system and second communication channels between fault-tolerance controlling processes, and wherein the fault-tolerance controlling process running on each said system is responsible for promoting processes running on that system to the active state, and for making processes running on that system exit from the active state.

19. The apparatus as recited in claim 10, further comprising:

a commonly available resource available to said first process in the active state and said corresponding replicate process in the standby state; and

wherein the fault-tolerance controlling means controls exit of said first process in an active state to said standby state, in dependence upon sensing that said commonly available resource is being seized by said corresponding replicate process.

20. The apparatus as recited in claim 10, further comprising:

commonly available resources available to said first process in the active state and said corresponding replicate process in the standby state; and

wherein the fault-tolerance controlling means controls exit of said first process in an active state to said standby state, in dependence upon sensing that said first process in an active state is not allowed to obtain access to said commonly available resources.

21. A fault tolerant platform having at least one process in an active state of execution and at least one process in a standby state, switchover capabilities for promoting a process in the standby state to an active state, and commonly available resources dedicated to the processes in the active and standby states, and means for preventing split brain syndrome, said means comprising:

a) means for limiting simultaneous access to the commonly available resources to a maximum number of processes in the active state;

b) means for giving priority of access to the commonly available resources to processes last promoted to the active state; and

c) means for terminating an active state of a process that is in an active state of healthy execution when said process in an active state of healthy execution is not allowed access to the commonly available resources.

22. The fault tolerant platform according to claim 21, further comprising:

d) means for causing an active process that leaves its active state to inform other processes in a standby or active state.