Computer system for preventing inter-node fault propagation

• US 7,243,257 B2
• Filed: 05/06/2003
• Issued: 07/10/2007
• Est. Priority Date: 05/14/2002
• Status: Expired due to Fees

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Kawaguchi, Shinichi
• Primary Examiner(s): Baderman; Scott
• Assistant Examiner(s): LOHN, JOSHUA A
• Application Number:
  US10/429,682
• Publication Number:
  US 20040078649A1
• Time in Patent Office: 1,526 Days
• Field of Search: 714/9, 714/27
• US Class Current: 714/4.2
• CPC Class Codes: G06F 11/0712 in a virtual computing platform, e.g. logically partitioned systems
• CPC Class Codes: G06F 11/0724 in a multiprocessor or a multi-core unit multiprocessors per se G06F15/80
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40
• CPC Class Codes: H04L 41/0681 Configuration of triggering conditions
• CPC Class Codes: H04L 43/0811 by checking connectivity

Abstract

In a computer system in which computers each having a plurality of processors are connected with each other, said each computer comprises a system controller for, at the time of a failure within the computer system body, disconnecting own computer from other computer in which said failure has occurred, without informing own processor of such failure.

Claims

1. A computer system in which a plurality of computers, each including a plurality of processors, are connected to each other, said computer system comprising:

a system controller in each said computer for, at a time of a failure within the computer system, disconnecting its own computer from another computer in which said failure has occurred, without informing said its own processors of such failure,

wherein said system controller includes an inter-computer read failure detector, and

wherein the inter-computer read failure detector comprises;

an inter-computer read register registering an identification of an inter-computer read issued by the system controller for any of the computers;

an inter-computer read timer measuring an elapse of a pre-specified period of time from when an inter-computer read is issued;

a dummy data reply generator for, after a timeout condition upon the elapse of the pre-determined period of time, generating a pre-defined fixed value for the computer issuing an inter-computer read for use as a temporary reply to the read; and

a dummy reply timeout setting register for registering a time elapsed before the temporary reply to the read is returned.

2. The computer system as set forth in claim 1, wherein the inter-computer read failure detector further comprises:

an inter-computer reply detector detecting that a read reply data has been returned successfully from any of the computers and instructing the inter-computer read register to remove the registration; and

a selector for outputting selectively, one of the read reply data from any of the computers and the temporary reply to the read data, to the computer issuing the inter-computer read.

3. The computer system as set forth in claim 1, wherein said system controller is placed subordinate to said each computer.

4. The computer system as set forth in claim 1, wherein said each computer comprises a main memory accessible by said plurality of processors.

5. The computer system as set forth in claim 1, wherein said system controller is placed subordinate to said each computer, and wherein said each computer comprises a main memory accessible by said plurality of processors.

6. The computer system as set forth in claim 1, wherein said each computer comprises a cluster driver for transferring data one of received and transmitted between said its own processors and processors of said other computer.

7. The computer system as set forth in claim 1, wherein said system controller is placed subordinate to said each computer, said each computer comprising a cluster driver for transferring data one of received and transmitted between said its own processors and processors of said other computer.

8. The computer system as set forth in claim 1, wherein said system controller is placed subordinate to said each computer, said each computer comprising:

a main memory accessible from said plurality of processors; and

a cluster driver for transferring data one of received and transmitted between said its own processors and processors of said other computer.

9. The computer system as set forth in claim 1, wherein said system controller comprises:

means for transmitting a signal to said other computer if there is no reply from said other computer to data that said its own processors has transferred to a processor of said other computer; and

means for disconnecting said its own processors from said other computer if there is no reply to said signal within a pre-specified period of time.

10. The computer system as set forth in claim 9, wherein said system controller comprising a timer for measuring said specified period of time.

11. The computer system as set forth in claim 1, wherein said system controller is placed subordinate to said each computer, said system controller comprising:

means for transmitting a signal to said other computer if there is no reply from said other computer to data that said its own processors has transferred to a processor of said other computer; and

means for disconnecting said its own processors from said other computer if there is no reply to said signal within a pre-specified period of time.

12. The computer system as set forth in claim 1, wherein said each computer comprises:

main memory accessible by said plurality of processors; and

wherein said system controller comprises;

means for transmitting a signal to said other computer if there is no reply from said other computer to data that said its own processors has transferred to a processor of said other computer; and

means for disconnecting said its own processors from said other computer if there is no reply to said signal within a pre-specified period of time.

13. The computer system as set forth in claim 1, wherein said system controller is placed subordinate to said each computer, said each computer comprising:

a cluster driver for transferring data one of received and transmitted between said its own processors and a processor of said other computer; and

wherein said system controller further comprises;

means for transmitting a signal to said other computer if there is no reply from said other computer to data that said its own processors have transferred to a processor of said other computer; and

means for disconnecting said its own processors from said other computer if there is no reply to said signal within a pre-specified period of time.

14. The computer system as set forth in claim 1, wherein said system controller is placed subordinate to said each computer,

said each computer comprising:

a main memory accessible by said plurality of processors; and

wherein said system controller comprises;

means for transmitting a signal to said other computer if there is no reply from said other computer to data that said its own processors has transferred to a processor of said other computer; and

means for disconnecting said its own processors from said other computer if there is no reply to said signal within a pre-specified period of time.

15. The computer system as set forth in claim 1, wherein said each computer comprises:

a cluster driver for transferring data one of received and transmitted between said its own processors and a processor of said other computer; and

wherein said system controller comprises;

means for transmitting a signal to said other computer if there is no reply from said other computer to data that said its own processors has transferred to a processor of said other computer; and

means for disconnecting said its own processors from said other computer if there is no reply to said signal within a pre-specified period of time.

16. The computer system as set forth in claim 1, wherein said system controller is placed subordinate to said each computer, said each computer comprising:

a main memory accessible from said plurality of processors; and

a cluster driver for transferring data one of received and transmitted between said its own processors and a processor of said other computer, and

wherein said system controller comprises;

means for transmitting a signal to said other computer if there is no reply from said other computer to data that said its own processors have transferred to a processor of such other computer; and

means for disconnecting said its own processors from said other computer if there is no reply to said signal within a pre-specified period of time.

17. A computer system in which computers, each including a plurality of processors, are connected to each other, said computer system comprising:

in each computer, a system controller for, at a time of a failure within the computer system, disconnecting its own computer from another computer in which said failure has occurred, without informing its own processors of such failure, thereby allowing said its own processors to continue processing,

wherein said system controller includes an inter-computer read failure detection means for detecting read failures between computers, the inter-computer read failure detection means comprising;

an inter-computer read registering means for registering an identification of an inter-computer read issued by the system controller;

an inter-computer read timer means for measuring an elapse of a pre-specified period of time from when an inter-computer read is issued;

a dummy data reply generating means for, after a timeout condition upon the elapse of the pre-determined period of time, generating a pre-defined fixed value for the computer issuing an inter-computer read for use as a temporary reply to the read;

a dummy reply timeout setting register means for registering a time elapsed before the temporary reply to the read is returned;

an inter-computer reply detection means for detecting that a read reply data has been returned successfully from any of the computers and instructing the inter-computer read registering means to remove the registration; and

a selector means for outputting selectively, one of the read reply data from any of the computers and the temporary reply to the read data, to the computer issuing the inter-computer read.