JP2002366535A - Maintenance system of parallel computers - Google Patents

Abstract

PROBLEM TO BE SOLVED: To dissolve temporal restriction of performance of a replacing work by performing incorporation of a logical package in a network by decision of software itself. SOLUTION: This maintenance system of parallel computers is provided with a means 100 to set completion of replacement of a faulty logical package, a CPU interface control part 101 to report the completion of replacement of the faulty logical package to the software 200 and a structure control part 103 to set connecting relation between the logical package and the network in the network according to an instruction from the software.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a parallel computer system in which the operation status of software is determined by the software itself, and a logic package can be incorporated into a network.

[0002]

2. Description of the Related Art In the conventional method, when a failed logical package is replaced, the operation status of software is collated on a console or the like, and it is confirmed that the logical package can be incorporated into a network. The input device executes the process of incorporating the logical package into the network.

[0003]

In the above-mentioned prior art, the replacement of the faulty logic package and the operation of incorporating the replaced logic package into the network are performed in a series, so that the time for performing the replacement work is greatly restricted.

[0004] It is an object of the present invention to eliminate the time constraint of the exchange work by incorporating a logical package into a network at the discretion of the software itself.

[0005]

FIG. 1 is a block diagram showing the present invention.

Means 100 for setting completion of replacement of a failed logical package, and CPU interface controller 10 for reporting completion of replacement of a failed logical package to software 200
1, a configuration control unit 102 for setting a connection relationship between a logical package and a network to a network according to an instruction from software, a unit 201 for holding a failure state of a logical package, and a unit 103 for holding a connection state of a logical package and a network. Have.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The completion of replacement of a failed logical package is set by the logical package replacement work control unit 100. Replaced failed logical package number with CP
Software 200 through U interface control unit 101
Report to The software stores the replacement completion logical package number in the failure part information 201. When the software determines that the network configuration can be changed,
The configuration control unit 102 is instructed to incorporate the replacement completion logical package held in the failure site information 201. The configuration control unit reconfigures the network configuration information 103 and sets the connection of the exchange completion logical package to the connection information 310 to 313.

An embodiment of the present invention will be described with reference to FIG.

For example, assume that the logic package 210 has failed. When the replacement of the failed logical package 210 is completed, the number of the replaced logical package 210 is input to the logical package replacement work control unit 100 (step S1).

The input logical package number 210 is
The information is reported to the interface control unit 101, and is reported to the software 200 from the CPU interface control unit 101 (step S2).

The software 200 changes the failure status of the logical package 210 in the failure site information 201 to “replacement completed” (step S12).

The software 200 determines whether the configuration change of the computer system may be executed when the job being executed is completed (step S13).

When it is determined that the configuration can be changed,
The failure state of the logical package in the failure part information 201 is checked, and if there is a logical package in the “replacement completed” state, the connection state of the logical package in the failure part information 201 is changed to “connected” (step S14). ), And instructs the configuration control unit 102 to incorporate the system via the CPU interface control unit 101 (step S15).

The configuration control unit 102 reconfigures the network configuration information 103 in which the logical package 210 is incorporated (step S4), and the logical package 2 in the network 300
The ten connection information 310 is changed from "disconnection" to "connection" (step S5).

[0015]

As described above, according to the present invention, the replacement operation of the faulty logic package can be completed irrespective of the operating state of the software for controlling the computer system. Further, at the time when the operation state becomes convenient for software control, the replaced logical package can be incorporated into the network.

[0016] These mechanisms enable parallel maintenance of a faulty logic package with less time constraints.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing the whole of the present invention.

FIG. 2 is a configuration diagram of information holding a connection state between a network and a logical package.

FIG. 3 is a configuration diagram of information holding a failure state of a logical package.

FIG. 4 is a flowchart for implementing the present invention.

[Explanation of symbols]

100: Logical package exchange work control unit, 101: CPU interface control unit, 102: Configuration control unit, 103: Network configuration information, 200: Software, 201: Fault location information, 21
0-213 ... Logical package, 300 ... Network, 310-31
3… Network and logical package connection information, 500,600
… Logical package number, 501,601… Logical package number 5
00: connection state, 602: failure state of logical package number 500, S1, S2: number of procedure (step).

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Teruyuki Harada 1st Horiyamashita, Hadano-shi, Kanagawa Prefecture Inside Nichi Information Technology Co., Ltd. (72) Inventor Akihiro Ueda 5-22-1, Kamizuhoncho, Kodaira-shi, Tokyo F-term (reference) 5B045 GG11 JJ42 in Hitachi Super LSI Systems Co., Ltd.

Claims (1)

[Claims] In a parallel computer system in which a plurality of logical packages connected by a network are mounted, software for controlling the computer system, means for retaining failure information of the logical package, and replacement of the failed logical package are completed. Means for setting the connection of the logical package, means for notifying the software that the replacement of the failed logical package has been completed, means for maintaining the connection state between the network and the logical package, and control of the connection information between the logical packages on the network. A maintenance method for a parallel computer system, which has a means for performing the above operation, and enables the incorporation of a logical package, whose software has been arbitrarily exchanged, into a network according to the operation state of the software.