JPH08235009A - Automatic system degeneration detection method - Google Patents

Abstract

PURPOSE: To switch a primary system to a stand-by system even though the former system has no its down when a fault of a device is processed in an entire system and the degree of the fault exceeds a fixed level. CONSTITUTION: In an information processing system including a primary system, a stand-by system and a centralized monitoring device connected to both former systems, a fault management table is provided on the centralized monitoring device to describe the weight information 11, the system present performance deterioration value 12 and the system performance deterioration allowable value 13 which are given to the fault parts designated previously in the devices of each system as well as these fault parts. The centralized monitoring device receives the fault information from the primary system and analyzes this information to specify a fault part. Then the monitoring device reads the weight information on the fault part out of the fault management table, updates the value 12 based on the weight information to compare it with the value 13, and produces an interruption to switch the primary system to the stand-by system when the value 12 exceeds the value 13.

Description

Detailed Description of the Invention

[0001]

INDUSTRIAL APPLICABILITY The present invention is applied to a computer system which is required to process a large amount of data in an information processing system instantaneously at all times and a system having a high degree of publicity. It relates to an automatic system degradation detection method that makes it possible to maintain.

[0002]

2. Description of the Related Art In a conventional computer system, as disclosed in JP-A-6-52131,
A method of monitoring the system of another system between processors has been taken, but such a method is a method for detecting a system down, and a hardware failure (TLB, BS, etc.) that does not lead to the system down. There is a limit in detecting the effect of system degradation due to performance degradation due to a partial use prohibition state (degra) due to a failure, path blockage of an I / O device having a multipath function, etc. Since the failure detection and the blockage are performed by the failure processing of the single device, it is necessary to investigate the detailed display of the failed part and the state of the device in each device. Since it is impossible to determine how much the system required performance is affected by the failure of each device, it is not possible to switch to the standby system (duplex system) unless the system goes down, and resources are effectively used. Since the overflow of processing amount (performance) that occurs because it is not done moves to the standby system that was switched over due to system down,
There was also a problem that the processing after the system went down became large due to peak processing.

[0003]

[Patent Document 1] Japanese Patent Application Laid-Open No. 6-52131
With a fault monitoring method such as the one used in No.1 or the like, performance deterioration due to hardware failure cannot be detected, so performance deterioration is detected early and system stability is always guaranteed without impairing system reliability. I couldn't do it. An object of the present invention is to manage what kind of failure occurs in which device in the entire system, and even if the system does not go down when the degree of failure exceeds a certain level, It is to be able to switch to the standby system and to always guarantee stable system processing capacity.

[0004]

In order to achieve the above object, the present invention provides an automatic system degeneration system in an information processing system including a main processing system, a standby processing system, and a centralized monitoring device connected to these systems. The detection method, the centralized monitoring device, the prespecified failure site of each device of each processing system and weight information given to the failure site, the current system performance degradation value, and the system performance degradation allowable value. Equipped with the described fault management table,
The fault information from the main processing system is received, the fault information is analyzed to identify the device in which the fault has occurred and its fault part, the weight information given to the fault part is read from the fault management table, and the weight is calculated. When the system current performance degradation value is updated by information, the updated system current performance degradation value and the system performance degradation allowable value are compared and determined, and the updated system current performance degradation value exceeds the system performance degradation allowable value. A performance-decreasing interrupt for instructing the main processing system to switch from the main processing system to the standby processing system is performed.

[0005]

By the above means, it is possible to promptly know the performance deterioration due to the occurrence of a failure in the system, and it is possible to switch to the standby system when the performance deterioration exceeds a certain level. As a result, system performance can always be maintained in a stable state, and system reliability can be improved.

[0006]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is an overall block diagram of a system showing an embodiment of the present invention. The system includes CPU # 1 to CPU
#N and IOC # 1 to IOC # n are Ethernet L
The configuration is such that the centralized monitoring device 6 is connected via the AN 5. The centralized monitoring device 6 is a communication adapter 7 for communication.
And a failure management table 8 in which is registered information about a previously specified failure part of all connected devices.
And a system configuration management table 10 for displaying the state of the connected device and showing the state of connection with the CPU, and a fault processing program 9 for analyzing the fault information from the connected device and determining performance degradation. ing.

FIG. 2 is a diagram showing the failure management table 8 in which one setting information example is set. This table information includes a prespecified fault part of all devices connected to the centralized monitoring device 6 and weight information 11 given to the fault part for calculating the importance of each device fault. It consists of failure part information, system current performance degradation value 12 that numerically represents the failure status of the entire managed system, and system performance degradation allowance value 13 that specifies in advance how much the system allows performance degradation. . Current system performance degradation value 12
Causes a performance degradation interrupt when the system performance degradation allowable value becomes 13 or more.

FIG. 3 is a flow chart showing an outline of the setting information creation processing of the failure management table 8. The faulty part information for every device is read from a file registered in advance corresponding to all device names. Although the weight information 11 given to each failure part is set for each target device, basically, since the connected device is automatically known based on the system configuration management table 10, the weight information registered in advance is used. The standard value is set automatically. At this time, for example, when only two paths are connected to the CPU for which the four-path connection is standard, the weight information 11 for path failure of the target IOC.
Is set to double. When the standard value is not used, each information can be changed. In addition to this, the performance degradation allowable value 13 of the managed system is set as management table information. This information can be the same daily, but it is flexible and can be changed.

FIG. 4 is a flow chart showing an outline of the failure analysis processing by the failure processing program 9. When a failure occurs in any of the CPU / IOC devices that make up the system, the device that has the failure causes an Ethernet LAN
All individual device fault information is sent to the centralized monitoring device 6 via 5. When the centralized monitoring device 6 receives the failure information via the communication adapter 7, the failure processing program 9 analyzes the content of the failure to identify the device and the part where the failure has occurred, and read and read the failure weight 11. The current performance degradation value 12 of the corresponding system is updated by adding all the weights, and the system configuration management table 10 is processed to reflect that a failure has occurred in the relevant device. To update. Thereafter, the system performance degradation allowable value 13 is read, and the tolerance value 13 and the current system performance degradation value 12 of the corresponding system are read.
Performance degradation determination processing is performed to compare and determine.

FIG. 5 is a flow chart showing an outline of the performance deterioration determination process. Current system performance degradation value 1
If 2 is within the allowable value, the process is completed without performing any processing.
If the allowable value is exceeded, the centralized monitoring device 6 refers to the system configuration management table 10 and sends to the CPUs # 1 to #n of the system that have exceeded the allowable value via the Ethernet LAN 5 Performs performance degradation interrupt processing and reports that the relevant system has a performance degradation failure. Each system can promptly know the performance deterioration of the related system including itself by the failure report due to this interrupt, switch to the standby system, and switch the derived IOC and hot standby. It can be done automatically. As a result, the system can always ensure stable processing capacity.

Further, the centralized monitoring device 6 instructs the system configuration management table 10 for each device in the system having the performance deterioration fault, updates the table contents,
It is also possible to display the status of a device failure and sound an alarm when a failure occurs to notify the system administrator of the failure occurrence. Further, when the system is switched, the centralized monitoring device 6 can receive an instruction from the system about each device of the standby system that is in the operating state, update the table contents, and display each device of the switched system. is there.

[0012]

According to the present invention, since it is possible to switch to the standby system by always knowing the performance deterioration promptly, the system performance can always be maintained in a stable state and the system reliability is improved. be able to. Further, it is possible to confirm the display of the faulty part and the device state before and after the switching, and it is possible to perform the centralized monitoring of the hardware.

[Brief description of drawings]

FIG. 1 is an overall block diagram of a system showing an embodiment of the present invention.

FIG. 2 is a diagram showing a failure management table in which an example of information is set.

FIG. 3 is a flowchart showing an outline of a failure management table setting information creation process.

FIG. 4 is a flowchart showing an outline of failure analysis processing by a failure processing program.

FIG. 5 is a flowchart showing an outline of performance deterioration determination processing.

[Explanation of symbols]

 1, 2 Central processing unit (CPU) 3, 4 Input / output device (IOC) 5 Ethernet LAN 6 Centralized monitoring device 7 Communication adapter 8 Fault management table 9 Fault processing program 10 System configuration management table

Claims (1)

[Claims] 1. An automatic system degeneration detection method in an information processing system comprising a main processing system, a standby processing system, and a centralized monitoring device connected to these systems, wherein the centralized monitoring device comprises A failure management table that describes a previously designated failure part of each device of the system and weight information given to the failure part, a system current performance degradation value, and a system performance degradation allowable value, and is provided from the main processing system Received the failure information of, the failure information is analyzed to identify the device in which the failure has occurred and its failure part,
The weight information given to the failure part is read from the failure management table, the system current performance degradation value is updated by the weight information, and the updated system current performance degradation value and the system performance degradation allowable value are compared and determined. When the updated current system performance degradation value exceeds the system performance degradation allowable value, a performance degradation interrupt for instructing the main processing system to switch from the main processing system to the standby processing system is performed. Automatic system degeneration detection method.