JPH0293856A - Trouble processing system in multiprocessor system - Google Patents

Abstract

PURPOSE:To decrease the load of respective processors and to exactly execute processing at a high speed when the generation of a trouble is detected by providing an exclusive maintenance processor to execute trouble detection and circuit processing. CONSTITUTION:A main memory 20 has a monitor information storing area 60 to house the monitor information of respective processors 1-N and monitor information updating means 70 in the plural processors 1-N update the information. A processor timer monitor means 80 of a maintenance processor 10 reads the monitor information with constant time intervals, discriminates the generation of the trouble in the processors 1-N and executes the trouble recovery processing of the processors 1-N, in which the trouble is generated, by a recovery means 105 of the maintenance processor 10 on the basis of trouble information to be collected by an information collecting means 100. Thus, the load of a whole multiprocessor system is decreased and when the trouble is detected, the collection of the fine trouble information and the trouble recovery processing can be executed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a failure handling method in a multiprocessor system, and more particularly to a processor failure handling method using a maintenance processor for handling failures.

[Conventional technology]

Conventionally, a fault handling method in this type of multiprocessor system has as external devices fault monitoring timer devices for the number of processors that the multiprocessor system has, and each processor and fault monitoring timer device are in one-to-one correspondence. In this way, a signal that each processor outputs at fixed time intervals to a corresponding failure monitoring timer device is used as a health signal indicating that the processor is normal.

When the health signal from the corresponding processor is interrupted, the fault monitoring timer device determines that an abnormality has occurred in the processor, that is, a fault has occurred, and generates an interrupt to the corresponding processor to collect fault information and recover from the fault. It was being processed.

[Problem to be solved by the invention]

In the conventional fault handling method in the multiprocessor system described above, a fault monitoring timer device is provided for each processor, so there is a drawback that fault monitoring timer devices are required for the number of processors.

Furthermore, since each processor outputs a health signal to the corresponding fault monitoring timer device, the multiprocessor system as a whole has a drawback in that the overhead of fault detection processing is large.

Furthermore, when the fault monitoring timer device detects a fault in the corresponding processor, it generates an interrupt to the faulty processor and collects the fault information, so that normal fault information is not detected. The disadvantage is that it is not always collected.

In view of the above-mentioned points, an object of the present invention is to reduce the ft load of the entire multiprocessor system by providing a dedicated maintenance processor that performs fault detection and recovery processing, and to collect detailed fault information when faults are detected. An object of the present invention is to provide a failure handling method in a multiprocessor system that can perform recovery processing.

[Means to solve the problem]

The failure handling method in the multiprocessor system of the present invention includes a main memory having a monitoring information storage area for storing monitoring information of each processor, and a monitoring information updating means for updating the monitoring information in the monitoring information storage area on the main memory. a plurality of processors including a plurality of processors; processor timer monitoring means for reading monitoring information from a monitoring information storage area on the main memory at regular intervals; A fault detection means for determining the occurrence of a fault, an information collection means for collecting fault information from monitoring information of a processor whose occurrence has been detected by the fault detection means, and a fault detection means for detecting a fault based on the fault information collected by the information collection means. and a maintenance processor that performs a recovery process for the processor that has occurred.

[Effect]

In the fault handling method in the multiprocessor system of the present invention, the main memory has a monitoring information storage area for storing monitoring information of each processor, and the monitoring information updating means for the plurality of processors is configured to have a monitoring information storage area on the main memory. The processor timer monitoring means of the maintenance processor reads the monitoring information in the monitoring information storage area on ta at regular time intervals, and the failure detection means of the maintenance processor updates the monitoring information read by the processor timer monitoring means. The occurrence of a fault in a processor is determined based on the information, the information collecting means of the maintenance processor collects the fault information from the monitoring information of the processor whose fault has been detected by the fault detecting means, and the recovery means of the maintenance processor collects the fault information using the information collecting means. Performs failure recovery processing for the failed processor based on the collected failure information.

〔Example〕

Next, the present invention will be explained in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a failure handling method in a multiprocessor system according to an embodiment of the present invention. The failure handling method in the multiprocessor system of this embodiment is based on a plurality of processors l, 2, . ..., N (N is a positive integer), and processor timer monitoring means 80. Fault detection means 90. A maintenance processor IO including an information collection means 100 and a recovery means 105, a main memory 20 including a monitoring information storage area 60, a communication control device 30, an input/output control device 40, and an external storage. The main part is composed of the device 50.

The failure information storage area 60 stores information about each processor l, 2-...
, N monitoring information is stored.

Each processor 1.2. ..., N monitoring information update means 7
0 updates the corresponding monitoring information in the monitoring information storage area 60 on the main memory 20.

The processor timer monitoring means 80 reads processors l, 2 from the monitoring information storage area 60 on the upper storage 20 at fixed time intervals.
．． . . . reads the monitoring information of N.

The failure detection means 90 detects the processors 1.2, . . . . determines whether a failure has occurred in N.

The information collecting means 100 collects fault information from the processor monitoring information when the fault detecting means 90 detects the occurrence of a fault in the processor K (K is a positive integer of l≦MSN), and controls the human output control device 40 by collecting fault information from the processor monitoring information. The data is output to the external storage device 50 via the external storage device 50.

The recovery means 105 performs failure recovery processing on the processor in which a failure has been detected based on the failure information collected by the ↑n information collection means 100.

Referring to FIG. 2, processors 1.2. . . . The process in N consists of a process execution step 110 and a monitoring information update step 120.

Referring to FIG. 3, the processing in the maintenance processor 10 includes a processor timer monitoring step 130;
It consists of a step 140 for generating monitoring information, a step 150 for detecting and determining a failure, a step 160 for collecting information about the failure, and a step 170 for recovering from the failure.

Next, the operation of the failure handling method in the multiplex [1 setter system] of this embodiment configured as described above will be explained.

Prosenona 1, 2. ..., N is each Broseno →) 12
..., step 1 when N executes the processing to be performed.
10) The monitoring information in the monitoring information storage area 60 on the main memory 20 is updated by the monitoring information updating means 70 (step 120).

- On the other hand, in the maintenance processor 10, the processor timer monitoring means 80 monitors the timer of the processor 1.2N (a timer created by software in the maintenance processor IO) (step +30),
Main memory 20”2 monitoring information record 10 area 6 at fixed time intervals
The monitoring information of each processor 1, 2°N is read from the monitoring information storage area 60 (step 140), and the monitoring information of each processor 4, 1.2. . . , N, the failure detection means 90 detects each processor l, 2 .・・・
. , N is determined to have occurred (step 150).

If the failure detection means 90 detects the occurrence of a failure in the processor, ('i't) the information collection means 100
Collect failure information from the monitoring information of the first sensor in which the failure has been detected and output it to the external storage device 50 via the input/output control device 40 (Step 160), recovery means 105
11′? Based on the fault information collected by the information collection means 100, fault recovery processing is performed on the processor (step 170). If the failure detection means 90 does not detect the occurrence of a failure, the process is immediately terminated.

〔Effect of the invention〕

As described above, the present invention is capable of reducing the load on each processor by providing a dedicated maintenance processor that performs fault detection and recovery processing, and also makes it possible to fine-tune 1'? This has the advantage of being able to collect seven reports and perform failure recovery processing.

[Brief explanation of the drawing]

Figure 7PI1 is a block diagram showing the configuration of a failure handling method in a multiprocessor system according to an embodiment of the present invention, Figure 2 is a flowchart showing the processing in the processor in Figure 1, and Figure 3 is a maintenance diagram in Figure 1. It is a flow chart showing processing in a processor. In the figure, 1.2. ..., N...Processor, 10...Maintenance processor, 20
・・・・・・・・・Main memory, 30・・・・・・Communication control device, 40・・・・・・Person output control device, 50・・・・・・
...External record jfJ device, 60...Monitoring information record jl>area, 70...Monitoring information update means, 80...Processor timer monitoring means,
90...Fault detection means, 100...Information collection means, 105.
...It is a means of recovery.

Claims (1)

[Scope of Claims] A main memory having a monitoring information storage area for storing monitoring information of each processor; and a plurality of processors including monitoring information updating means for updating the monitoring information in the monitoring information storage area on the main memory; processor timer monitoring means for reading monitoring information from a monitoring information storage area on the main memory at regular time intervals; and failure detection means for determining occurrence of a fault in the processor based on the monitoring information read by the processor timer monitoring means. an information collecting means for collecting failure information from monitoring information of a processor whose occurrence of a failure has been detected by the failure detection means; and failure recovery of the processor in which a failure has occurred based on the failure information collected by the information gathering means. 1. A failure handling method in a multiprocessor system, comprising: a maintenance processor including a recovery means for performing processing.