JPH07121395A - Method for preferentially selecting auxiliary device - Google Patents

Abstract

PURPOSE:To shorten the recovery time of a system by avoiding the reload of a software by preferentially selecting the same auxiliary device as that of a fault processor for a software previously loaded to a memory. CONSTITUTION:When any fault is generated at a processor unit 1 under operating, first of all, a switch 41 for connection is opened and next, any auxiliary processor is selected. In this case, the same software as the software class of the fault detected processor unit 1 loaded, any processor unit (such as a processor unit N, for example,) under auxiliary waiting is selected, a switch 4N for connection is closed, and the processor unit N is connected to a bus line 11 for processor link. Thus, the operation can be immediately restarted only by transferring peculiar data to each processor unit without the load processing of the software.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system recovery function when a processor device fails in a computer system,
In particular, the present invention relates to a spare processor switching method in a multiprocessor configuration.

[0002]

2. Description of the Related Art Conventionally, in a function-distributed computer system (multiprocessor system) as shown in FIG. 2, when a processor device fails, the preset spare processor device is loaded with the software to switch the control. The distributed function multiprocessor configuration is
The functions realized by the entire system are divided into several functional units, and the divided functional units are shared by individual processor devices, and the shared functions are realized by software loaded into the memory of each processor. It is a thing. In FIG. 2, each processor device 1 to N (N:
Each of the integers of 1 or more) is composed of a set of a processor (CPU) and a memory (MM), and a plurality of these processor devices are connected by a bus line 11 connecting the processor devices. Further, one processor control device 21 is used for operation management of a plurality of processor devices.
To provide. A file storage device 31 for storing software for running on each processor device is connected to the processor control device 21 for operation management. The software is read from the file storage device as necessary and is transferred to the software. By transferring the data to each processor device via the communication line 61, software load processing is realized in the memory of each processor device. Further, connecting switches 41 to 4N are provided between the processor devices 1 to N and the bus line 11, and the processor device 21 for operation management is switched from the processor device switching control line 51.
By connecting / disconnecting the connection switches 41 to 4N under the control via, the connection / disconnection of each processor unit to / from the bus line can be performed. In such a system,
These loosely-coupled multiprocessor devices, which are provided in plurality, stop their functions due to failure of any processor device or abnormality of software running on those processor devices, and as a result, a part of the processor system. Or all the functions may stop. As a recovery means therefor, in a conventional multiprocessor system, an optional redundant number of processors is provided as a spare device, and in the unlikely event that an arbitrary operating processor device fails for some reason, the processor control device In accordance with the following procedure, the function of automatically switching back the function stopped by the failure is realized by switching the spare processor apparatus as a substitute for the failed processor apparatus according to the following procedure. The procedure for restoration is shown below. "Procedure 1" (1) Detach the faulty device from the bus line (2) Select any spare processor unit (3) Connect the spare processor unit to the bus line (4) Install software to the memory in the spare processor unit Transfer (5) Transfer of data unique to each processor to the memory in the spare processor unit (6) Activation of software of the spare processor unit The operating states of the processor unit in operation and the processor unit in standby standby are N processor units. Of the (total number of processor devices: N), M units are operating processor devices and the remaining J units are spare processor devices. In this case, M, J, and N are positive integers, and N = M + J is satisfied. Here, consider a case where the processor devices 1 to M in FIG. 2 are operating devices and M + 1 to N are standby standby devices. As a specific example, it is assumed that M = 5 and J = 3, the operating processor units are the processor units 1 to 5, and the standby standby processor units are the processor units 6 to 8 as shown in FIG. Yes (5 devices in operation, 3 devices in standby). Further, the software loaded on the memory of each processor device is not a single software, but as described above,
There are multiple types to realize a multiprocessor configuration by function distribution. In the configuration of FIG. 3, it is assumed that three types of software are loaded as in the processor management table of Table 1.

[Table 1] Now, assuming that the processor device 1 has failed, when the processor device management control device 21 detects the failure, the control device follows the above “procedure 1”, specifically, the recovery of the following “procedure 2”. Perform processing. “Procedure 2” (1) The connection switch 41 is opened via the processor device switching control line to disconnect the processor device 1 from the inter-processor coupling bus line 11. (2) Select any one device from the spare processor devices (processor devices 6 to 8). Now, it is assumed here that the processor device 6 having the lowest number is selected. (3) The connection switch 46 is closed and the spare processor device 6 is connected to the bus line 11. (4) Referring to Table 1, select the same program (type A) as that used in the faulty processor device 1. (5) The program (type A) is read from the file storage device 31, and the software transfer line 61 and the bus line 1 are read.
1 to the processor unit 6 via. (6) Data unique to each processor is transferred to the memory of the processor device 6. (7) The software for the processor device 6 is activated to replace the function provided by the processor device 1 before the failure of type A. (8) Table 1 is rewritten according to the switching, and the contents of Table 2 are used as the processor management table.

[Table 2]

[0003]

In the above prior art, when a failure occurs in an arbitrary processor device, a load processing of the same software as that running in the failed device is performed on the spare processor device. Since this was done, the software load time was included in the interruption time until the operation was restarted (restored) in the standby device, which was a cause of delaying the overall recovery time. An object of the present invention is to improve such problems and restore the software loading process (the process (4) of the “procedure 1” and the process (5) of the “procedure 2”) to the spare device from the occurrence of the device failure. It is an object of the present invention to provide a method for selecting a backup device preferentially, which can be excluded from the time until, and shortens the overall recovery time.

[0004]

In order to achieve the above-mentioned object, a method for selecting a spare device priority according to the present invention comprises a plurality of processor devices whose operating specifications are variable depending on the type of software loaded on a memory and a spare spare device. In a loosely coupled multiprocessor computer system equipped with a processor unit, when a spare processor unit is incorporated upon the failure of an arbitrary processor unit, the type of software loaded on the memory of the failed processor unit is recognized ( 1), if there is a spare processor device in which the same type of software is preloaded on the memory (103 in FIG. 1), that spare processor device is preferentially selected (104 in FIG. 1). ) Is characteristic.

[0005]

According to the present invention, the same number or more of redundant spare processor units as the type of operating software are provided, and the same type of software as each operating software is loaded in advance into the memory of the spare processor unit during standby. . The operating status and software type of each processor device are registered and managed in the processor device management status table. When a failure occurs, the management status table is used to check whether or not there is a spare processor unit already loaded with the same software type as the failed processor, and if so, the spare processor unit is preferentially selected. Have logic. As a result, if there is a spare processor device that has been loaded with the software, it is not necessary to load the software to the spare processor device, which is required for each process from the occurrence of a failure in any processor device to recovery in the conventional method. Becomes Therefore, the software load processing time for the spare device is not included in the time from the occurrence of the device failure to the recovery, and the recovery time from the failure occurrence to the recovery can be shortened.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. 4 and 5 are configuration diagrams showing an example of a multiprocessor system to which the backup device priority selection method of the present invention is applied. The backup device priority selection method of the present invention can be applied not only to the configuration shown in FIG. 2 but also to the computer system having the configurations shown in FIGS. 4 and 5. Figure 4
2 is a form in which the software file transfer line 61 of FIG. 2 is directly connected to the memory control unit of each processor device without the processor connection bus line 11 interposed therebetween. FIG. (11, 12 in FIG. 5),
Moreover, the connection form of the switch is such that it is connected to either one of the processor coupling buses. In these systems, the bus line for file transfer is separated from the transmission path used for inter-processor communication between operating processor units and the path for software transfer, thereby lowering the bus transfer capability due to the file transfer process, and of the failed processor. The purpose is to avoid signal disturbance on the bus line due to abnormal operation.

Next, for comparison with the prior art, the backup device priority selection method of this embodiment will be described based on the configuration shown in FIG. 6 and 7 are diagrams showing the operating state before and after the switching of the multiprocessor system in the embodiment of the present invention. For comparison with the prior art, the number of processor devices N = 8. In the processor device management control device 21 of the present embodiment, the operating condition of each processor device is grasped by the processor device status management table 71 shown in Table 3.

[Table 3] In this way, in addition to the software information (processor number and software type) corresponding to each processor device shown in Tables 1 and 2, the operating state (processor device state) of each processor device is also managed. 6 and 7
As described above, the processor device state management table 71 when the number of processors N = 8 is as shown in Table 4 (before switching) and Table 5 (after switching). Further, the loading state of the operating software of these specific examples to each processor device is the same as that shown in Table 1, and the type of software installed in this multiprocessor system is also the same as in Table 1. , A type, B type, and C type.

[Table 4]

[Table 5] Thus, at least the processor units in standby are
The number of vehicles corresponding to the type of running software is prepared. That is, since the number of types of software is 3, three processor units 6 to 8 in standby are prepared. Further, in the present embodiment, three types of software, such as type A to the processor device 6, type B to the processor device 7, and type C to the processor device 8, are loaded into the processor device in the standby standby, and the software is initialized. Execute and wait.

Next, the process of selecting the backup device of this embodiment will be described. FIG. 1 is a flowchart showing a backup device priority selection method according to an embodiment of the present invention. Consider a case where the processor device 1 in operation now fails. In this case, first, the connection switch 41 is opened in accordance with the failure detection of the processor device 1, and then the spare processor device is selected. It should be noted that the recovery logic is the same when any other operating processor device fails. In the present embodiment, since the software type of the processor device 1 in which the failure is detected is the A type, the software of the A type is loaded and the processor device in the standby state is selected (steps 101 to 103). From Table 4 above, the processor device 6 is the spare device that satisfies this condition, so the connection switch 46 is closed and the processor device 6 is connected to the processor coupling bus line 11. Since the processor device 6 has loaded the software of the same type as that of the failed processor device 1 on the memory in advance and has completed the initial setting, it is possible to perform the processing of step 104. That is, after the connection of the spare processor to the bus line 11 is completed (step 104a), the data unique to each processor device is simply transferred without the software loading process (step 104b), and the operation is immediately restarted (step 104c). )be able to. Here, the transfer of the data unique to each processor device means a transfer process of the data used for continuing the operating state of the processor device before the failure in the spare processor. Cannot be transferred to the processor control device 21, and data to be transferred at the time when the faulty processor device determines it is transferred from the processor control device 21. In addition,
This data transfer process may be omitted depending on the system. As a result of this device switching, the processor device management state table 71 is changed from the above table 4 (before switching) to table 5 (after switching), and the operating states and software types of the processor devices 1 to 8 are as shown in FIG. From (before switching) to FIG. 7 (after switching).

On the other hand, if it is determined in step 103 that there is no standby standby processor device loaded with software of the same type as the faulty processor device, step 1
Go to 05. That is, as in the conventional case, an arbitrary spare processor device is selected (step 105a), software of the same type as the faulty processor is loaded into this (step 105b), and then data unique to the spare processor device is transferred. (Step 105c), operation is started (Step 105d). Note that normally, the standby device in which the software is loaded in advance is on standby, and therefore the process of step 105 is rarely performed. Therefore, the time required for the software loading process shown in step 105b can be shortened.

Regarding the faulty processor unit, when it is repaired and normality is confirmed, it is made to stand by as a spare processor unit. In this case, in principle, the faulty processor unit is running. The same kind of software as the above-mentioned software is loaded and made to stand by as a spare processor device. In this embodiment, when a failure occurs in the processor device 1 in the operating state (801 in FIG. 8), the cause of the failure is analyzed from the in-failure state (802 in FIG. 8), and then the failed processor device is repaired. Appropriate measures are taken, and then the standby state (804 in FIG. 8)
In this case, the processor device management control device 21 loads the software of type A into the software file storage device 31.
Read out from the device and transferred to the memory of the processor device 1 which has completed the failure repair via the software transfer line 51 (803 in FIG. 8).

[0011]

According to the present invention, an alternative spare processor unit is triggered by a failure of an arbitrary one processor unit in a computer system having a multiprocessor configuration including a plurality of processor units whose operation specifications are variable depending on the type of software. Avoids software reload processing by preferentially selecting a spare processor unit in which the type of software loaded in memory in advance is the same as the software loaded in the failed processor in the built-in failure recovery processing Therefore, the system recovery time is shortened.

[Brief description of drawings]

FIG. 1 is a flowchart showing a backup device priority selection method according to an embodiment of the present invention.

FIG. 2 is a diagram showing a basic configuration of a system including loosely coupled multiprocessor devices.

FIG. 3 is a diagram showing an operating state before switching of a conventional multiprocessor system.

FIG. 4 is a basic configuration diagram showing an example of a multiprocessor system to which the backup device priority selection method of the present invention is applied.

FIG. 5 is a basic configuration diagram showing an example of a multiprocessor system to which the backup device priority selection method of the present invention is applied.

FIG. 6 is a diagram showing an operating state before switching of the multiprocessor system according to the embodiment of the present invention.

FIG. 7 is a diagram showing an operating state before switching of the multiprocessor system according to the embodiment of the present invention.

FIG. 8 is a diagram showing a state transition of the processor device in the embodiment of the present invention.

[Explanation of symbols]

 1 multiprocessor device 2 multiprocessor device 3 multiprocessor device 4 multiprocessor device 5 multiprocessor device 6 multiprocessor device 7 multiprocessor device 8 multiprocessor device N multiprocessor device 11 interprocessor coupling bus line 12 interprocessor coupling bus line 21 Processor Device Management Control Device 31 Software File Storage Device 41 Connection Switch 42 Connection Switch 43 Connection Switch 44 Connection Switch 45 Connection Switch 46 Connection Switch 47 Connection Switch 48 Connection Switch 4N Connection Switch 51 Processor Device switching control line 61 Software transfer line 62 Software transfer line 71 Processor device management status table

Claims (1)

[Claims] 1. A method for selecting a spare processor unit in a loosely coupled multiprocessor computer system comprising a plurality of processor units whose operation specifications are variable depending on the type of software loaded on a memory and a spare processor unit for substitution. In the case of incorporating the spare processor device in response to the failure of an arbitrary processor device, the type of software loaded in the memory of the failed processor device is recognized, and software of the same type as the type is loaded in the memory in advance. A spare device preferential selection method characterized in that, when there is a spare processor device loaded, the spare processor device is preferentially selected.