JPH0916425A - Information processing system - Google Patents

Abstract

(57) [Summary] [Purpose] It is possible to avoid the situation where the entire system becomes inoperable due to the failure of one device in the processing device. [Configuration] An information processing system is configured by a plurality of processing devices 11 and 21 configured by a plurality of devices 12 and 22 and external resources 31 and 32 that are shared by a plurality of processing devices and are controlled exclusively. The monitoring devices 13 and 23 in the processing device monitor the presence / absence of an abnormality in the device in the same processing device, and when an abnormality is detected, output a reset command for releasing the exclusive use of the shared resource to the interface with the outside. . Provide availability confirmation means to the device. This availability confirmation means confirms the availability of the monitoring device, and when the availability of the monitoring device cannot be confirmed,
It outputs a reset command to release the monopolization of shared resources to the interface with the outside.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing system composed of a plurality of information processing devices.

[0002]

2. Description of the Related Art Information processing systems have been increasing in scale year by year with the development of communication networks. The larger the system, the greater the impact of the system being shut down due to a failure or the like. For this reason, there is an increasing demand for ensuring system reliability.
It is natural to secure the reliability of each component in order to enhance the reliability, but as the number of system constituent units increases, it becomes more difficult to secure the reliability of the entire system.

As one of the methods for ensuring reliability, the main constituent units of the system are duplicated. For this, a plurality of information processing apparatuses having an environment in which the same work can be performed, that is, an active system and a standby system are prepared. Then, when the active system fails and it becomes impossible to continue the work thereafter, the failed active system processing device is switched to the system that has been waiting until then, and the business is continued by the standby system processing device.

On the other hand, in a system that does not have a standby system, when a certain processing device fails, the work that can be executed by another operable processing device is continuously executed and A method that allows degeneracy only for the work that was being performed is taken. Both of the above two methods are widely used as methods for continuing operation as a system. Then, in each of the methods, in addition to the dedicated external resource of each information processing device, a shared resource is provided outside the processing device in order to exchange data with another information processing device or to share the data itself. It is common to have.

FIG. 3 shows a conventional information processing system. The illustrated information processing system includes two processing devices 71 and 81. Each processing device 71, 81 includes a device 72,
82, monitoring devices 73 and 83, and input / output devices 76 and 86
Consists of The devices 72 and 82 are, for example, CPUs, and the processing devices 71 and 81 are connected via the input / output devices 76 and 86.
Manipulate data outside of. The monitoring devices 73 and 83 manage and control the operation status of each device in the processing devices 71 and 81.

The first processing unit 71 has resources 77 and 78 dedicatedly allocated thereto, and the second processing unit 81 has resources 87 and 88 dedicatedly allocated thereto. Further, both processing devices 71 and 81 share resources 91 and 92 for the purpose of data exchange and the like. The shared resources 91 and 92 are generally exclusively controlled to prevent destruction of internal data. Therefore, if the shared resource 91 is secured in the first processing device 71 at some point, the operation of the shared resource 91 from the second processing device 81 is rejected. In this case, after the processing of the first processing device 71 is completed and the shared resource 91 is released, the second processing device 81 secures the shared resource 91 and performs the processing.

Here, the shared resource 91 is allocated to the first processing unit 7
When a failure in which the first processing device 71 is inoperable occurs in the state where 1 is secured, the monitoring device 73 that detects the failure issues an instruction to the input / output device 76, and the shared resource is provided to the interface with the outside. Issue a reset to cancel the exclusive use of. As a result, the shared resource 91 secured by the failure processing device 71 is released from the failure processing device 71,
The operation from the other processing device 81 becomes possible.

[0008]

In the above method, if the processing device 71 is in a failure state due to a failure of the monitoring device 73 itself, the reset of the shared resource cannot be issued. If the shared resources 91, 92 remain reserved on the fault processing device 71 side at this point, it becomes impossible to operate the shared resources 91, 92 from another processing device 81, and the shared resources 91, 92 cannot be operated. The operation of the information processing system which requires 91 and 92 cannot be continued.

It is an object of the present invention, in an information processing system composed of a plurality of information processing devices, to avoid a situation in which the entire system becomes inoperable due to a failure of one device in the processing device. To do.

[0010]

In order to achieve the above object, an information processing system comprising a plurality of processing devices composed of a plurality of devices and an external resource shared by the plurality of processing devices and exclusively controlled In (1), the processing device monitors the presence / absence of an abnormality in the device in the same processing device, and when the abnormality is detected, outputs a reset command for releasing the exclusive use of the shared resource to the interface with the outside, and the monitoring device. It is provided in a device other than the device, and has availability confirming means for confirming the availability of the monitoring device and outputting a reset command for releasing the exclusive use of the shared resource to the interface with the outside when the availability of the monitoring device cannot be confirmed. It shall be.

Further, the plurality of processing devices may be an active processing device and a standby processing device. Furthermore, the availability confirmation means can be provided in the CPU of the processing device.

[0012]

The monitoring device checks the presence / absence of an abnormality in the devices in the same processing device, and when an abnormality is detected, outputs a reset command for releasing the exclusive use of the shared resource to the interface with the outside. Also, availability confirmation means provided in a device other than the monitoring device confirms the availability of the monitoring device, and when the availability of the monitoring device cannot be confirmed, a reset command for canceling the exclusive use of the shared resource is given to the interface with the outside. Output.

When a failure occurs in a device within the processing device by the reset command, the exclusion of the shared resource occupied by the processing device is surely released. As a result, it becomes possible to operate the shared resource from another processing device. Therefore, it is possible to avoid a situation where the entire system becomes inoperable due to a failure of one device in the processing device.

[0014]

1 shows an information processing system according to an embodiment of the present invention. The illustrated information processing system includes two processing devices 1.
1, 21. Each processing device 11, 21 has
It is composed of devices 12 and 22, monitoring devices 13 and 23, and input / output devices 16 and 26. The devices 12 and 22 are, for example, CPUs and operate data outside the processing devices 11 and 21 via the input / output devices 16 and 26. Monitoring device 13,
Reference numeral 23 manages and controls the operation status of each device in the processing devices 11 and 21. Furthermore, the devices 12, 22 are
Check availability of the monitoring devices 13 and 23,
When 23 is abnormal, the I / O devices 16 and 26 are requested to reset the interface.

The first processing unit 11 has resources 17 and 18 dedicatedly allocated thereto, and the second processing unit 21 has resources 27 and 28 exclusively allocated thereto. Further, both processing devices 11 and 21 share resources 31 and 32 for the purpose of data exchange and the like. The shared resources 31 and 32 are exclusively controlled to prevent destruction of internal data. Therefore, when the shared resource 31 is secured in the first processing device 11 at some point, the shared resource 3 from the second processing device 21 is
Operation 1 is rejected. In this case, the first processing device 11
After the processing of 1 is completed and the shared resource 31 is released,
The processing device 21 of the above secures the shared resource 31 and performs the processing.

Here, the shared resource 31 is assigned to the first processing device 1
When a failure that cannot be operated occurs in the first processing device 11 in a state where 1 is secured, the shared resource 31 secured by the failure processing device 11 is released from the fault processing device 11 and the other processing device 21 It is possible to operate from. For this reason, the monitoring device 13 includes the devices 1 in the same processing device 11.
The device 12 in the same processing device 11 diagnoses the availability of the monitoring device 13 by managing and controlling the operation status of item 2.

As a result of the diagnosis of the device 12, if the monitoring device 13 is available, the monitoring device 13 gives an instruction to the input / output device 14 to cause the interface with the outside to issue a reset for releasing the monopolization of the shared resource. When the monitoring device 13 is not available, the device 12 that has made a diagnosis issues an instruction to the input / output device 14 to cause the interface with the outside to issue a reset for canceling the monopolization of the shared resource.

As a result, the shared resource 31 secured by the failure processing device 11 is released from the failure processing device 11 and can be operated by another processing device 21. The case where the first processing device 11 has a failure has been described above, but the same is true when the second processing device 21 has a failure. FIG. 2 shows a specific example of the system shown in FIG.

The illustrated information processing system is composed of two clusters 41 and 51. Cluster 41, 51
Is the minimum constituent unit of the information processing apparatus. Each cluster 41, 51 has a central processing unit (CPU) 42, 52, a service processor (SVP) 43, 53 for controlling each unit, and a main storage unit (MS).
U) 44, 54 and main memory control unit (MCU) 45, 5
5 and channel processors (CHP) 46 and 56.

The central processing units 42 and 52 operate data outside the clusters 41 and 51 via the channel processors 46 and 56. The service processors 43 and 53 are
It manages and controls the operating state of each operation in the clusters 41, 51. The first cluster 41 has dedicated I / O resources 47, 48, and the second cluster 51 has dedicated I / O resources 57, 58. Further, both clusters 41 and 51 share I / O resources 61 and 62 for the purpose of data exchange and the like.

The correspondence between FIG. 2 and the above-mentioned FIG. 1 will now be described. The clusters 41 and 51 in FIG. 2 correspond to the processing devices 11 and 21 in FIG. The central processing unit 42 of FIG.
52, main storage devices 44 and 54, and main storage control devices 45 and 5
5 corresponds to the devices 12 and 22 of FIG. Although the availability of the monitoring devices 13 and 23 is diagnosed by the devices 12 and 22 of FIG. 1, this diagnosis is performed by each device 4 in FIG.
Any of 2, 44, 45 or 52, 54, 55 may be performed. Here, the central processing units 42, 5
2 will be described as an example. The service processors 43 and 53 of FIG. 2 correspond to the monitoring devices 13 and 23 of FIG. 1, and the channel processors 46 and 56 of FIG. 2 correspond to the input / output devices 14 and 24 of FIG.

The shared I / O resources 61 and 62 are generally exclusively controlled in order to prevent destruction of internal data.
Therefore, if the shared I / O resource 61 is secured in the first cluster 41 at some point, the operation of the shared I / O resource 61 from the second cluster is rejected. In this case, the first
After the processing of the cluster 41 is completed and the shared I / O resource 31 is released, the second cluster 41 secures the shared I / O resource 31 and performs the processing.

Here, a case where a failure occurs in the cluster 41 will be described. It should be noted that when a failure occurs in the cluster 51, the description is omitted, but the same processing as the following description is performed. When a failure occurs in the cluster 41,
If the central processing unit 42 is operable, the central processing unit 4
2 confirms the availability of the channel processor 43 (operation diagnosis). As a specific confirmation method, the hardware control area of the main storage device 44 is used to determine whether the flag set by the central processing unit 42 can be reset by the service processor 43 within a fixed time. The central processing unit 42
If the service processor 43 recognizes that it is available, the central processing unit 42 does nothing to the channel processor 46. When an error occurs in a device in the cluster 41, the service processor 42 issues an I / O system reset instruction to the channel processor 46.

On the other hand, when the service processor 43 fails and the flag set by the central processing unit 42 cannot be reset within a predetermined time, the central processing unit 4
2 determines that the service processor 43 has failed, and the central processing unit 42 itself sends I / O to the channel processor 46.
Issues a system reset instruction. It uses the same circuitry that conveys I / O instructions from central processing unit 42 to channel processor 46.

Further, when the central processing unit 42 and the service processor 43 are constantly diagnosing each other's availability, a failure occurs in either the central processing unit 42 or the service processor 43, and it is difficult to continue the processing. Even when it is determined that the central processing unit 42 or the service processor 43 is operable, the I / O system reset instruction is promptly issued to the channel processor 46.

The above processing can be handled by the control program of each device, and the availability of the system can be improved without adding a new circuit. In the above description, the I / O resource has been described as the external resource, but it is also possible to target the shared memory instead of the I / O resource.

[0027]

According to the present invention, in a system composed of a plurality of processing devices, it is possible to avoid a situation where the entire system becomes inoperable due to a failure of one device in the processing device. Further, the monitoring device itself can detect and block a fault occurring in the processing device system at an early stage by diagnosing its availability by another device in the same processing device.

As described above, the availability of the information processing system is improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an information processing system according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a specific example of the information processing system in FIG.

FIG. 3 is a block diagram showing a conventional information processing system.

[Explanation of symbols]

 11, 21 ... Information processing device 12, 22 ... Device 13, 23 ... Monitoring device 16, 26 ... Input / output device 17, 18, 27, 28 ... Resource 31, 32 ... Shared resource 41, 51 ... Cluster 42, 52 ... Central Processing device 43, 53 ... Service processor 44, 54 ... Main storage device 45, 55 ... Main storage control device 46, 47 ... Channel processor 47, 48, 57, 58 ... Resource 61, 62 ... Shared resource

Claims (3)

[Claims] 1. An information processing system, comprising: a plurality of processing devices composed of a plurality of devices; and an external resource shared by the plurality of processing devices and controlled exclusively, wherein the processing devices are the same processing device. The presence or absence of an abnormality in the device inside, when detecting an abnormality, a monitoring device that outputs a reset command for canceling the monopolization of shared resources to the interface with the outside, and a device other than the monitoring device,
The availability of the monitoring device is confirmed, and when the availability of the monitoring device cannot be confirmed, availability confirmation means for outputting a reset command for canceling monopolization of the shared resource to an interface with the outside is provided. Information processing system. 2. The information processing system according to claim 1, wherein the plurality of processing devices are an active processing device and a standby processing device. 3. The information processing system according to claim 1, wherein the availability confirmation unit is provided in the CPU.