JPH0567974B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Object of the Invention The present invention relates to a distributed processing system, and particularly to a system in a distributed processing system in which subsystems are related to each other and a failure of one subsystem may have an adverse effect on other subsystems. The present invention relates to a processing method in a distributed processing system that prevents downtime and achieves high reliability.

Prior Art The aspects of failure detection and diagnosis in conventional distributed processing systems and recovery processing based on the results are as follows.
FIG. 1 is positioned in comparison with an aspect of recovery processing in the present invention, which will be described in detail later. In FIG. 1, the vertical axis indicates the target of failure detection and diagnosis, and the horizontal axis indicates the target of recovery processing performed based on the results of the detection and diagnosis.

Among the aspects shown in Fig. 1, a failure within the own subsystem is detected by itself and based on the result,
A device that performs recovery processing from a failure in its own subsystem or another subsystem is called a self-diagnostic tester. In this self-diagnosis tester, each subsystem "all other subsystems completely perform fault detection, diagnosis, and recovery processing based on the results."
The system is constructed based on this premise. Therefore, if each subsystem fails or makes an error in fault detection, diagnosis, or recovery processing, there is a problem in that other subsystems are affected by the failure and the system goes down. It also detects and diagnoses failures in other subsystems, and as a result,
A device that causes other subsystems to perform recovery processing is called a centralized tester. Failure or error in failure detection, diagnosis, or recovery processing of this centralized tester also directly affects other subsystems, resulting in a system failure.

In contrast, the present invention provides an autonomous tester that detects and diagnoses failures in other subsystems, and performs recovery processing to protect its own subsystem from failures in other subsystems based on the results. It is used. FIGS. 2 and 3 illustrate the functions of the three types of testers mentioned above.

There are two types of self-diagnostic testers, A and B in Figure 2, and the subsystem 1 tester detects a failure in its own subsystem and performs recovery of its own subsystem 1 based on the results of diagnostic processing 11 or 12. It instructs the process 21 or the recovery process 22 or 22' for the other subsystems 2 and 3. If this process is not performed normally (x marks in Figure 3 A and B), other subsystems 2, 3 will be affected 31, 32 or 32'.
and cause its failure.

In the centralized tester (see Figure 2 D), the tester for subsystem 1 detects and diagnoses failures in other subsystems 2 and 3 14 and 14', and performs recovery processing 24 and 24' based on the results. Subsystem 2,
3. Therefore, a failure in subsystem 1 will directly spread to other subsystems 2 and 3.
4,34'.

In contrast, an autonomous tester (see Figure 2 C)
Then, the tester of subsystem 1 detects and diagnoses failures of other subsystems 2 and 3 13 and 13',
Based on the results, the own subsystem performs recovery processing 23 to protect the failure from spreading to other subsystems. Therefore, a failure or error in failure detection, diagnosis, or recovery processing of the subsystem 1 will not affect the other subsystems 2 and 3.

Purpose of the Invention As stated in the above description, the purpose of the present invention is to solve the above-mentioned problems in distributed processing systems using conventional self-diagnosis testers or centralized testers, and to prevent system downs. The object of the present invention is to provide a processing method in a highly reliable distributed processing system.

General Description of the Invention The above object of the present invention is to provide a distributed processing system in which a plurality of subsystems are connected to each other on an equal basis, in which each of the subsystems is capable of diagnosing failures of other subsystems. This is achieved by a distributed processing system characterized in that it has a function to perform the above-mentioned functions and is configured to perform protection processing for its own subsystem based on the result of diagnosing the failure of the other subsystem.

Embodiments of the Invention Hereinafter, embodiments in which the present invention is applied to a loop transmission system will be described in detail based on the drawings.

FIG. 4 is an overall configuration diagram of a distributed processing system showing an embodiment of the present invention. This system consists of a pair of transmission control devices installed on dual loop transmission paths having mutually opposite transmission directions. Paired transmission control devices 100 and 110, 200 and 2
10,... are mutually detour routes 100A, 11, respectively.
They are connected by 0A; 200A, 210A;... Furthermore, the range of the subsystems is indicated by broken lines. For example, subsystem 1 includes the transmission control device 100, the loop transmission path 1200, and the detour 100A, and subsystem 2 includes the transmission control device 200, the loop transmission path, and the detour 100A. 2300 and Detour 20
Consists of 0A. Furthermore, subsystem 1 is connected only to subsystems 2, 4, and 5.

Figure 5 shows details of subsystem 1.
A processing device (host) 1000 is connected to the paired transmission control devices 100 and 110 via host-to-host transmission paths 111 and 211. The transmission control devices 100, 110 and the processing device 1000 also include testers (BIT) 100B, 110B and 100.
0B is built-in. This BIT is a function of the transmission control device that detects failures in other subsystems,
It performs diagnosis and recovery processing. In addition, tester (EXT) 1 that identifies system failure points
010 is provided in the processing device 1000. The EXT
1010 outputs the location of the failure to the display device 1020 to inform maintenance personnel, as will be described in detail later. Although not shown, other subsystems 2 and 3
... is also configured in exactly the same way as the subsystem 1 described above.

Below, the tester BIT100B, 110B, 1
The operations of 000B and EXT1010 will be explained in detail using FIGS. 6A to 6G, FIGS. 7 and 8. In addition, in the following explanation, the transmission control device (hereinafter referred to as "NCP") 400, 4
Assume that 10 is down.

Now, suppose that the NCP 200 sends a message 201 onto a transmission path (also referred to as a "loop") 2300. If the message does not return to the sender NCP200 after a certain period of time _T1 , the NCP2
00 retransmits the same message for confirmation.
If retransmission reaches a certain number of times N ₁ or more, BIT20
0B determines that a failure has occurred on the transmission path (see FIG. 6A). BIT 200B then sends a small loop check signal 202 to check whether the message can be transmitted to NCP 300 of the adjacent subsystem. BIT3 of NCP300
When 00B receives the small loop check signal 202 from the transmission path 2300, it determines that a failure has occurred somewhere on the transmission path, and at the same time transmits the small loop check signal 202 to the NCP 310, it also A loop check signal 302 is sent onto the transmission line 3400.

NCP 310 receives small loop check signal 202 from the direction of detour 300A and sends it onto loop 3200. NCP 2 receiving the small loop check signal 202 from loop 3200
The No. 10 BIT 210B performs the same processing as the above-mentioned BIT 300B. As described above, when the small loop check signal 202 returns to the source NCP 200, the BIT 200B determines that it can send the message to the adjacent NCP 300, and thereafter sends the message on the loop 2300.

On the other hand, BIT200B informs NCP210 that a failure may have occurred on the inner loop.
Notify BIT210B. As a result, BIT21
0B sends out a small loop check signal 212 onto loop 2100, similar to BIT 200B described above. Each NCP that received a small loop check signal
BIT sequentially performs small loop checks in the same manner as described above. In this example,
Assuming that NCPs 400 and 410 are down, BITs 300B and 110B will not receive the small loop check signal back.

As a result, BIT300B is detoured to 300A.
, the BIT 110B constitutes a detour 110A, and from then on, each received message is routed through the loop 34.
00 and 1400, but only to the detour routes 300A and 110A (see FIG. 6B).

BIT300B and 110B that made up the detour are:
Detour route configuration notification signals 303 and 113 are sent out, respectively. NCP100 connected to BIT1010
Or 110 is the detour configuration notification signal 303,
113, it is transmitted to the processing device 1000 (see FIG. 6C).

The EXT 1010 of the processing device 1000 detects the failed part from the detour configuration notification signal 113.
3' (shaded area downward to the right in Figure 6D), and
Based on the detour configuration notification signal 303, the faulty portion is diagnosed as 303' (the diagonally shaded portion upward to the right in FIG. 6D). The EXT 1010 combines the above two diagnostic results and displays on the display 1020 that the faulty part is located somewhere between 113' and 303'.

BIT300B and 110B that made up the detour are:
To check whether the fault has been recovered, small loop check signals 302, 112 and large loop check signals 304, 114 are alternately and periodically sent out. Since the large loop check signal is not bypassed by any NCP, the large loop check signal 3
If 04 goes around the loop and returns to the source,
The BIT 300B determines that the failure on the loop has been recovered and cancels its own detour 300A. Note that even when the small loop check signal sent by the BITs 300B and 110B is returned, the BIT that sent it cancels its own detour (see FIG. 6E).

The BIT 300B that has canceled the detour sends out a detour cancellation notification signal 305. The NCP 100 connected to the EXT 1010 receives the above detour release notification signal 3.
05, it is transmitted to the EXT 101 (see FIG. 6F).

EXT1010 cancels the detour notification signal 303 from BIT00B based on the detour cancellation notification signal 305 and also cancels the failure diagnosis result on the loop 4100 in response to the above-mentioned diagnosis results 113' and 303'. The local diagnosis results are reduced to a range of 113'' (see Figure 6G).

Note that in order for transmission to occur normally, each
It is essential that the NCP addresses are set without duplication. Therefore, when turning on the power, each BIT checks whether another NCP has set the same address as its own address. Now, when NCP100 is started, it is already
NCP200, 300, 310, 210 and 1
10 is activated (see Figure 7A), BIT 100B of the activated NCP 100
sends out an address train 106 as detailed below. The address train 106 corresponds to each NCP.
They are received one after another and sent out after filling in each address. After this address train 106 has made two circuits on the transmission path, the address train 106 is
Transferred to EXT1010. FIG. 8 shows the process by which the address train 106 is configured each time it passes through each NCP. In each BIT, an address 106A is written, and if a detour is being configured, the detour flag 106B is set to "1".

The BIT 100B receives the address train 106 that has made two rounds on the transmission path, and returns its own address 100.
If the addresses of other subsystems in between are not the same in the first and second rounds, it is assumed that there is a duplicate address, and no further transmission is performed. Also, NCP100 connected to EXT1010,
Take in the address train 106 and write it to EXT1
Notify at 010. The EXT 1010 updates the system configuration to 10 by checking the detour flag and address in the address train 106.
6' (see FIG. 7B), and the display device 102
Output to 0.

The BIT in each processing device has a function of switching the transmission direction to the NCP 110 or 100 when transmission to the NCP 100 or 110 is impossible.

In the above embodiment, an example in which the present invention is applied to a loop transmission system is shown, but the present invention is not limited to this, but can be similarly applied to other distributed processing systems such as line-type, mesh-type distributed processing systems, etc. It goes without saying that it is possible.

For details of the operation shown in the above embodiment, please refer to the "loop transmission system" (Japanese Patent Application No. 55-93924) and the "transmission control device" (Utility Application No. 56-87835) proposed earlier by the applicant. ) etc. may be referred to.

Effects of the Invention As described above, according to the present invention, in a distributed processing system in which a plurality of subsystems are connected to each other in an equal relationship, each of the subsystems diagnoses the failure of the other subsystems. The system is configured to perform protection processing for its own subsystem based on the diagnosis results of failures in other subsystems, thereby preventing system failure in the distributed processing system and creating a highly reliable distributed processing system. This has the remarkable effect of being achievable.

[Brief explanation of the drawing]

Figure 1 shows the functional classification of various testers, Figure 2
3 are diagrams schematically showing the functions of various testers, FIGS. 4 and 5 are diagrams showing a schematic configuration of an embodiment of the present invention, FIGS. 6 A to G, and FIGS. 7 A, B is a diagram showing the operation of the embodiment, and FIG. 8 is a diagram showing the address train. 1, 2, 3, 4, 5: subsystem, 100,
110; 200, 210;...: NCP, 100B,
110B; 200B, 210B;...: BIT, 10
00: Processing device, 1010: EXT, 1020:
Display device.

Claims (1)

[Scope of Claims] 1. In a distributed processing system having a plurality of subsystems connected via a transmission medium, each subsystem has a method for diagnosing failures in other subsystems and the transmission medium connected thereto. , transmits a check signal to the transmission medium, diagnoses the failure based on the response to the check signal, and when the failure is detected as a result of the diagnosis, controls to perform protective processing within its own subsystem. , a signal indicating the control state is transmitted onto the transmission medium, and a processing device provided corresponding to at least one of the plurality of subsystems determines a failure location based on the signal indicating the control state. A processing method characterized by outputting. 2. After the above-mentioned failure is detected and each subsystem is controlled to perform its own protection process,
A check signal for diagnosing recovery from the fault is transmitted to the transmission medium, recovery from the fault is determined based on the response to the check signal, and when recovery from the fault is detected, recovery from the fault is detected. Processing in the distributed processing system according to claim 1, characterized in that a signal indicating the status is transmitted onto the transmission medium, and the processing device outputs the recovery status of the failure based on the signal indicating the recovery status. Method.