JPH0414934A - Communication line fault detecting method in distributed processing system - Google Patents

Abstract

PURPOSE:To locate a fault section in a communication line when one system is faulty when plural systems of communication lines forming a loop are present thereon by providing a token counter on each processor. CONSTITUTION:A token counter reading instruction from a system management equipment 101 is transmitted to a CPU 404 of each processor 30 via a reception buffer, a buffer control section, and a CPU interface 403. The CPU 404 reads the counted value of the token counter 402 via the CPU interface 403 accordingly and reports the counted value to the system management equipment 101 by using a sound communication line 311. The system management equipment 101 detects the fault location of a faulty communication line from the result of report. Thus, even when any location of a loop communication line 310 has a fault, the fault location in the communication line 310 is specified.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a communication path failure detection method in a distributed processing system.

[Prior Art] A plurality of processing devices and one system management device are connected in a loop via a communication path, each processing device has its own processing function, and each processing device exchanges information with each other via the communication path. Distributed processing systems are known in which the tasks of one system are shared and processed while exchanging data.

FIG. 4 is a block diagram showing such a distributed processing system. In the figure, 101 is a system management device that manages the distributed processing system, and 102a to 1
02e is a processing device, and 103a and 103b are loop-shaped communication paths (sometimes simply called communication paths or loops).

In a distributed processing system that uses the token hashing method, as a means of establishing communication between each node (processing device or system management device), frame-structured data called a token that can be granted transmission rights is inserted into the communication path. When the token is circulated and a node that wants to communicate captures the token, that node is allowed to use the communication path and communicate.

In the distributed processing system shown in Figure 4, one
Since two systems, 03a and 103b, are provided, if one communication path fails, the remaining communication path can be used to exchange information between nodes.

Now, in such a distributed processing system, if 103
If a failure occurs in the communication path 103a, the system management device 101 needs to detect in which section of the communication path 103a the failure occurs. A conventional communication path failure detection method in such a distributed processing system will be described below with reference to FIG.

FIG. 5 shows each processing device (including a system management device) 102.
FIG. 2 is a block diagram showing a loop interface section of the FIG. Since the loop interface units 201 are provided corresponding to the number of communication paths, in this case, the communication path 103a
, 103b are provided, but in FIG.
The one for communication path 103a is shown.

In FIG. 5, 202 is a reception buffer, 203 is a buffer control unit, 204 is a frame control unit, 205 is a frame discard communication control unit, 206 is a transmission selector, 207 is a transmission buffer, 208 is a CPU interface, 209 is a CPU, 210 is memory.

Normally, tokens and data are taken into the loop interface unit 201 by the receive buffer 202. The buffer control unit 203 transmits data necessary for the own device among the captured data to the CPU 208 via the CPLI interface 208, and sends data to be transmitted from the own device to the transmission selector 206. The frame control unit 204 passes received data that is unnecessary for the own device through the frame discard communication control unit 205, or controls termination. The transmission selector 206 is
The data to be transmitted is selected and sent to the transmission sofa 207.

Suppose now that the system management device 101 detects that a failure has occurred in the communication path 103a. Then, the system management device 101 removes the communication path 103a from active use and sends a test command for the communication path 103a to each processing device 102 using the healthy communication path 103b.

Then, in each processing device 102, the loop interface unit 201 corresponding to the communication path 103a creates a communication path as shown by the broken line from the sending communication path 207 to the receiving communication path 202, and the communication path 203→206→207→202→2 is created.
03 is formed within the device itself, a test frame is sent through this loop, and based on the transmission/reception status, it is determined whether or not there is a failure in the communication path 103a within the device itself, and the result is transmitted using a healthy communication path 103b. and reported to the system management device 101.

[Problems to be Solved by the Invention] In the conventional communication path failure detection method in a distributed processing system as described above, it is possible to detect failures in a part of the reception sofa 202, a part of the frame discard communication control unit 205, a part of the transmission selector 206, etc. If a failure occurs in the communication path between the processing devices or a failure occurs in the communication path between the processing devices, there is a problem in that the location of the failure cannot be determined and the failure section cannot be specified.

An object of the present invention is to solve the problems in the prior art and to provide a communication path failure detection method in a distributed processing system that can identify a failure section in a communication path no matter where the failure occurs. There is a particular thing.

[Means to solve the problem]

To achieve the above object, the present invention provides a distributed processing system in which a plurality of processing devices and one system management device are connected in a loop via communication paths, in which a plurality of communication paths forming the loop are provided. In this case, each processing device is provided with a token counter in order to detect the fault section of the fault loop when one of the systems fails.

[Effect]

When the system management device somehow detects that a failure has occurred in one of the multiple loops, it instructs each processing device to reset its own token counter via the normal loop. Then, by generating a token and sending it through the faulty loop where the failure occurred, the processing device that received the token counts the number of times the token has been received using its token counter, and then passes it to the next processing device. When the system management device detects that the token has not been returned to the system management device even after a certain period of time has elapsed, the system management device updates the token counter of each processing device at that time. The count value is reported through the normal loop, and the fault section of the fault loop is detected from the report result.

In other words, the system management device sequentially looks at the count value in the token counter starting from the processing device upstream in the loop, and determines where the count value in the subsequent processing device decreases by one after that processing device is reached. find,
It is determined that the area between the boundary device and the adjacent device is a failure area.

〔Example〕

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

FIG. 1 is a block diagram showing a loop interface section of each processing device used to explain one embodiment of the present invention.

In the figure, 302 is a processing device, 310 and 311 are loop-shaped communication paths, 401 is a loop interface section corresponding to the communication path 310, 402 is a token counter, 403 is a CPU interface, 404 is a CPU,
It is.

Compared to FIG. 5, there is no other structural change except that a token counter 402 is added.

FIG. 2 is an overall configuration diagram of a distributed processing system configured by a processing device 302 having such a loop interface section and a system management device 301. Processing device 30
2, 302a and 302b are shown as examples.

Next, the operation of the communication path failure detection method in the distributed processing system according to the present invention will be explained with reference to FIGS. 1 and 2.

Now, assume that the system management device 301 detects by some method that a failure has occurred in the communication path 310. The system management device 101 then removes the communication path 310 from active use, and uses the healthy communication path 311 to send a command to each processing device 302 to reset their respective token counters. That is, the send command from the system management device 101 is sent to the CPU via the reception buffer, buffer control unit, and CPU interface 403 in each processing device 302.
The information is transmitted to the PU 404, and the CPU 404 responds to the C
A signal is sent to the token counter 402 via the PU interface 403 to reset it.

Next, the CPU in the system management device 301 generates a frame worth a token, and sends the frame via the CPU interface, buffer control unit, transmission selector, and transmission frame.
It is sent to the communication path 310 where the failure has occurred.

When a token is received in each processing unit 302, it is transmitted to the CPtJ 404 via the receiving Hanwha, buffer control unit, and CPtJ interface 403, and the CPU 404 accordingly sends a signal to the token counter 402 via the CPU interface 403. Send it out and count up the number of degrees. The token is then passed further downstream to a processing device.

The token is lost at a failure point during the process and is not returned to the system management device 301. When the system management device 301 uses a timer or the like to detect that the token has not returned to the system management device 301 even after a certain period of time has passed, it sends the token to each processing device 302 using a healthy communication channel 311. Send a command to read and report the count value of each token counter.

That is, the token counter read command from the system management device 101 is transmitted to the CPU 404 in each processing device 302 via the reception buffer, buffer control unit, and CPU interface 403, and the CPU 4
04 reads the count value of the token counter 402 via the CPtJ interface 403 and sends it to the system management device 101 using the healthy communication channel 311.
Report to. The system management device 101 detects the failed section of the failed communication channel from the report result.

That is, the system management device 101 sequentially checks the count values in the token counter starting from the upstream processing device side of the communication path, and determines whether the count value of the subsequent processing devices decreases by one after that processing device is reached. It is determined that the area between the device on the boundary and the device adjacent to it is a faulty area.

Please refer to FIG. 3, which shows the processing flow of the communication path failure detection method in the distributed processing system described above. There is no need to further explain Fig. 3.

〔Effect of the invention〕

As described above, according to the present invention, in a distributed processing system, there is an advantage that a fault section in a communication path can be identified no matter where in the loop communication path a fault occurs. As a result, if the probability that a failure has occurred in the loop interface section is higher than the probability that a failure has occurred in the communication channel itself, it is possible to replace the loop interface in that identified section. , it is possible to quickly recover from a failure and improve maintenance efficiency.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a loop interface section in each processing device when the present invention is implemented, FIG. 2 is a block diagram showing a distributed processing system when the present invention is implemented, and FIG.
4 is a block diagram showing an example of the configuration of a distributed processing system; FIG. 5 is a block diagram showing a loop interface section in a conventional processing device; It is. Explanation of symbols 101.301...System management device, 102゜30
2... Processing device, 202... Reception buffer, 203
...Buffer control unit, 204...Frame control unit,
205...Frame discard communication control unit, 206...Transmission selector, 20'l...Transmission Hanwha, 208,4
03...cpu interface, 209.404...
・CPU, 210...Memory, 402...Token counter agent Patent attorney Akio Namiki

Claims (1)

[Claims] 1) In a distributed processing system in which multiple processing devices and one system management device are connected in a loop via communication paths, if there are multiple communication paths forming the loop, one of the systems may fail. A communication path failure detection method for detecting a failure section of the failure loop, wherein a token counter is provided in each of the processing devices, and the system management device detects a failure in one of the plurality of loops. When it detects that something has happened, it instructs each processing unit through the normal loop to reset its own token counter, then generates a token and sends it through the failed fault loop. By doing this, the processing device that receives the token counts the number of times the token has been received using its token counter, then sends the token to the next processing device, and the system management device confirms that the token will not be returned to the system management device. , the system management device causes the count value in the token counter of each processing device at that time to be reported via the normal loop, and detects the fault section of the faulty loop from the report result. A communication path failure detection method in a shape processing system.