JPH0696040A - Fault detection system - Google Patents

Abstract

PURPOSE:To speedily and easily detect a fault and prevent the fault from being overlooked by providing each computer with a fault detection program and an inspection program which performs various inspection. CONSTITUTION:Respective computers 1-3 have fault detection programs 11, fault inspection program registration files 13, and talkers 15 as transmission programs and listeners 17 as reception programs which constitute communication control programs for a communication among the respective fault detection programs 11 provided on the respective computers 1-3. In this case, the contents (program name) of inspection performed in the respective computers 1-3 and file names of inspection results are registered in the fault detection program registration files 13 on the computers 1-3. Then, the fault detection programs 11 starts the respective inspection programs while referring to the fault detection program registration files 13 and transfer result files. Therefore, the addition and corrections of the inspection contents and the customerization of the inspection contents corresponding to a system and computers are facilitated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fault detecting a fault in a distributed processing system in which a network is composed of a plurality of computers and a plurality of programs operating on each computer perform predetermined processing while mutually communicating. Regarding detection system.

[0002]

2. Description of the Related Art A method of controlling various manufacturing lines, for example, using a plurality of computers and utilizing network communication is widely used. For example, a manufacturing management system of ASIC LSI will be described as an example. In such a system, among a plurality of computers, a plurality of programs that operate separately on each computer (always running) exchange messages with each other, and schedule lots on the LSI manufacturing line. It manages the progress and collects data. As described above, in a system in which a network is formed between a large number of computers and distributed processing is performed, stopping a certain computer or a program causes the entire system to stop. Therefore, there is a need for a highly reliable system that can detect such failures early.

Such an automatic failure detection system can be used for a production management system, a monitoring system, etc. of various manufacturing lines for controlling a plurality of devices using a computer network.

The computer to which the present invention is applied includes not only a computer having a CRT terminal but also a control terminal (sequencer or the like) equipped with a programmable ROM, a board, and the like. It is adaptable. Further, the network includes not only a LAN (local area network) and a WAN (wide area network) that can communicate with each other, but also a connection form with the sequencer and the like.

[0005]

As described above, in order to detect a failure in a processing system for performing distributed processing by utilizing communication between a plurality of computers via a network, conventionally, for example, a system is individually set on each computer. It is necessary to check the operating programs by inputting the commands of the above commands and to check whether communication between computers is possible, and it takes time and effort to check the entire system. In addition, there are problems such as missing check.

The present invention has been made in view of the above,
It is an object of the present invention to provide a fault detection system capable of easily and automatically inspecting the operating status and faults of a distributed processing system using a network.

[0007]

In order to achieve the above object, the fault detection system of the present invention configures a network with a plurality of computers, and performs a predetermined communication while mutually communicating between a plurality of programs operating on each computer. Is a fault detection system for detecting a fault in a distributed processing system that performs the above process, wherein each computer is provided with a fault detection program and an inspection program for executing various inspections, and the fault detection program provided in each computer is a communication control program. The failure detection program installed in each computer is connected to the computer via the communication control program and the failure inspection program in another computer to activate the inspection program in the other computer. Start up and check the normal operation of programs that should always be started in each computer and Together to perform the normal operation check of the communication between the computer each other, and subject matter to receive the contents of the test results from each computer.

[0008]

In the fault detection system of the present invention, each computer is provided with a fault detection program and an inspection program for executing various inspections, and each fault detection program activates each inspection program and is always activated in each computer. The normal operation check of the program to be performed and the normal operation check of the communication between the computers are performed, and the contents of the inspection result are received from each computer.

[0009]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a fault detection system according to an embodiment of the present invention. In the fault detection system shown in the figure, a network is composed of a plurality of computers, that is, a primary computer 1, a secondary computer 2, and a tertiary computer 3, while mutual communication is performed between a plurality of programs operating on each computer. Each computer 1 detects an operating condition and a failure in a distributed processing system that performs processing.
Reference numerals 3 to 3 respectively include a fault detection program 11, a fault inspection program registration file 13, and a talker 15 and a reception program which are transmission programs constituting a communication control program for communicating between the fault detection programs provided in each computer. Have a listener 17.

The fault detection program 11 transfers the inspection contents and the inspection result between the function of inspecting the fault in the own computer and the fault detection program in another computer,
It has a function to transfer a common file used in the system and collate its contents. It is possible to select whether to check all computers or only between certain specific computers by specifying flags at startup.

When the failure detection program is activated, each inspection program is activated and the inspection is executed with reference to the failure detection registration file, and the inspection result is transferred to the boot source computer. As a result, the test results of all the computers can be centrally managed by the boot computer.

As functions of the inspection program, a function of checking whether a program that should be always started in each computer is operating normally, a function of checking whether communication between computers is normally performed, and a plurality of functions A function of checking whether or not various definition files commonly used between computers (hereinafter referred to as common files) match between computers is included. In addition to this, if there is different environment setting information for each computer such as what kind of device is connected to each computer, double definition is added to the definition content,
Functions such as checking for inconsistencies such as registration omissions can be given.

Fault detection is broadly divided into three stages. In the first stage, the fault in the computer (primary computer 1) that activates the fault detection program is checked. In the second stage, a fault in another computer (secondary computer 2) connected to the system and the consistency of the common file used in the system are checked. In the third stage,
Perform mutual communication check between each computer connected in the system. It should be noted that the check in the third stage can be omitted when checking only between certain specific computers. In some systems, a plurality of terminal computers are connected under one host computer, and mutual communication between the terminals may not be performed. In this case, the third stage check is unnecessary.

Communication in the fault detection program 11
This is performed using a talker (transmission program) 15 and a listener (reception program) 17, which are communication control programs that perform communication between programs activated in the system. In the case of a specific communication message (identified by the message number or its content), the communication control program
Recognizing as a message from the fault detection program 11,
Perform special processing. That is, the content of the communication message is
If the message is a "request message" such as an activation request or inspection request to the failure detection program, the listener 17 receiving the message activates the detection program and passes the message. If the communication message content is a "reply message" which is a response from the failure detection program to another failure detection program, the message is transmitted to the destination failure detection program. If there is no response from the listener of the partner computer, the talker outputs error information to the failure detection result file.

The fault detection result is not only converted into a file on each computer, but also transferred to the primary computer and output as a file. Therefore, it is possible to confirm the inspection result of the operating states of all the computers connected to the system at the terminal that has activated the failure detection program.

The contents (program name) to be inspected in each computer and the file name of the inspection result are registered in the fault detection program registration file (hereinafter simply referred to as registration file) 13 on each computer. The failure detection program 11 is configured to start each inspection program while referring to the file 13 and transfer the result file. Therefore, it is easy to add / correct the inspection contents and customize the inspection contents according to the system and the computer.

Next, the operation procedure of the fault detection program will be described.
An example of checking only between specific computers (computer 1-computer 2) will be described with reference to FIG.

The fault detection program started on the computer 1 refers to the registration file on the self computer to detect the fault in the self computer. The inspection item can be changed by registering the registration file.

For example, in the case of LSI production line production control, the following items are inspected.

(1) Checking the computer information connected to the system (2) Checking the common file (checking the file format) (3) Checking the startup status of the program that should always be started in the computer itself (4) Communication check between programs in the computer (response check) (5) Database setting environment check (6) Disk free space check (7) Environment setting information check for each computer (computer system settings, device connection information, etc.) Individual inspection The inspection contents and method of the target will be described later.

The inspection result of (1) is output as a result file.

A request to activate the fault detection program is issued to the computer 2. Even before the completion of all the inspections, if the communication control program (talker / listener) can be confirmed to be activated, the activation request may be performed first. In that case, since the inspections of the computers 1 and 2 are performed in parallel, the inspection time can be shortened.

In the computer 2 which has received the activation request, the listener activates the fault detection program. The fault detection program refers to the registration file on its own computer to detect the fault. The processing contents are the same as those in (1) to (7). Regarding the check on the environment setting information file of (7), the file content differs for each computer, but the processing content such as the inspection procedure is the same. However, the check contents of the common file of (2) are slightly different from those of.
That is, this file is originally a file that is commonly used in the system (for example, various name registration files).
Therefore, not only the presence / absence of files and the consistency of the format, but also the data content is checked.

The method will be described later.

The inspection result of (1) is output to the computer as a result file. The output file name at this time is registered in the registration file.

When the result file is designated as a transfer file in the registration file, the inspection result file is transferred from the computer 2 to the computer 1.

Computer 1 that received the inspection result file
Then, in order to distinguish it from the inspection result of the self computer, an identifier indicating that this file is the inspection result from the computer 2 is added to the file name, and the result file is output.

With the above, a series of processing is completed. Such processing may be activated by the operator by inputting a command to the failure detection program when necessary, or may be set in the computer system so as to be activated periodically.
For example, in a system in which the OS (operating system) of the computer is UNIX, it can be started up periodically by registering it in the frontab file. In this case, the inspection program is started unattended and the inspection result is output to a file. If there is error information by referring to this file, you can immediately notify the occurrence of a failure by preparing another program that issues an alarm or message. For example,
If you provide a column at the top of the failure detection result file to set whether or not an error has occurred in the inspection program of each computer as a flag (1: with error, 0: without error), just see this flag. Presence of error can be identified. For details of the error content, it is sufficient to refer to the following output in characters.

When checking all the computers in the system, it is checked whether or not the computers connected to the system can communicate with each other by sending a message and checking whether or not there is a reply. In the mutual communication check, the primary computer instructs the secondary computer to instruct the other computer to perform the check and is activated, and the check result is received. The secondary computer performs communication check using the instructed computer as the tertiary computer and returns the result to the primary computer via the secondary computer. At this time, the computer connected to the system is recognized by the "computer information file". This computer information file consists of the name of the computer used in the system and the type (or number) of programs that should always be started on each computer, the details of which will be illustrated later in FIGS.

In the mutual communication check, the communication check paths are set so that there is no duplication. For example, there are five computers, 1 to 5, and No. When 1 is a primary computer, the communication path may be set as shown in FIG. It should be noted that the interval 1-5 can be omitted because it is already inspected when 5 is inspected as a secondary computer. Further, the computer in which the communication error is detected on the way can be excluded from the target of the mutual communication check from that point, so that the increase of the inspection time due to the retry or the like can be prevented.

The general structure, function, and usage of the fault detection program have been described above. It is necessary to change the inspection contents in each computer according to the system configuration and usage. An embodiment in which the fault detection program of the present invention is applied to a distributed LSI production management system will be described below.

The LSI production management system is a system for scheduling the processing schedule (creating a processing plan line table) of each wafer to be introduced into the LSI manufacturing line, grasping the progress status thereof, and accumulating processing data. In this system, the name of each wafer on the line (or lot of multiple wafers), information about processing conditions, conditions of each processing step, name of connected equipment, information about processing conditions / time, etc. , A program for managing various information and creating a schedule (called a line management program), a user interface for giving work instructions to an operator, a device control program for controlling each device and collecting data from the device, a database A plurality of programs, such as a database management program that manages the structure, data, search conditions, and the like, operate while communicating via the communication control program.

In such a system, since a plurality of computers are connected to each other and perform separate device control and data management, the types of programs to be run on each computer are different. For example, a certain computer is a host computer for line management, and a line management program operates. This line management computer also has a terminal function to register information such as equipment used in the system, lots to be put into the line, processing condition names, process names, etc., and the information registered here is stored in each computer. Distributed through the network. At this time, if a computer happens to be down, the information will not be transmitted. After that, even if the computer is booted up, the contents of the registration information (common file) will be corrupted, so when using the registration information, only that computer will result in an error.

Further, the other computer is a computer for controlling a plurality of devices, and a device control program and a database management program which are different for each device always operate. Further, a computer may be a terminal dedicated to the user interface used by the operator. In this way, the programs that should always run differ from computer to computer. Figure 4
Is an example showing such a situation. In FIG.
The shaded programs are always-on programs.

In order to recognize the type of program that should be always started on each computer as described above,
Prepare a "computer information file". This computer information is commonly used by each computer, and is composed of a name of each computer and a list of programs that should run on the computer (should constantly run). This information is
The names of programs to be started may be listed, the process number at startup may be determined for each program, and the process number may be registered, or the program to be started may be specified by a flag or the like. May be. Figure 5
5 shows an example of a "computer information file" of the system of FIG. In this example, the types of programs to be started are designated by flags of 1 (always started) and 0 (not started).

The fault detection program checks the program started at that point in its own computer, and refers to this "computer information file" to check whether the program that should be executed at all times Detect if anything is running. If there is an error, output that information as a file. In the case of a secondary computer, the result is 1
Transfer the file to the next computer. If a program that has a parent-child relationship is also specified, it is possible to detect an error in which only the child process is running even if the parent process is stopped. it can. For example, when the boot process number is registered in the "computer information file", a parent-child relationship can be identified by providing the boot process number of the parent process in an identification field for identifying the parent and child. In the case of designating with a flag as shown in FIG. 5, it is possible to identify the parent and child by predetermining that the parent is 1, the child is 2, and so on.

Further, on the secondary computer side, the "computer information file" sent from the primary computer and the "computer information file" of its own computer (secondary computer) have the same contents before the program start inspection. Check if it is. By doing so, it is possible to prevent an error in the case of recognizing the transmission destination by referring to the "computer information file".

Such a common file (a list table of various names used in the system, a computer information file,
As a method of checking the consistency of information files such as which computer the database of each device is on), there is a method of directly checking whether the file contents match, but in the case of a file with a large amount of information Takes time to check. Therefore, it is easy to perform the method shown in FIG. 6 instead of checking whether or not all the data contents match. At the time of creating / updating each common file, file information such as the creation (updating) date and time of the file and the number of data items is written at the beginning of the file. This information is written when the common file is created on the line management computer side in FIG. The entire file including the information may be distributed on the network. Alternatively, if the file is large and you want to communicate only the newly registered or deleted part instead of sending all the data every time, receive that information and create this file information on the receiving computer. You may do it.

When checking the file consistency, first, the primary computer side reads this file information, adds this file information, and issues an inspection request to the secondary computer. Two
The next computer reads the file information of its own computer and checks whether the contents (update date and number of data items) match those transmitted from the primary computer. The inspection result is output or transferred to the primary computer. If the file contents do not match as a result of the inspection, transfer the file with the newest update date and time to another computer (older date and time), and match the information, to eliminate the system operation error. It can be avoided. In addition, when the data is added but not deleted, it is possible to assume that the one with the larger number of data is correct and transfer the contents to the smaller one. If only one computer is registered each time (assuming that a computer is the host computer and the information on it is correct), the file information of the host computer is compared with the file information of each computer. It is also possible to check. This kind of file content consistency check is not limited to the "computer information file", but the registration contents of various information files such as the lot name table, recipe (processing condition) information file, database structure definition file, etc. It can be used to check whether they match.

Some startup programs and definition files differ depending on the computer. For example, the control program of the device is activated by the same device control program depending on the connected device. (There is also a method to start a program with a different name for each device,
In that case, setting of a communication partner at the time of communication from another program is troublesome, and therefore it is usually better to recognize the difference between the devices by referring to the environment setting table. ) In this case, it is necessary to check the system environment settings such as whether the correct control program for each device is started, whether the communication parameters at startup are correct, and whether the database for each device is set. , It is not enough to simply check the existence of files and the name of the startup program. An example of such an inspection is shown below.

In the device control, it is checked what kind of device is connected to each computer and whether or not there is a contradiction in the contents of the device control-related setting file for setting the communication parameters and the like.

The setting files to be checked and their check contents are as follows.

(1) Device registration file: Registration of devices connected to the own computer (correspondence table of computers and connected devices) This is the basis of the following checks, and each device (number is registered in this file Check whether the following settings are made for each.

(2) SECS communication parameter setting file: Communication parameters with each device of the LSI manufacturing device Normally, SEC is used for communication between the device and the control program.
The S standard is used. Device I during this communication
It is checked whether communication parameters such as D, connection port number, and timeout time are specified, there is no double registration, the connection port name is correct, and there is no registration failure.

(3) Device control program startup setting file: Designation of process ID, option, etc. at the time of startup Whether the process ID, option, etc., when starting a different program for each device are specified, or the corresponding program Check if is installed, etc.

(Check for Necessary Program / Directory / Registered File Existence) Also, as different from device to device, it is necessary to check the database setting environment such as the structure of the database when storing it in the database. This is to check whether or not the database of an originally intended device exists on the designated computer. Furthermore, it is also necessary to check whether the tables (files) showing the correspondences of the databases on each computer match on each computer. This check
This can be checked by using the method shown in FIG.

In addition, by checking the free disk capacity (disk usage rate) of the database, etc., effective use of the disk is achieved by saving some information and changing the connection before the disk is full. be able to. This not only prevents a data storage error due to the disk becoming full, but also prevents an increase in search time or the like due to an increase in data (such as data being concentrated and accumulated in one computer).

As described above, the LS using the computer network
The example in which the fault detection system of the present invention is applied to the I production management system has been described. As described above, the fault detection system of the present invention can be used not only for controlling and monitoring various manufacturing lines using a computer network, but also for various applications as described below.

(1). Entrance / exit control system for buildings (2). Reservation system for various vehicles, lodging facilities, leisure facilities, etc. (3). Various manufacturing / sales management systems

[0051]

As described above, according to the present invention,
Each computer is equipped with a fault detection program and an inspection program that executes various inspections. Each fault detection program activates each inspection program to check the normal operation of the programs that should always be activated in each computer, and between the computers. Since the normal operation check of communication is performed and the contents of the inspection result are received from each computer, the operating states of the programs on all other computers can be simultaneously detected from a given computer which is one terminal. Therefore, it is not necessary to remotely log in to each computer and check each one as usual, and not only can faults be detected quickly and easily, but also oversight of faults can be prevented. In addition, because the consistency of various definition files that are commonly used among computers is also checked, even if it seems to be operating normally, some data in the definition file may be missing and Occasionally, a failure that causes malfunction can be prevented, and the reliability of system operation can be greatly improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a fault detection system according to an exemplary embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an operation procedure of the failure detection system shown in FIG.

FIG. 3 is a diagram showing a method of setting a communication path when performing a computer mutual communication check.

FIG. 4 is a diagram showing a system configuration when the fault detection system of the present invention is applied to an LSI production management system.

5 is a diagram showing an example of a computer information file in the embodiment of FIG.

FIG. 6 is a flowchart showing a procedure for checking file consistency between computers.

[Explanation of symbols]

 1, 2 and 3 computer 11 failure detection program 13 inspection program registration file 15 talker (transmission program) 17 listener (reception program)

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yasushi Wada 1-1-6 Uchisaiwaicho, Chiyoda-ku, Tokyo Nihon Telegraph and Telephone Corporation (72) Masato Mitomi 1-5 Nihonbashi-Horidomecho, Chuo-ku, Tokyo No. 7 in NTT Fanet Systems Co., Ltd.

Claims (2)

[Claims] 1. A network comprising a plurality of computers,
A fault detection system for detecting a fault in a distributed processing system that performs a predetermined process while mutually communicating between a plurality of programs operating on each computer, and a fault detection program and an inspection program for executing various tests on each computer The failure detection program provided in each computer is connected via the communication control program, and the failure detection program provided in each computer activates the inspection program in its own computer, and the communication control program and other Starting the inspection program in other computers through the fault inspection program in the computer, to make sure that the normal operation check of the program that should be always started in each computer and the normal operation check of the communication between the computers are performed. Faults characterized by receiving the contents of the inspection result from each computer Out system. 2. The fault detection program has a function of inspecting whether or not the contents of a definition file commonly used in the distributed processing system match between computers directly or through the inspection program. The fault detection system according to claim 1, wherein: