JP3139536B2 - Distributed batch job processing system and automatic job restart method in the event of failure - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a distributed batch job processing system comprising a plurality of computers each having a batch job processing function, and more particularly to a method for automatically restarting when a failure occurs.

[0002]

2. Description of the Related Art Conventionally, in this type of distributed batch job processing system, automatic re-execution of a job is performed as follows. That is, the distributed batch job processing system includes a plurality of computers each of which executes a batch job, and a re-input computer which schedules and inputs the batch job. Then, when a failure occurs in one computer that executes the batch job, the re-submission computer re-submits the batch job being processed by the failed computer to another non-failed execution computer. By doing so, automatic restart is realized.

An example of such a conventional distributed batch job processing system is disclosed in JP-A-7-175766 (hereinafter, referred to as prior art 1). There is. In the prior art 1, when a failure occurs in the n-th host computer, the job being executed is immediately re-executed by the designated m-th host computer. That is, in the first host computer, the job control language translating means translates the job control language and registers the designation of the m-th host computer for re-executing the job of the n-th host computer in the job management information holding means. I do. The host failure recognition unit recognizes a failure notification from the host monitoring device. The job re-execution preparation unit updates the job control information of the job management information holding unit, and requests re-execution of the job being executed on the failed n-th host computer.
The job scheduler reschedules the job and requests the mth host computer to re-execute the job. The job activating means activates an execution program of the job requested to be executed.

Japanese Patent Application Laid-Open No. Hei 8-227368 (hereinafter referred to as Prior Art 2) discloses that a processing unit in which a failure has occurred can be easily detected, and a job can be efficiently re-executed. A “job re-execution method” is disclosed. In the prior art 2, a log execution control unit and a log file for individually monitoring, at a processing unit level, the configuration and execution state of a batch processing job operating on a computer system; A state display control unit and a display device for displaying the execution state of a unit on a display screen, and a command for designating the job and the process and instructing re-execution from the display screen of the display device, and It includes a re-execution control unit for re-executing the process after the designated process, a process execution control unit, and a job configuration management table.

[0005] Further, Japanese Patent Application Laid-Open No. 2-253441 (hereinafter referred to as Prior Art 3) discloses a "computer system device" which can automatically switch and re-execute when a device failure occurs. Switching system "is disclosed. In the prior art 3, when a device failure occurs during job execution, the device failure receiving unit receives a notification of the failure, and the device restoration impossible determination unit determines whether restoration is possible or impossible. When it is determined that the device can be recovered, the device recovery instruction output means instructs the recovery of the device, the recovery work is performed by this instruction, and the job re-execution means automatically re-executes the job. When it is determined that the apparatus cannot be restored, the automatic apparatus switching means automatically switches the assignment of the job being executed to another apparatus, and the medium mount message output means sets the medium in the switched apparatus. A job is output and a job is performed, and the job re-executing unit automatically re-executes the job being executed. Thus, even if a device failure occurs, the job can be automatically switched and the job can be re-executed without causing an error in the job. If the job can be restored, the job can be automatically re-executed after the restoration. .

[0006]

The conventional (prior art 1) automatic job re-execution method described above has the following problems. In addition to the computer that executes the job, a re-entry computer that performs the re-entry processing is required. If a failure occurs in the re-entry computer that performs the re-entry processing, the restart processing is not performed. It is technically difficult to correctly detect an abnormality in a computer, and there is a risk that a job will be processed twice if an abnormality is detected by mistake.

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a distributed batch job processing system comprising a plurality of computers for performing batch job processing. Of the present invention is to provide a distributed batch job processing system having a simple configuration capable of re-execution.

Note that none of the prior arts 2 and 3 is a distributed batch job processing system including a plurality of computers.

[0009]

According to a first aspect of the present invention, at least one active computer that performs batch job processing in a normal state, and performs processing in place of a failure in the active computer. At least one alternative computer to be performed, failure detection means for detecting a failure of the active computer, and connection switching means for changing a connection from the active computer to the alternative computer when a failure of the active computer occurs A shared disk device, wherein the substitute computer has a job resubmitting means group for taking out information from the shared disk device and resubmitting a job to the substitute computer when a failure occurs in the active computer.
The job re-submitting means group includes the shared disk device.
Job information to retrieve the job information stored in the
Output means and the job information reading means.
The job indicated by the job information
Job re-submission means for re-submission to the computer and submission
Deleted job information from the shared disk device
Distributed batch job processing system according to claim Rukoto a job deletion means is obtained.

According to a second aspect of the present invention, there is provided a distributed batch job processing system having a plurality of computers for performing a batch process, the system comprising a failure detecting means for detecting a failure of the computer, Each of the plurality of computers is connected to a shared disk device that stores job information of the computer, and each of the shared disk devices is a computer that switches jobs from a computer that performs normal batch processing when a failure occurs. In the distributed batch job processing system having a connection switching unit for changing the connection to the server, each of the plurality of computers receives information from the shared disk device when a failure occurs in the computer performing the normal batch processing. To take out the job and submit the job to the alternative computer again E Bei job reintroduction means group, the di
The job re-entry means are shared by another computer.
To retrieve job information stored in a disk drive
Job information reading means, and the job information reading means
The job indicated by the retrieved job information
Job to re-enter the substitute computer
Input means and the input job information to the shared disk device.
Distributed batch job processing system according to claim Rukoto a job deletion means for deleting from above is obtained.

[0011]

The shared disk device is always connected to only one computer. The connection is switched to another predefined computer when the failure detection unit detects a failure in the currently connected computer. Job information is recorded on the shared disk device. During normal operation, the process is performed as it is, and the job information is deleted from the shared disk device when the job execution ends. If a failure occurs on one of the computers executing the job, a normally operating computer retrieves the job information from the shared disk device, re-enters it on its own computer, and simultaneously deletes the job information remaining on the shared disk device. delete.

[0012]

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

Referring to FIG. 1, a distributed batch job processing system according to a first embodiment of the present invention will be described. Note that the program executed by the computer constituting the distributed batch job processing system may be recorded on a recording medium (not shown). Here, "recording medium"
The term "computer" refers to a computer-readable recording medium on which a program is recorded. Specifically, CD-ROM,
Includes magnetic disks such as plexible disks, semiconductor memories, and the like. Further, the recording medium may be paper on which a program is recorded. In this case, the computer
It is only necessary to have a reading device such as an (optical character reading device) and a compiler that translates characters (codes) read by the reading device into a machine language that can be recognized by a computer. Anyway, by installing the program recorded on the recording medium into the computer, the computer can perform a predetermined process.

The illustrated distributed batch job processing system comprises:
First and second computers 1 and 2 that always perform batch job processing, an external storage device 3 connected to the first computer 1, and a shared device connected to the first and second computers 1 and 2. And a disk device 4. This shared disk device 4 is a type of external storage device.

The first computer 1 has a first job processing means group 10-1 and a job re-submission means group 20.
The job processing means group 10-1 is for normally processing a batch job. The job re-submitting means group 20 is the second
When a failure occurs in the computer 2, information is taken out from the shared disk device 4 and a job is input to the job processing means group 10 again.

The second computer 2 is a first job processing means group 10-1 mounted on the first computer 1.
Has the same second job processing means group 10-2.

The external storage device 3 stores the first job information 31-1 input to the first computer 1 and the first job execution generated by the job when the first computer 1 executes the job. Save the result 32-1.

The shared disk device 4 stores the second computer input to the second computer almost in the same manner as the external storage device 3.
And the second job execution result 32-2 generated by the job when the job is executed by the second computer 2. further,
The shared disk device 4 includes a failure detection unit 41 and a connection switching unit 42. The failure detecting means 41 monitors the operations of the first and second computers 1 and 2 and, in the event of an abnormality, instructs the connection switching means 42 to change the switching, and re-executes the job re-entering means group 20. Instruct. The connection switching unit 42 is connected to the first computer 1 and the second
Of the computer 2 so that only one of them can be connected at the same time.

In the first embodiment, the second computer 2 is called an active computer, and the first computer 1 is called an alternative computer.

The first group of job processing means 10-1
Job input means 101-1, first job receiving means 102-1 and first job information recording means 103-1
A first job execution unit 104-1, a first job end processing unit 105-1, and a first operator notification unit 10.
6-1.

The first job input means 101-1 accepts a job input from an operator or the like. The second job receiving unit 102-1 determines whether to actually receive the job input by the first job input unit 101-1. The first job information recording unit 103-1 stores the job received by the first job receiving unit 102-1 in the external storage device 3 as first job information 31-1. The first job execution unit 104-1 retrieves the first job information 31-1 recorded in the external storage device 3 and executes the job. First job end processing means 105-
1 performs a process of deleting the first job information 31-1 stored in the external storage device 3 after finishing the first job execution. The first operator notifying unit 106-1 notifies the operator that the job after the completion of the operation of the first job end processing unit 105-1 has ended. Further, the first operator notifying unit 106-1 is provided with the first job receiving unit 10-1.
The operator is notified that the acceptance of the job has been rejected according to 2-1.

Similarly, the second job processing means group 10-2
Is a second job input unit 101-2, a second job receiving unit 102-2, and a second job information recording unit 10-2.
3-2, a second job execution unit 104-2, a second job end processing unit 105-2, and a second operator notification unit 106-2.

The second job input means 101-2 receives a job input from an operator or the like. The second job receiving unit 102-2 determines whether or not to actually receive the job input by the second job input unit 101-2. The second job information recording unit 103-2 stores the job received by the second job receiving unit 102-2 as the second job information 31-2 in the shared disk device 4. The second job executing unit 104-2 extracts the second job information 31-2 recorded on the shared disk device 4 and executes the job. The second job end processing unit 105-2 performs a process of erasing the second job information 31-2 stored in the shared disk device 4 after the end of the second job execution. Second operator notification means 106-2
Notifies the operator of the completion of the job after the completion of the operation of the second job end processing means 105-2.
Further, the second operator notifying unit 106-2 notifies the operator that the second job receiving unit 102-2 has refused to accept the job.

The job re-submitting unit group 20 includes a job information reading unit 201, a job re-submitting unit 202, and a job deleting unit 203. Job information reading means 20
1 indicates the second job information 31-from the shared disk device 4.
2 is to be read. Job resubmitting means 20
2 corrects the second job information 31-2 read by the job information reading means 201 into appropriate information, and attempts to resubmit the job by the first job receiving means 102-1. The job deletion unit 203 deletes the unnecessary second job information 31-2 by being re-input from the shared disk device 4.

Next, a distributed batch job processing system according to the first embodiment will be described with reference to FIGS.

In a normal state, the first computer 1 and the second computer 2 perform the operation of the flow F3 using the first and second job processing means groups 10-1 and 10-2, respectively. I have. Also, the shared disk device 4
Are connected to the second computer 2 in the initial state.

First, a computer operator inputs a job using the first and second job input means 101-1 and 101-2 (step S31). The input job is inspected for abnormalities in job information such as attributes and authority of the operator (step S32).

If the input of the job is permitted, the first computer 1 records the first job information 31-1 in the external storage device 3 by the first job information storage means 103-1. 2 is the second job information 3
1-2 is recorded in the shared disk device 4 by the second job information storage unit 103-2 (step S33). If the job submission is denied, the first computer 1 uses the first operator notification means 106-1 and the second computer 2 uses the second operator notification means 106 (Step S3)
6).

After the job information is recorded, in the first computer 1, the first job execution means 104-1 executes the job based on the recorded first job information 31-1, and the second computer executes the second job. In the computer 2, based on the recorded second job information 31-2, the second job execution unit 104-2
Executes the job (step S34). At this time, the first computer 1 outputs the execution result to the external storage device 3, and the second computer 2 outputs the execution result to the shared disk device 4.

When the job execution ends, the first job end processing means 105-1 in the first computer 1 deletes the first job information 31-1 in the external storage device 3, and
The second job end processing means 105
-2 is the second job information 31-2 of the shared disk device 4
Is deleted (step S35).

After the job information deletion is completed, the first computer 1 notifies the user of the normal completion of the job execution using the first operator notifying means 106-1, and ends the processing.
Second computer notification means 106-2
To notify that the execution of the job has been completed normally, and terminate the process (step S36).

Next, the operation when a failure occurs in the second computer 2 will be described. First, the shared disk device 4 detects a failure by the failure detecting means 41 (step S11). Next, the failure detection unit 41 instructs the connection switching unit 42 to switch the connection from the second computer 2 to the first computer 1 (step S1).
2).

The failure detecting means 41 further transmits the second job information 31-2 on the shared disk device 4 to the job information reading means 201 on the normally operating first computer 1. An instruction is given to insert the battery (step S13).

Job information reading means 201 receiving the instruction
Is the second job information 31-2 from the shared disk device 4.
Is read (step S21). The job is resubmitted using the read second job information 31-2 using the job resubmitting means 202 (step S22). The job processing after the re-submission is processed by the first job processing means group 10-1 in the same manner as in the normal state (point 2 in flow F3).

Apart from the flow of the job processing, the job information deleting means 42 deletes the second job information 31-2 created by the second computer 2 on the shared disk device 4 in the next step S23.

Next, the operation of the distributed batch job processing system shown in FIG. 1 will be described in detail.

In the initial state, the line of the shared disk device 4 is connected to the second computer 2. First, an operator submits a job A to the second computer 2. This input processing is performed by the second job input unit 101-2 on the second computer 2. It is assumed that the determination of the job attributes and the authority is performed by the second job receiving unit 102-2, and as a result, the job is normally input. Next, the job A is recorded on the shared disk device 4 in the form of the second job information 31-2 by the second job information recording unit 103-2. At this point, the input processing of job A has been completed. Next, the information of job A is extracted by the second job execution unit 104-2,
The job A is executed on the second computer 2. The job A outputs the execution result output file to the second computer 2.
The second job execution result 32-2 is created on the shared disk device 4 which is also an external storage device.

Assume that a failure has occurred in the second computer 2 at this point. Then, the failure detection means 41 on the shared disk device 4 detects the event. The failure detecting unit 41 issues a switching instruction to the connection switching unit 42. In response to this instruction, the connection switching unit 42 stops the connection with the second computer 2 and starts the connection with the first computer 1. Since the connection with the second computer 2 is disconnected, the execution result of the job A operating on the second computer 2 cannot be updated, and the execution of the job is substantially continued. become unable. Also, since the second job information 31-2 cannot be referred to from the second computer 2, a new job is not input.

Further, the failure detecting means 41 is a means for reading job information on the first computer 1 which is operating normally.
01 is instructed to start the operation.

In the first computer 1, the job information reading means 201 transmits the job A from the shared disk device 4.
Of the second job information 31-2. This job A
To the first job receiving means 102 of the first computer 1
Input for -1. Here, it is assumed that the reception processing has been performed normally. The information of the job A is recorded as the first job information 31-1 in the external storage device 3 of the first computer 1 this time. Next, the first job executing means 104-1 of the first computer 1 extracts the information of the job A and executes the job. At this time, job A is shared disk device 4
It is also possible to refer to the previous execution result remaining on the upper part, and it is not impossible to execute the job continuously depending on the implementation of the job. If no reference is made, a completely new job A is executed.

When the execution of the job A is completed, the first job end processing means 105-1 deletes the first job information 31-1 of the job A on the external storage device 3. When the processing of the job A is completed as described above, the operator is notified of the completion of the job execution using the first operator communication unit 106-1.

If the job A is submitted to the first job receiving means 102-1 of the first computer 1, the information of the job A in the second job information 31-2 is incomplete. In some cases, processing of job A may not be possible. In this case, the first job receiving unit 102-1
Detects the abnormality, refuses to accept the job, and notifies the operator using the first operator notification means 106-1.

A distributed batch job processing system according to a second embodiment of the present invention will be described with reference to FIG. In the distributed batch job processing system according to the second embodiment, the external storage device 3 for the first computer 1 in the first embodiment is replaced with a similar shared disk for the second computer 2. The use of the device. Here, the shared disk device for the first computer 1 is denoted by reference numeral 4-1 to be referred to as a first shared disk device, and the shared disk device for the second computer 2 is denoted by reference numeral 4-2. Then, it is referred to as a second shared disk device. Then, the second computer 2 is the first computer 1 in the first embodiment.
And a job re-submitting means group. Here, the first
Reference numeral 20 denotes the job re-submitting means group of the computer 1
-1 is referred to as a first job re-submitting unit group, and the job re-submitting unit group of the second computer 2 is denoted by reference numeral 20-.
2 will be referred to as a second job re-submitting means group. That is, the first computer 1 and the second computer
The configuration of the data 2 is exactly the same.

At this time, the two shared disk devices 4-1
And 4-2 are initially connected to the first shared disk device 4
-1 is connected to the first computer 1, and the second shared disk device 4-2 is connected to the second computer 2. Further, the two shared disk devices 4-1 and 4-2
The fault detecting means 41-1 and 41-2 share information about faults and operate synchronously. In other words , when the first failure detection unit 41-1 of the first shared disk device 4-1 detects a failure, the fact that the failure has been detected is simultaneously notified to the second failure detection unit 41-1.
Second failure detecting means 41- of the shared disk device 4-2 of FIG.
2 is also notified. The reverse is also true.

Next, the operation when a failure occurs in the second computer 2 will be described. The difference from the first embodiment is that not only the second shared disk device 4-2 but also the first shared disk device 4-1 detects a failure. However, since the first shared disk device 4-1 is connected to the first computer 1 by the initial connection, the connection is not changed. Therefore, the operation at this time is similar to that of the first embodiment.

When a failure occurs in the first computer 1, the connection of the first shared disk device 4-1 is switched, and the connection of the second shared disk device 4-2 is not changed. . Thereafter, the first shared disk device 4-1
The second job re-entry means group 20-2 on the second computer 2 is activated by the first failure detection means 41-1 to perform the re-entry processing.

Thus, the first computer 1 and the second computer 1
If any of the computers 2 fails,
Mutual automatic re-execution of jobs can be realized.

The present invention is not limited to the above-described embodiment, and various changes and modifications can be made without departing from the spirit of the present invention. For example, in the above-described embodiment, the case where there are two computers is described. However, the present invention can be similarly applied to a case where there are three or more computers. Further, in the above-described embodiment, the shared disk device includes the failure detection unit. However, a failure detection unit may be provided separately from the shared disk device.

[0049]

As described above, the present invention has the following effects.

The first effect is that, when a failure is detected, the risk of double start of a job can be avoided when the job is resubmitted. The reason is that when a failure is detected,
By forcibly switching the connection between the job information used by the failed computer and the shared disk device that records the job execution results, the jobs running on the failed computer can be executed on the shared disk device. This is because the execution result cannot be updated. Therefore, the execution of the job by the computer in which the failure has occurred is substantially stopped. Further, the failed computer cannot refer to the job information on the shared disk device, and cannot execute a new job.

The second effect is that when the failed computer is recovered or when the shared disk device is reconnected, the computer cannot retrieve job information from the shared disk device. This is to prevent double activation. The reason is that, when the job re-input processing is completed, the job information relating to the extracted job is deleted from the shared disk device used by the failed computer.

The third effect is that it is not necessary to prepare an extra computer for re-submitting a job. The reason is that the job re-input processing is performed by the computer that executes the job.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a distributed batch job processing system according to a first embodiment of the present invention.

FIG. 2 is a flowchart for explaining the operation of the distributed batch job processing system shown in FIG. 1;

FIG. 3 is a block diagram showing a distributed batch job processing system according to a second embodiment of the present invention.

[Explanation of symbols]

1, 2 Computer 3 External storage device 4, 4-1, 4-2 Shared disk device 10-1, 10-2 Job processing means group 20-1, 20-2 Job re-submission means group 31-1, 31-2 Job information 32-1, 32-2 Job execution results 41, 41-1, 41-2 Failure detection means 42, 42-1, 42-2 Connection switching means 101-1 and 101-2 Job input means 102-1, 102-2 Job receiving means 103-1 and 103-2 Job information recording means 104-1, 104-2 Job execution means 105-1, 105-2 Job end processing means 106-1, 106-2 Operator notification means 201 , 201-1, 201-2 Job information reading means 202, 202-1, 202-2 Job re-submission means 203, 203-1, 203-2 Job deleting means

Claims (5)

(57) [Claims] 1. At least one active computer (2) that performs batch job processing during normal operation, and at least one alternative computer (1) that performs processing when the active computer fails. Failure detection means (41) for detecting a failure of the active computer
And a shared disk device (4) having connection switching means (42) for changing a connection from the active computer to the alternative computer when a failure occurs in the active computer. e Bei job reintroduction means group for the retrieving information from the shared disk device when a failure occurs on the working computer performs the insertion of job again to the alternative computer (20), the job cycling means group (20 ) Is the shared disk
Job information for retrieving job information stored in the device
Information reading means (201) and the job information reading means.
The job indicated by the retrieved job information
Job to re-enter the substitute computer
Input means (202), and the input job information
Job deletion means (203) for deleting from the disk device;
Distributed batch job processing system according to claim Rukoto equipped with. 2. An at least one active computer (2) that performs batch job processing in a normal state, and at least one alternative computer (1) that performs processing when a failure occurs in the active computer. A distributed batch job processing system comprising: a shared disk device (4) for storing job information of the active computer; and a failure detection unit (41) for detecting a failure of the active computer. A method of automatically restarting a job when a failure occurs in the distributed batch job processing system, comprising: a connection switching unit (42) configured to change a connection from the active computer to the substitute computer when a failure occurs in the active computer. On the alternative computer, the job information stored in the shared disk device is A step of issuing Ri, in the alternative computer, the step of deleting and re-submit the job indicated by the retrieved job information to the alternative computer, in the alternative computer, a submitted job information from the shared disk device on And a method for automatically restarting a job when a failure occurs in the distributed batch job processing system. 3. A retrieving job information stored in the shared disk device process, a process of re-submit the job indicated by the retrieved job information to an alternate computer, the job information indicating a job that is該再turned A recording medium storing a program for causing the alternative computer to execute the process of deleting from the shared disk device. 4. A distributed batch job processing system having a plurality of computers (1, 2) for performing batch processing, wherein failure detecting means (42-1 and 42-2) for detecting a failure of the computers. And a shared disk device (4-1, 4-2) for storing job information of the computer is connected to each of the plurality of computers, and each of the shared disk devices is In the distributed batch job processing system having connection switching means (42-1 and 42-2) for changing a connection from a computer that performs normal batch processing to a computer that replaces a job, each of the plurality of computers includes: When a failure occurs in the computer that performs the normal batch processing, information is taken out from the shared disk device and the replacement is performed. Job cycling group of means for performing the insertion of job again to computer (20-1, 20-2) Bei example, said job cycling means group (20-1, 20-2) is another
On a shared disk device connected to another computer
Job information reading means for retrieving selected job information
(201-1, 201-2) and the job information reading
The job indicated by the job information retrieved by the column
Job to re-submit the job to the alternative computer
Re-charging means (202-1 and 202-2) and
Job for deleting job information from the shared disk device
Distributed batch job processing system according to claim Rukoto a deletion means (203-1, 203-2). 5. A distributed batch job processing system having a plurality of computers (1, 2) for performing batch processing, wherein failure detecting means (42-1 and 42-2) for detecting a failure of the computers. And a shared disk device (4-1, 4-2) for storing job information of the computer is connected to each of the plurality of computers, and each of the shared disk devices is The method of automatically restarting a job when a failure occurs in the distributed batch job processing system, comprising a connection switching unit (42-1 and 42-2) for changing a connection from a computer that performs normal batch processing to a computer that replaces jobs. A step in which each computer retrieves job information stored in a shared disk device connected to another computer. In each of the computers, a step of re-submitting the job indicated by the extracted job information to the substitute computer, and a step of deleting the input job information from the shared disk device in each of the computers A method for automatically restarting a job when a failure occurs in a distributed batch job processing system.