JPH0359730A - Job abort control method - Google Patents

Abstract

PURPOSE:To prevent a normal job from being illegally aborted by extending the reference abort time of a job in accordance with the variation of system load. CONSTITUTION:When the system load is a reference value indicating the average confusion degree of respective apparatuses, a program inputs an equation for finding out the time required for processing from table 10. A reference ETIME calculating equation 11 is registered in the table in each program name and the equation 11 is previously formed based upon the statistics of effects and registered in the table 10. The system load is measured at a fixed interval during the execution of a batch job, and when the system load is higher than the reference value, the canceling time is extended. Even if the job processing time is extended due to the increment of the system load, the job can be prevented from being illegally aborted.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a job abort control method for optimally aborting jobs in response to system load fluctuations in a computer system that multi-processes batch jobs.

[Conventional technology]

Conventionally, in order to deal with abnormalities in cases where a batch job uses a device other than the CPU for a long time, job abort is provided based on the time required from the start to the end of the job (hereinafter referred to as ETIME). .

This ETIME termination setting is set by the user, who estimates the approximate value and specifies it in the job control statement (JCL) when registering the batch job to the system.Usually, users of computer systems that process multiple batch jobs Since it is not possible to grasp the level of congestion, that is, the system load, it is not possible to set ETIME termination in consideration of the system load.

Therefore, even if the same ETIME abort value is specified for the same job, if the system load is heavy, the job may be aborted even if there is no abnormality in the job itself.

For this reason, the current situation is that users either set the ETIME cutoff value to a very long value or do not set it at all.

For this reason, even when an abnormality occurred where the ETIME was several times longer, it was assumed that it was due to a heavy system load, and the investigation into the cause was delayed for nearly a year.

[Problem to be solved by the invention]

The above-mentioned conventional technology does not take into account the fact that the time required to process a job changes depending on the system load, and there is a problem that the job may be aborted even if the job is being processed normally if the system load is large. Ta.

An object of the present invention is to provide an optimal job termination control method in response to irregularly changing system loads.

As a result, if job abort occurs even though job abort is performed in accordance with the system load, it can be recognized that the job processing time has been extended due to a cause other than the system load.

[Means to solve the problem]

In order to achieve the above purpose, the system load monitoring program measures the system load at regular intervals during job execution, and if the system load is higher than the standard value, the cutoff time is updated according to the degree of load. This is what I did.

[For production]

The job abort control method of the present invention measures the system load at regular intervals during batch job execution, and extends the abort time if the system load is higher than a standard value.

This prevents the job from being unduly terminated even if the job processing time is extended due to a heavy system load.

Further, for a job whose processing time is extended due to a reason other than a large system load, the job is aborted because the abort time is not extended.

〔Example〕

Hereinafter, one embodiment of the present invention will be described in detail with reference to the drawings.

FIG. 1 shows, in the form of a problem analysis diagram, the specific configuration and operation of the process for setting the job termination time.

This problem analysis diagram is generally called Problem Ana
It is called lysisDiagraIII or PAD for short.

Usually, a job is made up of what are called job steps that correspond to programs. Processes 1 to 7 in FIG. 1 are repeated for each job step.

The process finds the first job step from the JCL in which the user has registered the job, and reads the program name, data type, and number of data.

In process 2, when the system load is a standard value indicating the average congestion of each device, a formula for determining the time required for the program to process is read from the table 10 shown in FIG. 2.

This table 10 shows the standard ET for each program name.
IME calculation formula 11 is registered. The formula is created in advance based on actual statistics and registered in this table 10.

Process 3 is the standard ETIME calculation formula 1 that reads the data type and number read from the JCL from table 1o.
The value is calculated by substituting it into 1.

In process 4, program processing time is system load.In this table 10, a system load influence coefficient 12 is registered for each program name. The value of the system load influence coefficient is determined statistically from actual results in advance. They are stored in a table for each job multiplex number and input/output device busy rate.

Processes 5 and 6 are repeated at regular intervals until the program process is completed or aborted. In FIG. 1, this interval is t seconds.

In process 5, the average value of the system load for 5 seconds is investigated using the existing "system operation status monitoring program" subroutine.

Processing 6 is based on ETIM, depending on the size of the system load.
E is being extended. The calculation method is that the remaining ETIME after 1 second is ETIME-t, and to this, the job multiplex number influence B and the input/output device busy rate influence in the system load influence coefficient 12 newly read from Table E0 in Figure 2 are calculated. Add the sum of degrees M multiplied by t seconds and use it as a new ETIME. That is, if the system load is larger than the standard value, ETIME is extended by the amount of influence of the system load on job processing time.

When the calculation in process 6 is completed, the process proceeds to process 7. In process 7, ETI
Determine whether ME has become below O.

If it is less than O, the process proceeds to process 8 and the job step is aborted. If it is not below O, return to process 5.

After this job step is completed, the process returns to process 1 in FIG. 1, and the next job step is processed in the same way.

Note that the average value of the system load for t seconds is
... Average Job Multiplicity [Effect of the Invention] According to the present invention, by extending the standard job termination time in accordance with fluctuations in the system load, it is possible to prevent normal jobs from being unduly terminated.

Therefore, even if the abort time has been extended in response to changes in the system load, if a job is aborted beyond the abort time, there may be an abnormality in which the job processing time is extended due to reasons other than the system load. Recognize.

[Brief explanation of drawings]

FIG. 1 is a PAD showing the configuration and processing flow of an embodiment of the present invention, and FIG. 2 is a table containing data necessary for job termination control of the present invention. Octopus

Claims (1)

[Claims] In a computer system that multi-processes batch jobs,
A job abort control method characterized by updating abort based on elapsed time of a batch job in accordance with changes in system load.