JPH0210437A - Management method for timer request - Google Patents

Abstract

PURPOSE:To accurately forecast the cancel occurrence time at a high speed by managing a timer request with a queue generated in the order of timer cancel occurrence. CONSTITUTION:When receiving a timer cancel request, a system 13 outputs a cancel reception completion signal. Then, a timer cancel processing part 14 gives the cancel reception completion signal to a process 10 and turns off a pertinent queue of a management table 15. Log information is gathered and gathered data is written in a log buffer 16. Data written in the log buffer 16 is sent to a statistical information part 12. A timer request processing part 11 totalizes data for each request source and each processing classification to statistically obtain the time, when a cancel request may occur, and manages queues of the management table 15 by the queue generated in the order of timer cancel request occurrence.

Description

[Detailed Description of the Invention] [Summary] A timeout notification request (
Regarding the method of managing timer requests (hereinafter simply referred to as timer requests), we have prepared a management table that determines the order (queuing) of timeout notification requests in order to more accurately predict the time when cancellation will occur as a system. After the process sends a timeout notification request to the system, it receives an acceptance completion signal from the system, and then after the process completes the specified processing, it sends a timer cancellation request to the system, and the system responds. When the cancellation reception completion signal is received, log information such as the timer value and actual processing time is collected, and based on the collected log information, the time when the cancellation request is expected to occur is statistically determined, and the time when the cancellation request is expected to occur is calculated. are managed in a queue generated in the order in which timer cancellation requests occur.

[Industrial Application Field 1] The present invention relates to a timer request management method when a process requests a timeout notification (hereinafter simply referred to as a timer request) to a system.

The process makes various timer requests to the system. In this case, this system is required to process various timer requests in the most efficient manner.

[Prior Art] When a process executes a certain type of task processing, the process first requests a timer to the system. When the timer in the system receives the request, it sends an acceptance completion signal to the process.

The process side receives an acceptance completion signal from the system,
A predetermined process is started, and after the process is finished, a timer cancellation request is made to a timer in the system. The timer counts a predetermined time, and when the count ends, the system side sends a timeout signal to the process side. On the process side, the predetermined task processing must be completed before this timeout signal is input. In other words, if the process is normal, it is necessary to request timer cancellation before the timeout signal is generated. If the process has not finished when the timeout signal is input to the process, abnormal processing is performed.

When making a timer request here, a table for timer management is provided. Here, the queuing method is to find the time when timer cancellation (timer cancellation request signal after task processing is completed) occurs in response to a timer request,
Buffers are linked in that order.

FIG. 8 is an explanatory diagram of how to determine timer cancellation occurrence time using a conventional method. Management buffer 1 contains timer 1.
0. The timer value, actual location time ratio, timer maximum cell generation time, etc. are stored. The time calculation unit 2 calculates the timer cancellation occurrence time from the timer value of the management buffer 1 and the actual time ratio j, and stores it in the timer cancellation occurrence time area of the management buffer 1. The timer cancellation occurrence time is calculated by adding the actual required time to the current time. The timer cell generation time is calculated using the following formula: Current time + (timer value) / (actual required time percentage) where:
The timer value is the time until the time limit for task processing, and the actual required time is the ratio of the actual required time to the requested timer value. Note that the timer value and the ratio of the actual required time are given as parameters from the process requester.

On the other hand, the management table 3 stores the first buffer address to the last buffer address of the queue for timeout notification requests, and is stored in the order of timer requests. Based on the timer cancellation occurrence time of management buffer 1, the management table buffer? When the desired buffer is found, queuing is performed and the queue is turned off.

[Problems to be Solved by the Invention] The ratio of the actual required time described above is a parameter from the requester and depends on how the requester specifies it. Therefore,
The time at which timer 1 uncell occurs is precisely the time at which timer cancellation is predicted to occur, and the order in which the timer request queue is generated based on this cannot be said to be sufficiently reliable. . In this way, if the time at which timer cancellation occurs is held as a fixed value, timer cancellation does not necessarily occur in the order in which the queues are arranged, making queue searches inefficient.

The present invention has been made in view of such problems, and is a more accurate key! ? The purpose of this invention is to provide a timer request management method that allows the system to predict the time when a cell occurs.

[Means for Solving the Problem] FIG. 1 is a flowchart showing the principle of the method of the present invention. When a process requests a timeout notification to the system, the present invention prepares a management table that determines the order (queuing) of timeout notification requests.Step 1) The process sends a timeout notification request to the system. After that, the process receives the acceptance completion signal responded from the system (step 2), and then sends a timer cancellation request to the system after the process completes the predetermined processing.Step 3) receives the cancellation acceptance completion signal responded from the system. When received, log information such as timer values and actual processing time is collected (Step 4), and based on the collected log information, the time at which a cancellation request is expected to occur is statistically determined (Step 5).

The present invention is characterized in that the queues in the management table are managed using queues generated in the order in which timer cancellation requests are generated (step 6).

[Operation] Manage timer requests in a queue generated in the order in which timer cancellations occur. This makes it possible to provide a timer request management method that allows the system to predict the time at which cancellation will occur more accurately and faster than conventional methods.

[Example] Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

The general sequence when a process uses a timer is to determine a time limit for waiting for the result of a process (let's say A) and request a timeout notification. The time limit at that time depends on the content of processing A that the process attempts to perform, and is set to a value obtained by adding α to the time required for the actual processing (in other words, a value multiplied by multiple times).

The method of the present invention collects 1a obtained from the parameters when a process requests a timeout notification request, the actual processing time, etc. as log information, statistically calculates the percentage of the actual time required, and calculates the timer cancellation occurrence time. This reduces calculation errors.

FIG. 2 is a diagram showing an example of the sequence of the method of the present invention. When an event occurs in the process 10, the process 10 issues a timeout notification request (timer request). The timer request processing unit 11 receives this request, and the statistical information unit 1
If there is no information in the system 13, a timer request is made to the system 13.

Upon receiving the reception completion signal from the system 13, the timer request processing unit 11 transmits the reception completion signal to the process 10. In the process 10, a timer cancellation request is issued to the timer cancellation processing unit 14 at the same time as the 98th process Δ ends. The timer cancellation processing section 14 transmits this request to the system 13.

When the system 13 receives a timer cancellation request, it outputs a cancellation acceptance completion signal. The timer cancellation processing unit 14 transmits the cancellation reception completion signal to the process 10.
At the same time, the corresponding queue in the management table 15 is turned off. Also, log information is collected and the log buffer 16
Write the collected data to. The figure shows the requester identifier and processing type.

An example is shown in which parameters such as a timer value, actual processing time, and number of requests are written. Logbuff? The data written in 16 is sent to statistical information section 12.

When the number of requests exceeds a statistically significant number, the timer request processing unit 11 aggregates data by request source and type of processing, determines and stores a standard value.

FIG. 3 is an explanatory diagram of standard value determination. The horizontal axis is the percentage, and the vertical axis is the number of requests made by the requester. For example, the center of the distribution is set as the standard value for each process Δ and process 8. This standard value is taken as the actual required time ratio in equation (1). Then, when there is a next timeout notification request (=1), the system 13 is given a value determined from the result of the log information, not a parameter, and the cancellation occurrence time is determined and queued.

Thereafter, the standard value is updated at a certain period (number of hours or number of counts). Thereby, the time of occurrence of cancellation can be determined more accurately. The timer request processing unit 11 then manages the queues in the management tape file 15 in the order in which timer cancellations occur.

FIG. 4 is a diagram showing an example of a system configuration for implementing the method of the present invention. Components that are the same as those in FIG. 2 are designated by the same reference numerals. In the figure, reference numeral 20 denotes a timer control section, which includes the above-mentioned timer request processing section 11. In addition to the timer cancellation processing section 14, a timeout processing section 17 and an actual required time ratio calculation section 1
It is composed of 8. The actual required time ratio calculation unit 18
It is activated by an initial program loader (IPL) 30. The professional wrestling 1o is a timer requesting section 10a.

The timer request unit 10a issues a timer request to the timer request processing unit 11, and the timer cancellation request unit 101) issues a timer cancel request to the timer cancellation processing unit 14. do.

Here, the logging of the actual required time is collected by the timer cancellation processing unit 14 as described above. On the other hand, the logic for calculating the actual processing time ratio based on the collected data is performed in a separate process, that is, the actual processing time ratio calculation unit 18, and a table for it is generated.

The timing of activation is based on the extension of the IPL location.

The timer request processing unit 11 reads the actual processing time ratio from the table created there, and uses that value when the next timeout notification request is made.

FIG. 5 is a flowchart showing the operation of the timer cancellation processing section 14 (see FIG. 4). Here, a timer ID is used as a parameter.

First, it is checked whether the requested cancellation has already timed out (step 1).

If so, the process ends with an error. If the timeout has not been completed, the current time (cancellation completion time) is determined (step 2), and the actual processing time is determined. Here, the actual processing time can be calculated using the following equation.

Killen cell completion time - timeout notification request time Once the actual processing time is determined in step 3, collect log information (step 4), count the number of requests (step 5),
Since the corresponding buffer has already been used, remove it from the queue (Step 6), and release the timer ID (Step 7).
.

FIG. 6 is a flowchart showing the operation of the actual processing time ratio calculating section 18 (see FIG. 4). First, check whether you have a statistically sufficient amount of data (step 1).
If not, return.

If there is, the log information is read (step 2) and the information is divided for each request source (step 3). Next, it is checked in step 4) whether the calculation to calculate the standard value has been performed for all requesters, and if it has been performed, the process returns.

If not, the information is divided by processing content (step 5), and for the same requestor, it is checked whether standard values have been obtained for all processing (step 6).

If standard values for all processes are required, step 4
Returning to step 7, if it has not been determined, the actual time required for one timer cancellation process is calculated (step 7). Here, the actual required time ratio is determined using the following formula.

Actual processing time/timer value Then, a count-up (totalization of data) is performed (step 8), and it is checked whether all data have been totaled within the same processing content (step 9).

If all data has not been aggregated, return to step 7.
If all data has been aggregated, determine the standard value (Step 10), and store the determined standard value (Step 11).
, clear the log information (step 12), and return to step 6.

FIG. 7 is a flowchart showing the operation of the timer request processing section 11 (see FIG. 4). First, determine the timer ID number (step 1), then check whether there is statistical information (step 2).

If there is, check whether there is information about the requester (step 3), and if there is, check whether there is information about the specified process (step 4).
). If there is, the actual required time ratio determined from the statistical information is used (step 5). If the check result in steps 2 to 4 is No, the actual required time ratio of the parameters specified by the requester is used (step 6).

In addition, in steps 2 to 4, we check what process the process that requested the timeout notification is trying to use the timer for, and we find statistical information (actual time required percentage) about that process. This is a search to find out whether or not.

Next, using the determined actual required time ratio, the timer cancellation occurrence time is determined by equation (1) (step 7), and a management buffer is set (step 8). Here, the buffer is acquired by a process that requests processing to timer control. Finally, queuing processing is performed (step 9).

[Effects of the Invention] As explained in detail above, according to the present invention, by managing timer requests in a queue generated in the order in which timer cancellations occur, the system can more accurately and quickly predict the time when cancellations will occur. It is possible to provide a method for managing timer requests.

[Brief explanation of the drawing]

Figure 1 is a flowchart showing the principle of the method of the present invention, Figure 2 is a diagram showing an example of the sequence of the method of the present invention, Figure 3 is an explanatory diagram of standard value determination, and Figure 4 is a system implementing the method of the present invention. Figure 5 is a flowchart showing the operation of the timer cell processing unit; Figure 6 is a flowchart showing the operation of the actual processing time ratio calculation unit; Figure 7 is a flowchart showing the operation of the timer request processing unit. Flowchart FIG. 8 is an explanatory diagram of how to find the timer cancellation time using the conventional method. In FIG. 4, 10 is a process, 10a is a timer request unit, 10b is a timer cancellation request unit, 11 is a timer request processing unit, 13a is a timer, 14 is a timer cancellation processing unit, 17 is a timer 1~processing unit, 1 day 3 is an actual processing time ratio calculation unit, and 30 is an IPL. Patent applicant: Fujitsu Limited
Company representative Patent attorney Ijima
Flowchart showing the origin of the method of the present invention 1 Figure 3 Figure 3 The method of the present invention ■ Sequence 1 Figure 2 System configuration for carrying out the method of the present invention Diagram 4 showing an example Flowchart showing the operation of the timer cancel processing unit 0

Claims (1)

[Claims] When a process requests a timeout notification to the system, a management table is prepared that determines the order (queuing) of timeout notification requests (step 1), and the process sends a timeout notification to the system. After sending the request,
Receive the reception completion signal responded from the system (step 2), then send a timer cancellation request to the system at the same time as the process finishes the specified process (step 3), and receive the cancellation reception completion signal responded from the system. Once received, log information such as the timer value and actual processing time is collected (step 4), and based on the collected log information, the time at which the cancellation request is expected to occur is statistically determined (step 5), and the information in the management table is A timer request management method characterized in that the queue is managed in a queue generated in the order in which timer cancellation requests occur (step 6).