JP2000222227A - Process management system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the abnormal finish of a system by automatically releasing unnecessary resources held by a process when there is leakage in a resident process. SOLUTION: A monitoring process 11 monitors the resident process 12 of a monitoring object and when the quantity of its occupying resources, a passing time from starting and the number of finished processing's exceed a designated critical value, temporarily finishes the process 12 and restarts it again. Thus, since the resident process is restarted at a proper interval though it generates leakage, there is an effect of releasing unnecessary occupying resources.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a resident process management method, and more particularly to a process management method suitable for managing a server that releases computer resources in a client-server system.

[0002]

2. Description of the Related Art Generally, a process in a computer program performs processing while dynamically acquiring and releasing resources such as a memory from a computer system. However, after a process dynamically acquires and processes resources, it may not be released even though it is no longer needed, and may remain held by the process. Such a leak of releasing resources is called a leak.

[0003] The leaked resources are occupied by the acquired process, and cannot be used by other processes until the process ends. In the case of a resident process that, once activated, does not terminate in principle until a termination request is issued, a new resource is allocated each time it is executed. Typically, this is a server process in a client / server system.

[0004] A server process resides in the system and provides services to clients. Since the server process operates every time there is a request from the client, if the server process is leaking, it acquires new resources one after another without releasing the resources once secured, and eventually uses the resources of the system. Use up.

In such a case, a resource allocation request from another process or its own process to the system fails and terminates abnormally. In some cases, the entire system may be stopped.

[0006] As described above, leak is a serious problem for computer systems, and its countermeasures are important. To solve this problem, as described in, for example, Japanese Patent Application Laid-Open No. 9-244948, a memory management mechanism for debugging provided with a memory allocation table that tracks memory allocation and release detects a memory leak location. , A method of removing it from the program is conceivable.

[0007]

In the prior art, it is up to the program creator to detect and remove leaks of resources such as memory. However, it is very difficult to detect and remove a leak, and in practice, a program often leaves a leak to a user.

An object of the present invention is to release unnecessary resources held by a resident process when there is a leak in the resident process, thereby preventing abnormal termination of the system.

[0009]

The above object is achieved by automatically stopping and restarting a resident process which is leaking.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows an example of the configuration of a client-server system according to the present invention. A client device 14 and a server device 10 are connected by a network 15.
A management table 13 for managing a state of the server process 12 in addition to a server process 12 for performing processing in response to a request from the client device 14, a monitoring process 11 for monitoring and starting and stopping the server process 12,
There is an upper limit value definition table 16 that stores an upper limit of a state that the server process 12 can take. Here, only one client device is shown, but generally there are a plurality of client devices, and a server device provides a plurality of services. Further, a plurality of server processes that provide the same service are activated for each service, and all the server processes are monitored.

FIG. 2 shows an example of the configuration of the management table 13. For each server process monitored by the monitoring process 11, the process number 20 of the server process and the server process among various services provided by the server device are shown. A service type 30 for identifying which service is provided, an execution counter 22 for indicating the number of processes currently being executed by the server process, an accumulation counter 23 for indicating the total number of processes executed by the server process up to now, A start time 24 indicating the start time and an end request flag 25 set when the monitoring process requests the server process to stop the server process are held as a set of information.

The termination request flag is set to "1" when requesting termination of the server process, and "0" when not requesting termination. Further, in this embodiment, it is assumed that the server process operates in a multi-thread and simultaneously processes requests from a plurality of clients.

FIG. 3 shows an example of the configuration of the upper limit value definition table 16. For each service type, a memory occupation amount 31 indicating the upper limit of the memory that can be occupied by the server process that provides the service, An operation time 32 indicating a continuously operable time and a processing number 33 indicating the number of requests that can be continuously processed by the server process are held as a set of information. If the server process exceeds any of these upper limits, the monitoring process 11 restarts the server process.

FIG. 4 shows an example of a processing flow of a monitoring process according to the present invention. Here, the process to be monitored is 1
Two cases are assumed. For each service type, the service provider:-Memory occupancy: automatically restarts when the memory capacity occupied by the server process exceeds a certain value.-Operating time: restarts automatically after the server process has run for a certain time. Yes-Number of processes: Defines the upper limit of "Restart automatically after the server process processes a certain number of requests" and starts the monitoring process 11. The monitoring process 11 stores the defined upper limit value in the upper limit value definition table 16 for each service type (step 400),
The server process 12 is started (Step 401), the management table is initialized (Step 402), and monitoring of the server process is started.

In the management table, the process number of the started server process, the server type, and the start time are written.
The execution counter, the accumulation counter, and the end request flag are each initialized to “0”. The monitoring process 11 periodically checks the server process 12 and the management table 13 from the start of monitoring to the end of monitoring.

First, the activation status of the server process 12 is checked (step 403). If the server process 12 is stopped, the information of the server process 12 is deleted from the management table 13 (step 404), and the same type of service is performed (that is, the service type). Start the process (Step 401)
And continue monitoring. Further, the service type 21 is acquired from the management table 13, the upper limit value of the server process 12 is referred to from the entry of the same service type in the upper limit value definition table 16, and is compared with the current state of the server process 12.

The result is that the occupied memory capacity of the server process 12 exceeds the defined occupied memory capacity limit (step 405).

The elapsed time from the start time 24 to the current time in the management table 13 exceeds the defined continuous execution time limit (step 406).

A cumulative counter 23 in the management table 13
Exceeds the limit value of the cumulative number of processes defined (step 407).

Monitoring process 1
If the end of 1 is requested by the service provider (step 408), the execution counter 22 in the management table 13 is checked (step 409), and a value other than "0" is set (request is being processed). In this case, "1" is set to the end request flag 25 in the management table 13, and monitoring of the server process is continued (step 410).

If "0" is set in the execution counter 22, the server process 12 is stopped (step 411).
If the service provider requests that the monitoring process 11 be terminated (step 412), the monitoring process 1
1 (step 413), and if not requested, monitoring of the server process is continued.

FIG. 5 shows an example of the processing flow of the server process in the present invention. The server process starts a service, and upon receiving a request from a client (step 502), performs a thread generation process (step 503). The created thread first searches for the process number of its own process from the management table 13.

Hereinafter, all references and updates of the management table 13 are made to entries of the own process. Next, the end request flag 25 in the management table 13 is checked (step 504). If the end request flag 25 is set, the request is rejected (step 512).
Then, the execution counter 22 in the management table 13 is changed to "
If "0", the process ends (step 511),
If it is not "0", the thread end processing (step 50)
Perform 9).

If the end request flag 25 is not set, the execution counter 22 and the accumulation counter 23 in the management table 13 are counted up (step 50).
5) Then, the request is processed (step 506).
When the processing of the request is completed, the execution counter 22 is counted down (step 507), and the management table 13
Is checked (step 508).

If the end request flag 25 is set, the execution counter 22 of the management table 13 is further checked (step 510). If "0", the process is ended (step 511). If there is, the thread is terminated (step 509). If the end request flag 25 has not been set, the thread is ended (step 509).

According to this embodiment, even if the server process has a memory leak, it is restarted when the occupied memory capacity exceeds the defined value, and unnecessary memory is released. In addition, the server is periodically restarted with respect to the leak of computer resources other than the memory, so that unnecessary resources are released. As described above, abnormal termination of the system can be prevented.

[0028]

According to the present invention, even if a resident process has leaked, it is restarted at an appropriate interval, so that there is an effect that unnecessary occupied resources are released.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration example of a client / server system according to the present invention.

FIG. 2 is a diagram showing an example of the contents of a management table.

FIG. 3 is a diagram showing an example of the contents of an upper limit value definition table.

FIG. 4 is a flowchart showing processing contents of a program for managing a server process in the server device.

FIG. 5 is a flowchart showing processing contents of a server process for providing a service to a client in the server device.

[Explanation of symbols]

10 server device, 11 monitoring process, 12 server process, 13 management table, 14 client device, 15 network, 16 upper limit value definition table.

Claims (1)

[Claims] A resident process that does not release computer resources acquired from the system after it is no longer needed.
The amount of resources occupied by this process has exceeded the specified value, the elapsed time since the start of the resident process has exceeded the specified value, and the number of processes that have completed execution has exceeded the specified limit A process management method characterized in that a resident process is temporarily terminated and then restarted to release unnecessary computer resources of the resident process.