JP5986974B2 - Process monitoring system and process monitoring method - Google Patents

Description

  The present invention relates to a process monitoring technique in which when a process on a computer system is terminated for some reason, the termination process is restarted to ensure normal operation of the system.

  There is a need for process monitoring technology that monitors a process on a computer system and captures this event and restarts the terminated process when the process is terminated due to a malfunction or terminated by a malicious third party. It is very expensive because various usage forms such as normal operation management of the entire system and elimination of an obstruction to the system by a malicious third party are assumed (see Non-Patent Documents 1 and 2).

Nozomi Anjo, “PostgreSQL Multifunctional Middleware pgpool-II Latest Usage Method”, February 12, 2013, SRA OSS, Inc. Japan, [October 9, 2013 Search], Internet <URL: http: // www.sraoss.co.jp/event_seminar/2013/20130212_pgpool_seminar_sraoss.pdf> Masato Kitamura, “UNIX / NT Server Strategy for Ensuring High Reliability”, June 2001, CAC Technical Report Magazine “SOFTECHS” Vol. 24 No. 1. [Search October 9, 2013], Internet <URL: http://www.cac.co.jp/softechs/pdf/st2401_05.pdf>

In the conventional process monitoring technology, there is a problem that a monitoring-dedicated program must be incorporated into a monitoring target program (hereinafter referred to as a monitoring target program), and the monitoring target program needs to be greatly modified.
Normally, if a process simply creates another process, the process has a parent-child relationship. In this case, however, the entire process tree is terminated according to an expression that ends when the parent process is terminated. There was also a problem that the monitoring and restarting of the process could not be performed and the monitoring could not be performed.

  The present invention provides a process monitoring technique that can prevent monitoring from being impossible while having little influence on a monitoring target program.

  The process monitoring technology of the present invention is a process monitoring technology for monitoring the life and death of each process by a plurality of monitoring target programs, and each of the plurality of process monitoring means corresponds to a corresponding monitoring target process among the plurality of monitoring target programs. Each process monitoring means monitors the operating status of at least one process monitoring means other than itself among the plurality of process monitoring means, and a plurality of process monitoring means Monitoring among a plurality of process monitoring means is configured so that the operation status of the self can be monitored by at least one process monitoring means other than itself. Each process monitoring unit stores process information in the storage unit when the process monitoring unit is activated, and each process monitoring unit displays the operation status of the process monitoring unit monitored by the process monitoring unit. Monitor. Furthermore, each process monitoring unit refers to the process information registered in the storage unit so that it is not included in the same process tree as the process monitoring unit when it is detected that the process monitoring unit to be monitored is stopped. Then, the monitoring target program corresponding to the stopped process monitoring means is executed.

  Alternatively, the process monitoring technique of the present invention is a process monitoring technique for monitoring the life and death of each process by a plurality of monitoring target programs, and each of the plurality of process monitoring means is a program separated from each monitoring target program. Each process monitoring means monitors the operating status of at least one process monitoring means other than itself among the plurality of process monitoring means, and at least one other than itself among the plurality of process monitoring means. Monitoring between a plurality of process monitoring means is configured so that the operation status of one process monitoring means can be monitored. Each process monitoring unit stores the process information of each process by the process monitoring unit and the monitoring target program that is the monitoring target of the process monitoring unit in the storage unit when the process monitoring unit is activated. Each process monitoring unit monitors the operation status of the process monitoring unit that is the monitoring target of the process monitoring unit and the life and death of the process by the monitoring target program that is the monitoring target of the process monitoring unit. Further, each process monitoring means refers to the process information registered in the storage means when the process stoppage by the process monitoring means or the monitoring target program that is the monitoring target is detected, and the same process as the process monitoring means. Restart the process by the stopped process monitoring means or the monitored program so that it is not included in the tree.

  According to the present invention, the process monitoring means is realized by a dynamic link library incorporated in the monitored process or a program independent of the monitored process, and the restart of the stopped process is included in the same process tree as the starting process. Therefore, the monitoring target program can be prevented from becoming unmonitorable although the influence on the monitoring target program is small.

The figure which shows the system configuration example of 1st Embodiment. The figure explaining division of the parent-child relationship of a process. The figure which shows the system configuration example of 2nd Embodiment.

<First Embodiment>
The process monitoring system 100 of the first embodiment includes N monitoring target programs 130 ₁ , 130 ₂ ,..., 130 _N−1 , 130 _N (where N is a predetermined integer equal to or greater than 2). .., 120 _N−1 , 120 _{N. The system} includes a registry 110 serving as storage means and N process monitoring means 120 ₁ , 120 ₂ ,. N process monitoring means 120 ₁ , 120 ₂ ,..., 120 _N−1 , 120 _N are each configured to be able to access the registry 110.

  The term “registry” in this specification does not assume a specific operating system, but generally refers to a storage area used by an application.

Process monitoring unit _{120 1, 120 2, ...,} 120 N-1, 120 each _N, the process monitoring means _{120 1, 120 2, ...,} 120 N-1, 120 at least one process monitoring means other than the self of the _N , And at least one process monitoring means other than itself among the process monitoring means 120 ₁ , 120 ₂ ,..., 120 _N−1 , 120 _N is monitored. Monitoring between the process monitoring means 120 ₁ , 120 ₂ ,..., 120 _N−1 , 120 _N is configured.
That is, the arbitrary process monitoring means 120 _i (i∈ {1, 2,..., N−1, N}) has at least one process monitoring means 120 _j (j∈ {1, 2,..., N−1). N}-{i}) and the self (ie process) from at least one process monitoring means 120 _k (k∈ {1,2,..., N-1, N}-{i}) The operating status of the monitoring means 120 _i ) is monitored.

In the example shown in FIG. 1, cyclic permutation of the subscript 120 assigned to identify a process monitoring unit between any process monitoring unit and the process monitoring unit to be monitored by the process monitoring unit. Since (N N-1... 2 1) is established, the monitoring between the plurality of process monitoring means is configured in a ring shape (or may be referred to as a circulation shape). That is, in this example, the process monitoring means 120 _N monitors the operation status of the process monitoring means 120 _N-1, the process monitoring means 120 _N-1 monitors the operation status of the process monitoring means 120 _N-2, · · -, process monitoring unit 120 ₂ monitors the operation status of the process monitoring means 120 _1, the process monitoring means 120 ₁ monitors the operation status of the process monitoring means 120 _N.

Process monitoring unit 120 _1, 120 _2, ..., respectively 120 _N-1, 120 _N, N-number of the monitoring target program 130 _1, 130 _2, ..., the corresponding monitoring target program among the 130 _N-1, 130 _N These programs are implemented as embedded programs, and these programs are each configured as a dynamic link library. In the example shown in FIG. 1, any process monitoring means 120 _i (i∈ {1, 2,..., N−1, N}) is realized by a dynamic link library incorporated in the monitoring target program 130 _i . By configuring the program for realizing the process monitoring means as a dynamic link library in this manner, the monitoring target program can be modified only by adding a code for specifying the library to be used. Can be easily realized.

  Note that the term “dynamic link library” in this specification does not presuppose a specific operating system, but generally refers to a library using dynamic links.

Arbitrary process monitoring means 120 _i (i∈ {1, 2,..., N−1, N}) includes process information registration means 121 _i , process stop detection means 122 _i , and independent process restarting means 123 _i . Including.

The process information registration unit 121 _i (i∈ {1, 2,..., N−1, N}) stores the process information of the monitoring target program 130 _{i in} the registry 110 when the process monitoring unit 120 _i is activated. Specific examples of process information include a process ID and an execution file path. In addition, it is assumed that a continuous registry number starting from 1 is assigned to each process monitoring means in the order of registry registration (in this specification, the subscript of reference numeral 120 assigned to identify the process monitoring means is a registry number. To match).

Process stop detection means _{122 i (i∈ {1,2, ...} , N-1, N}) , the process monitoring means 120 _j (j∈ {1,2 is monitored process monitoring unit 120 _i, ..., N-1, N}-{i}) is monitored. For example, if the cyclic monitoring configuration shown in FIG. 1, a process stop detection means 122 _i, the process monitoring means 120 _i-1 is monitored process monitoring unit 120 _i (However, in the case of i = 1 Process The operation status of the monitoring means 120 _N ) is monitored. Specific examples of this monitoring method include a method of acquiring a process handle to be monitored and receiving a signal indicating that the process has stopped, a method of monitoring an event released by the operating system based on a semaphore, and the like.

Independent process restart means _{123 i (i∈ {1,2, ...} , N-1, N}) , the process stops detecting means 122 _i is monitored processes monitoring means 120 _j (j∈ {1,2, ... , N-1, N}-{i}) is detected so that the process information registered in the registry 110 is not included in the same process tree as the process monitoring unit 120 _i . The monitoring target program 130 _j corresponding to the stopped process monitoring unit 120 _j is executed to restart the process by the monitoring target program 130 _j . For example, in the case of the annular monitoring configuration shown in FIG. 1, the independent process restarting means 123 _i is configured so that the process stop detecting means 122 _i is the process monitoring means 120 _{i-1 to be} monitored (provided that i = 1). When the stop of the process monitoring unit 120 _N is detected, the stopped process monitoring is performed so that it is not included in the same process tree as the process monitoring unit 120 _i by referring to the process information registered in the registry 110. run the monitored program 130 _i-1 corresponding to the unit 120 _i-1, when to restart the process by the monitoring target program 130 _i-1 (monitored program 130 _i-1 is executed, the process monitoring means 120 _{i-1 is} also activated).
Here, “so as not to be included in the same process tree” means that the parent-child relationship between the process monitoring unit 120 _i and the restarted process monitoring unit 120 _i-1 is divided. For example, it can be realized by a so-called double fork (see FIG. 2).
In other words, the independent process restarting means 123 _i (i∈ {1, 2,..., N-1, N}) is called when the process stop detecting means detects the end of the monitoring target, and generates (re-starts) a process that has ended. After starting), the own process (independent process restarting means) is terminated. As a result, the generation (restart) process becomes an orphan process having no parent process (for example, a process monitoring unit or an independent process restart unit). As a result, the parent-child relationship between the process monitoring unit and the generation (restart) process (that is, the restarted process monitoring unit) originally connected by one process tree is cut off and separated into an independent process tree. As a result, it is possible to prevent the monitoring target process from stopping according to the formula, and in the worst case, monitoring and restart cannot be performed.

  According to such a monitoring configuration, even if N-1 process monitoring means out of N process monitoring means are terminated for some reason, all the process monitoring means can be restarted. With this monitoring configuration, there is an advantage that each process monitoring means only needs to monitor the process of the preceding process monitoring means.

In the example of the circular monitoring shown in FIG. 1, an arbitrary process monitoring means 120 _i (i∈ {1, 2,..., N−1, N}) is converted into one process monitoring means 120 _j (j∈ {1,2). , ..., N-1, N}-{i}), but one process monitoring means monitors two or more process monitoring means, or two or more process monitoring means The configuration may be such that one process monitoring means is monitored.

Second Embodiment
The process monitoring system 200 according to the second embodiment includes N (where N is a predetermined integer of 2 or more) monitoring target programs 230 ₁ , 230 ₂ ,..., 230 _N−1 , 230 _N respectively. , 220 _N-1 , 220 _{N. The system} includes a registry 210 as storage means and N process monitoring means 220 ₁ , 220 ₂ ,. Each of the N process monitoring units 220 ₁ , 220 ₂ ,..., 220 _N−1 , 220 _N is configured to be able to access the registry 210.

Process monitoring unit _{220 1, 220 2, ...,} 220 N-1, 220 _N, respectively, the process monitoring unit _{220 1, 220 2, ...,} 220 N-1, 220 at least one process monitoring means other than the self of the _N , And at least one process monitoring means other than itself among the process monitoring means 220 ₁ , 220 ₂ ,..., 220 _N-1 , 220 _N is monitored. Monitoring between the process monitoring means 220 ₁ , 220 ₂ ,..., 220 _N−1 , 220 _N is configured.
That is, the arbitrary process monitoring means 220 _i (i∈ {1, 2,..., N−1, N}) has at least one process monitoring means 220 _j (j∈ {1, 2,..., N−1). N}-{i}) and the self (that is, the process) from at least one process monitoring means 220 _k (k∈ {1,2,..., N-1, N}-{i}) The operating status of the monitoring means 220 _i ) is monitored.

In the example shown in FIG. 3, cyclic permutation of the subscript 220 attached to identify a process monitoring unit between any process monitoring unit and the process monitoring unit to be monitored by the process monitoring unit. Since (N N-1... 2 1) is established, the monitoring between the plurality of process monitoring means is configured in a ring shape (or may be referred to as a circulation shape). That is, in this example, the process monitoring means 220 _N monitors the operation status of the process monitoring means 220 _N-1, the process monitoring means 220 _N-1 monitors the operation status of the process monitoring means 220 _N-2, · · -, process monitoring unit 220 ₂ monitors the operation status of the process monitoring means 220 _1, the process monitoring means 220 ₁ monitors the operation status of the process monitoring means 220 _N.

Process monitoring unit _{220 1, 220 2, ...,} 220 N-1, 220 N , respectively, N pieces of monitoring target program 230 _1, 230 _2, ..., are also separated from any of 230 _N-1, 230 _N Realized by an independent program.

Arbitrary process monitoring means 220 _i (i∈ {1, 2,..., N−1, N}) includes process information registration means 221 _i , process stop detection means 222 _i , and independent process restarting means 223 _i . Including.

Process information registration unit _{221 i (i∈ {1,2, ...} , N-1, N}) is in registry 210 in the process monitoring unit 220 _i is activated, the process monitoring unit 220 _i and process monitoring unit 220 _i The process information of each process by the monitoring target program 230 _i that is the monitoring target is stored. Specific examples of process information include a process ID and an execution file path. In addition, it is assumed that a continuous registry number starting from 1 is assigned to each process monitoring means in the order of registry registration (in this specification, the subscript of reference numeral 120 assigned to identify the process monitoring means is a registry number. To match).

Process stop detection means _{222 i (i∈ {1,2, ...} , N-1, N}) , the process monitoring means 220 _j (j∈ {1,2 is monitored process monitoring unit 220 _i, ..., N-1, N}-{i}) and the life and death of the process by the monitoring target program 230 _i which is the monitoring target of the process monitoring means 220 _i are monitored. For example, if the cyclic monitoring configuration shown in FIG. 3, a process stop detection means 222 _i, the process monitoring means 220 _i-1 is monitored process monitoring unit 220 _i (However, in the case of i = 1 Process monitor the life and death of the operating conditions and process monitoring unit 220 _i process by the monitoring target program 230 _i is monitored monitoring unit is 220 _N). Specific examples of this monitoring method include a method for receiving a process stop signal by acquiring each process handle of a process by a monitoring target program to be monitored and a process monitoring means, or an operating system releasing based on a semaphore. For example, a method for monitoring an event.

Independent process restart means _{223 i (i∈ {1,2, ...} , N-1, N}) , the process processes stop detection means 222 _i is monitored monitoring unit 220 _j (j∈ {1,2, ... , N-1, N}-{i}) or process stop by the monitored program 230 _{i, the} process information registered in the registry 210 is referred to, and the same process as the process monitoring means 220 _i A program for realizing the stopped process monitoring unit 220 _j or the stopped monitoring target program 230 _i is executed so as not to be included in the tree, and the process by the process monitoring unit 220 _j or the monitoring target program 230 _i is restarted. Let For example, in the annular monitoring configuration shown in FIG. 3, the independent process restarting means 223 _i uses the process monitoring detecting means 222 _i as the monitoring target process monitoring means 220 _i-1 (where i = 1). Process monitoring means 220 _N ) or the process information registered in the registry 210 is detected in the same process tree as the process monitoring means 220 _i when a process stoppage by the monitoring target program 230 _i is detected. The program for realizing the stopped process monitoring unit 220 _i-1 or the stopped monitoring target program 230 _i is executed so that the process monitoring unit 220 _i-1 or the monitoring target program 230 _i is restarted. Let
Here, "so that it does not contain the same process-tree", the parent-child relationship between the process monitoring unit 220 _i-1 is activated again and the process monitoring unit 220 _i, or re and process monitoring unit 220 _i This is to divide the parent-child relationship with the process by the monitoring target program 230 _i to be started, and this can be realized by, for example, so-called double fork (see FIG. 2).
That is, the independent process restarting means 223 _i (i∈ {1, 2,..., N−1, N}) is called when the process stop detecting means detects the end of the monitoring target, and generates (re-starts) a process that has ended. After starting), the own process (independent process restarting means) is terminated. As a result, the generation (restart) process becomes an orphan process having no parent process (for example, a process monitoring unit or an independent process restart unit). As a result, the parent-child relationship between the process monitoring means and the generation (restart) process (that is, the restarted process) originally connected by one process tree is cut off and separated into an independent process tree. As a result, it is possible to prevent the monitoring target process from stopping according to the formula, and in the worst case, monitoring and restart cannot be performed.

  According to the second embodiment, compared with the first embodiment, the number of targets monitored by one process monitoring unit increases from 1 to 2, but the program for realizing the process monitoring unit is changed from the monitoring target program. Because they are separated, there is no need to make any changes to the monitored program.

  Also in the second embodiment, according to the above-described monitoring configuration, even if N-1 process monitoring means out of N process monitoring means are terminated for some reason, all process monitoring means are restarted. In particular, in the case of an annular monitoring configuration, there is also an advantage that each process monitoring means only needs to monitor the process of the preceding process monitoring means.

In the example of the annular monitoring shown in FIG. 3, any process monitoring means 220 _i (i∈ {1, 2,..., N−1, N}) is converted into one process monitoring means 220 _j (j∈ {1,2). , ..., N-1, N}-{i}), but one process monitoring means monitors two or more process monitoring means, or two or more process monitoring means The configuration may be such that one process monitoring means is monitored.

<Supplementary note>
The process monitoring system includes, for example, as a single hardware entity, an input unit to which a keyboard or the like can be connected, an output unit to which a liquid crystal display or the like can be connected, and a communication device (for example, a communication cable) capable of communicating outside the hardware entity. Connectable communication unit, CPU (Central Processing Unit) (may include a cache memory or a register), RAM or ROM as a memory, external storage device as a hard disk, and their input unit, output unit, communication unit , A CPU, a RAM, a ROM, and a bus connected so that data can be exchanged between the external storage devices. If necessary, a hardware entity may be provided with a device (drive) that can read and write a recording medium such as a CD-ROM. A physical entity having such hardware resources includes a general-purpose computer.

  The external storage device of the hardware entity stores a program necessary for realizing the above functions and data necessary for processing the program (not limited to the external storage device, for example, reading a program) It may be stored in a ROM that is a dedicated storage device). Data obtained by the processing of these programs is appropriately stored in a RAM or an external storage device.

  In the hardware entity, each program stored in an external storage device (or ROM, etc.) and data necessary for processing each program are read into a memory as necessary, and are interpreted and executed by a CPU as appropriate. . As a result, the CPU realizes a predetermined function (process monitoring means).

Claims (5)

A process monitoring system for monitoring the life and death of each process by a plurality of monitored programs,
Storage means;
A plurality of process monitoring means,
Each of the process monitoring means monitors the operating status of at least one process monitoring means other than itself among the plurality of process monitoring means, and at least one process monitoring means other than itself among the plurality of process monitoring means Monitoring between a plurality of the above process monitoring means is configured so that its own operation status is monitored,
Each of the above process monitoring means
Realized by a program incorporated in the corresponding monitoring target program among the plurality of monitoring target programs,
Process information registration means for storing process information in the storage means when the process monitoring means is activated;
A process stop detection means for monitoring the operating status of the process monitoring means to be monitored by the process monitoring means by monitoring an event released by the operating system based on the semaphore ;
When the process stop detection unit detects the stop of the process monitoring unit being monitored, the monitoring target program corresponding to the stopped process monitoring unit is executed with reference to the process information registered in the storage unit And an independent process restarting means for terminating the own process,
A process monitoring system, wherein the program for realizing each of the process monitoring means is configured as a dynamic link library. A process monitoring system for monitoring the life and death of each process by a plurality of monitored programs,
Storage means;
A plurality of process monitoring means,
Each of the process monitoring means monitors the operating status of at least one process monitoring means other than itself among the plurality of process monitoring means, and at least one process monitoring means other than itself among the plurality of process monitoring means Monitoring between a plurality of the above process monitoring means is configured so that its own operation status is monitored,
Each of the above process monitoring means
A process information registering unit for storing process information of each of the processes by the monitoring target program that is the monitoring target of the process monitoring unit and the process monitoring unit in the storage unit when the process monitoring unit is activated;
By monitoring an event released by the operating system based on the semaphore, the operation status of the process monitoring unit monitored by the process monitoring unit and the life and death of the process by the monitoring target program monitored by the process monitoring unit are monitored. Process stop detection means;
When the process stop detection unit detects a process stop by the process monitoring unit or the monitoring target program that is a monitoring target, the process monitoring unit or the monitoring target stopped by referring to the process information registered in the storage unit An independent process restarting means for terminating the process after restarting the process by the program,
A process monitoring system, wherein a program for realizing each of the process monitoring means is separated from each of the monitoring target programs. In the process monitoring system according to claim 1 or 2 ,
The number of the process monitoring means is the same as the number of the monitoring target programs,
Each of the process monitoring means monitors the operation status of one process monitoring means other than itself among the plurality of process monitoring means, and self-monitors from one process monitoring means other than itself among the plurality of process monitoring means. The process monitoring system is characterized in that the monitoring between the plurality of process monitoring means is configured in a ring shape by monitoring the operation status of the process. Storage means;
Each including a plurality of process monitoring means realized by a dynamic link library incorporated in a corresponding monitored process among the plurality of monitored programs,
Each of the process monitoring means monitors the operating status of at least one process monitoring means other than itself among the plurality of process monitoring means, and at least one process monitoring means other than itself among the plurality of process monitoring means A process monitoring method in a process monitoring system in which monitoring between a plurality of the process monitoring means is configured such that the operation status of the process is monitored from
Each of the process monitoring means stores a process information registration step for storing process information in the storage means when the process monitoring means is activated,
A process stop detection step in which each of the process monitoring means monitors an operation status of the process monitoring means that is a monitoring target of the process monitoring means by monitoring an event released by the operating system based on the semaphore ;
When each of the process monitoring means detects the stop of the process monitoring means to be monitored in the process stop detection step, the process monitoring means refers to the process information registered in the storage means to the stopped process monitoring means. A process monitoring method comprising: an independent process restarting step for terminating the own process after executing a corresponding monitoring target program. Storage means;
Each including a plurality of process monitoring means realized by a program separated from each of the plurality of monitoring target programs,
Each of the process monitoring means monitors the operating status of at least one process monitoring means other than itself among the plurality of process monitoring means, and at least one process monitoring means other than itself among the plurality of process monitoring means A process monitoring method in a process monitoring system in which monitoring between a plurality of the process monitoring means is configured such that the operation status of the process is monitored from
Process information in which each process monitoring unit stores process information of each process by the monitoring unit program monitored by the process monitoring unit and the process monitoring unit in the storage unit when the process monitoring unit is activated Registration step;
Each of the process monitoring means monitors an event released by the operating system based on the semaphore, so that the operation status of the process monitoring means that is the monitoring target of the process monitoring means and the monitoring target program that is the monitoring target of the process monitoring means A process stop detection step for monitoring the process life and death caused by
When each process monitoring unit detects a process stop by the process monitoring unit or the monitoring target program that is a monitoring target in the process stop detection step, the process monitoring unit refers to the process information registered in the storage unit A process monitoring method comprising: an independent process restarting step for terminating the own process after restarting the process by the stopped process monitoring means or the monitoring target program.

Images