KR20030052015A - System and Method for Process Recovery in Multiprocess Operating System - Google Patents

Abstract

PURPOSE: A system and a method for recovering a process on a multi-process OS(Operating System) are provided to fast recover a trouble by constructing and managing the information related to all processes on the multi-process OS, and to debug the trouble easily through the trouble occurring particulars. CONSTITUTION: A database(12) stores the generation information and the state information for each process dynamically generated and deleted on the multi-process OS. Management processes(11-1¯11-n) manage respective processes by using the information stored in the database(12), and recover the trouble by deleting and regenerating the corresponding process when the trouble is occurred on a specific process. The database(12) is configured by tree structure nodes matched with each process. A tree structure node includes the process generation information such as a process ID, a process priority, and a process name, and the process management information such as a recover flag, a no-signal flag, and a transmission signal number.

Description

System and Method for Process Recovery in Multiprocess Operating System

The present invention relates to a process recovery in a multi-process operating system, and more particularly, to a process recovery system and method in a multi-process operating system for recovering a failure by deleting and regenerating the corresponding process when a specific process failure occurs when operating the system. will be.

In a multiprocess operating system, the system operates while creating subprocesses from the main process and communicating with each other. However, when a specific process fails during the operation of the system, the system itself can recover from the failure.

In addition, if a particular process fails during system operation, the system operator or software developer may not only have difficulty debugging the system but also if the system is restarted for quick failure recovery, all process information will be deleted, and even the operation record of the failure history can be found. There are many cases that cannot be.

In other words, the conventional multi-process operating system proposes a method of creating and deleting a process, but fails to suggest a method of recovering a failed process when the process fails.

This is because failures that occur during system operation are often caused by software problems of software developers, but the operating system cannot have all information about processes created by software developers.

Therefore, when a specific process fails in an infinite loop or deletes the process from the operating system during system operation, the system operator or software developer can directly input a command through the device that monitors the process status. Either the system must be restarted or the entire system must be restarted.

However, in case of recovering a failure by directly inputting a command through the process monitoring device, when the system is actually installed in the field, the monitoring devices cannot be used and the system takes a long time to recover from the failure. If a catastrophic problem occurs and the system operator restarts the system to recover quickly from the failure, the operating records for the failure may not be found. The problem arises that causes difficulties.

The present invention is to solve the problems as described above, the purpose is to build and manage the relevant information for all processes in the multi-process operating system to quickly recover from a failure when a specific process failure, failure occurs The history makes it easy to debug the fault.

1 is a block diagram of a process recovery system in a multiprocess operating system according to the present invention;

FIG. 2 is a diagram showing the internal structure of a database in FIG. 1; FIG.

3 is a diagram showing a tree structure node in FIG.

4 is a flowchart illustrating a process recovery operation in a multiprocess operating system.

Explanation of symbols on the main parts of the drawings

11-1 ~ 11-n: Process 12: Database

12-1 ~ 12-n: Tree Structure Node 13: Process Management Process

A feature of the present invention for achieving the above object is a database for storing the creation information and status information for each process that is dynamically created and deleted in the multi-process operating system; Processes in the multi-process operating system, characterized in that for managing each process using the information stored in the database, and if a specific process failure occurs, the management process for recovering the failure by deleting and regenerating the corresponding processes To provide a recovery system.

Here, the database is a tree structure database, characterized in that composed of tree structure nodes corresponding to each process, the tree structure node, the process creation information such as process ID, process priority, process name, and the like, A recovery flag to determine which process to delete and recreate in the event of a process failure, a complaint flag to determine whether the process can receive a health check request signal from the managed process, and a failed process to signal a health check request from the management process And process management information such as a transmission signal number indicating the number of times of reception.

The management process checks a setting value of a recovery flag of a tree structure node corresponding to a failure process when a failure occurs in a specific process, and if the flag setting value is '0', the failure process and its child processes. And deleting both the parent process of the failing process and other child processes of the parent process, and regenerating the deleted parent process. If the flag setting value is '1', the failing process and the child process are deleted after deleting the failing process and the child process. By regenerating the process that occurred, it is possible to recreate the processes deleted in the event of a failure.

In addition, the management process is characterized in that when a failure occurs in a specific process, the failure occurrence history is stored in the auxiliary storage device for debugging.

Further, another aspect of the present invention, the process of the management process in the multi-process operating system to check whether the steady state signal is received within a certain time after sending a status check request signal to each process; Detecting that a failure has occurred in a process for which the steady state signal is not received, and checking a recovery flag of a tree structure node corresponding to the failed process in a database; The process for recovering a process in a multi-process operating system comprising determining a failure recovery range according to a setting value of a recovery flag of the tree structure node, recovering a failure by deleting and regenerating corresponding processes. To provide.

The status check request signal may be transmitted only to a process in which a management process checks a reception rejection flag of a tree structure node corresponding to each process in a database and receives the status check request signal.

In addition, the process of recovering a failure by deleting and regenerating the processes may be performed when the setting value of the recovery flag of the tree structure node corresponding to the failure process is '0', and the parent of the failed process, its child processes, and the failed process. Deleting both the process and other child processes of the parent process; Regenerating the deleted parent process to recover the failure, and regenerating the deleted processes when the failure occurs. The setting value of the recovery flag of the tree structure node corresponding to the failure process is '1'. Deleting the failing process and its child processes; And regenerating the deleted process to recover the failure, thereby recreating the deleted processes when the failure occurs.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

In the multi-process operating system according to the present invention, the process recovery system includes a plurality of processes 11-1 to 11-n, a database 12, and a process management process (hereinafter, referred to as 'management process') as shown in FIG. 13).

A plurality of processes (11-1 to 11-n) are generated by the parent process generated by the main process according to information such as process ID, process name, process priority, etc., and operate through mutual communication to provide a corresponding service. .

The database 12 stores generation information and management information of all processes 11-1 to 11-n, which are shown in each process 11-1 to 11-n as shown in FIG. Corresponds to the tree structure. As such, the structure of the database 12 is that the tree structure means that the user process creates child processes like tree branches from the parent process, so managing the process in the tree structure is created in the parent process and the parent process together. This is because it is easy to identify the sibling processes that are in use and to facilitate the process of managing the failure.

In the real time OS, each process is dynamically created and deleted, so when deleting a specific process, information about the process is deleted from the database 12 so that the database 12 always updates the latest information. To maintain.

Each tree structure node 12-1 to 12-n constituting the database 12 is as shown in FIG.

The tree structure nodes 12-1 to 12-n are created with process creation information and process management information created when the processes 11-1 to 11-n are created. The process creation information is a process ID and a process priority. Rank, process name, etc., and process management information includes a failover flag for determining a process to be deleted and regenerated (Alone flag: hereinafter referred to as a 'recovery flag'), and a corresponding process (11-1 to 11-n). A No Signal Flag that determines whether a status check request signal of the management process 13 can be received, or a process 11-1 to 11-n in which a failure occurs is the status check request of the management process 13. A Send Signal Number indicating the number of times a signal is received, and the like.

The management process 13 manages all the processes 11-1 to 11-n of the system by using the information stored in the database 12, recovers the specific process 11-1 to 11-n which has failed, and also The failure history of a specific process (11-1 to 11-n) is stored in a separate auxiliary memory (not shown) so that the system operator or software developer can refer to it when debugging the failure. do.

In other words, the management process 13 periodically reads the database 12 and sends a status check request signal to the processes 11-1 to 11-n to determine the status of the processes 11-1 to 11-n. Then, it is checked whether or not the steady state signal is received from the processes 11-1 to 11-n which have received the state check request signal within a predetermined time.

At this time, since the status check request signal may not receive the status check signal due to the operation characteristics of the process, the reception check flag of the corresponding tree structure nodes 12-1 to 12-n may be changed. By checking, only the processes 11-1 to 11-n have been deleted without transmitting a status check request signal to the corresponding processes 11-1 to 11-n having the rejection flag set to '1'. .

If the management process 13 does not receive a steady state signal from a specific process 11-1 to 11-n within a predetermined time period, the management process 13 detects that the corresponding process 11-1 to 11-n has failed and performs a failure recovery routine. To operate.

For example, the management process 13 signals the status check request every 30 seconds to all the processes 11-1 to 11-n whose reception flags of the tree structure nodes 12-1 to 12-n are set to '0'. To confirm whether a steady state signal is received from the processes 11-1 to 11-n that have received the status check request signal, and the processes 11-1 to 11-n that have received the status check request signal If no steady state signal is transmitted for 20 seconds, the transmission signal numbers of the tree structure nodes 12-1 to 12-n corresponding to the corresponding processes 11-1 to 11-n are increased by one. If the transmission signal number is '3' or more, it detects that a failure has occurred in the corresponding processes 11-1 to 11-n and operates a failure recovery routine.

The failure recovery routine is made by the management process 13 analyzing the recovery flags of the tree structure nodes 12-1 through 12-n for the failed processes 11-1 through 11-n.

In other words, the management process (13) analyzes the recovery flags of the corresponding tree structure nodes (12-1 to 12-n) to determine the recovery range and then recovers the failure according to the recovery range, where the recovery flag is '1'. If set to ', the failed process is repaired by deleting the failed process and the child process of the failed process, and regenerating the failed process. If the recovery flag is set to' 0 ', the failed process and its The failure is recovered by deleting both the parent process and other child processes of the parent process (that is, sibling process) and then regenerating the deleted parent process to regenerate the deleted processes in case of failure. In this case, when the recovery flag of the parent process is also set to '0', the parent process of the parent process and the process derived from the parent process of the parent process are deleted, and then the parent process of the parent process is regenerated.

The value of '0' or '1' set in the recovery flag of the process is determined by the software developer to set the recovery flag.

In addition, the management process 13 stores a failure history such as a failure process name, a failure time, and the number of failures in a separate memory when the process fails, and outputs a failure history when requested by a system operator or software developer. give.

The failure history can be printed at the request of the system operator or the software developer because the failure is recovered immediately during the operation of the failure recovery routine, but it is destroyed when the system is restarted and stored in a separate auxiliary memory device to prevent it.

A process recovery operation in the multi-process operating system having the above configuration will be described in detail with reference to FIG. 4 as follows.

First, the management process 13 reads each tree structure node 12-1 to 12-n stored in the database 12, and the corresponding process 11-1 to 11-n in which the complaint flag is set to '0'. The low status check request signal is transmitted (step S41), and it is checked whether or not the steady state signal is received from the processes 11-1 to 11-n that have received the status check request signal within a predetermined time (step S42).

If the management process 13 does not receive a steady state signal from the processes 11-1 to 11-n that have received the status check request signal within a predetermined time as a result of the checking, the management process 13 performs the corresponding process 11-1. 11-n), a failure flag of the tree structure nodes 12-1 to 12-n corresponding to the corresponding processes 11-1 to 11-n is analyzed (step S43), and the recovery flag is analyzed. It is checked whether the set value is '0' (step S44).

As a result of checking the setting value of the recovery flag, when the recovery flag of the tree structure nodes 12-1 to 12-n for the failure process 11-1 to 11-n is set to '0', a failure occurs. If you delete the process and its child processes, the parent process of the failing process, and all other child processes of that parent process, then recreate the parent process (steps S45, S46, S47), and the recovery flag is set to '1'. Next, after deleting the failed process and the child processes of the failed process, the failed process is regenerated (steps S48 and S49).

For example, assuming that a failure occurs in the process 11-3 (hereinafter, 'process' c') whose process name is' C ', it will be described below.

When the management process 13 detects that process 'c' 11-3 has failed, the tree structure node 12-3 (hereinafter tree structure) for process 'c' 11-3 in the database 12 is detected. After checking the recovery flag of the node 'c', the failure process is deleted by deleting the failed process 'c' (11-3) and the related process and regenerating the corresponding processes.

At this time, when the recovery flag of the tree structure node 'c' (12-3) is set to '0', the tree structure which is a tree structure node for the child process 11-4 of the process 'c' (11-3) Node 'd' 12-4 and tree structure node 'b' 12-2, which is the tree structure node for parent process 11-2 of process 'c' 11-3, and its parent process ( A failed process 'c' (11-3) using the process information stored in the tree structure node 'e' (12-5), which is a tree structure node for another child process 11-5 created in 11-2). And deletes process 'b' 11-2, process 'd' 11-4 and process 'e' 11-5.

Thereafter, the management process 13 regenerates the process 'b' 11-2 which is the parent process of the failed process 'c'. Then, the failure caused to process 'c' 11-3 is repaired by the parent process generating child processes in turn.

On the other hand, when the recovery flag of the tree structure node 'c' (12-3) is set to '1', the tree structure node for the child process 11-4 of the failed process 'c' (11-3) Process 'c' (11-3) after deleting process 'c' (11-3) and process 'd' (11-4) using process information stored in tree structure node 'd' (12-4) Regenerate the error to recover from the failure.

In addition, the embodiment according to the present invention is not limited to the above-mentioned, and can be implemented by various alternatives, modifications, and changes within the scope apparent to those skilled in the art.

As described above, the present invention builds and manages related information about all processes in a multi-process operating system in a tree structure, thereby quickly recovering a failure in the event of a failure in a specific process, and easily through a failure history. This has the effect of debugging.

Claims (9)

A database for storing creation information and state information for each process that is dynamically created and deleted in a multi-process operating system; Processes in the multi-process operating system, characterized in that for managing each process using the information stored in the database, and if a specific process failure occurs, the management process for recovering the failure by deleting and regenerating the corresponding processes Recovery system. The method of claim 1, The database is a tree structure database, process recovery system in a multi-process operating system, characterized in that composed of tree structure nodes corresponding to each process. The method according to claim 1 or 2, The tree structure node receives process creation information such as process ID, process priority, and process name, a recovery flag for determining a process to be deleted and regenerated when a specific process fails, and a process request signal for checking the status of a management process by the corresponding process. In the multi-process operating system, characterized in that it comprises a process management information, such as a reception rejection flag for determining whether or not it is possible, a transmission signal number indicating the number of times the failure process has received the status check request signal of the management process. Process recovery system. The method of claim 1, The management process checks the setting value of the recovery flag of the tree structure node corresponding to the failing process when a specific process fails, and if the flag setting value is '0', the failing process and its child processes and failures. After deleting both the parent process of the originating process and other child processes of the parent process, regenerate the deleted parent process, and if the flag setting value is '1', the failure process and the child process are deleted after deleting the failing process and its child processes. A process recovery system in a multi-process operating system, wherein the process is regenerated to regenerate deleted processes in the event of a failure. The method according to claim 1 or 4, The management process is a process recovery system in a multi-process operating system, characterized in that when a failure occurs in a specific process, the failure history is stored in the auxiliary storage device for debugging. Confirming whether a steady state signal is received within a predetermined time after the management process sends a status check request signal to each process in the multi-process operating system; Detecting that a failure has occurred in a process for which the steady state signal is not received, and checking a recovery flag of a tree structure node corresponding to the failed process in a database; And determining a failure recovery range according to a setting value of a recovery flag of the tree structure node, deleting and regenerating corresponding processes, and recovering from the failure. The method of claim 6, The status check request signal is a multi-process operating system, characterized in that the management process checks the rejection flag of the tree structure node corresponding to each process in the database, and transmits only to the process that can receive the status check request signal. To recover a process in Windows. The method of claim 6, The process of recovering from a failure by deleting and regenerating the processes may include: a failure process, its child processes, and a parent process of the failure process when the setting value of the recovery flag of the tree structure node corresponding to the failure process is '0'; Deleting all other child processes of the parent process; Regenerating the deleted parent process to recover the failure, thereby regenerating the deleted processes when the failure occurs. The method of claim 6, The process of recovering from a failure by deleting and regenerating the processes may include: deleting the failing process and its child processes when a setting value of a recovery flag of a tree structure node corresponding to the failing process is '1'; And regenerating the deleted process to recover the failure, thereby recreating the deleted processes when the failure occurs.