KR100309678B1 - Process Monitoring and Failure Recovery - Google Patents

Abstract

The present invention targets a plurality of application functions mounted and operated in a processor constituting a control system of an exchange system, and detects a failure through an operating system and uses a process monitoring function when a failure of a process, which is a minimum unit of these functions, occurs. The present invention relates to a method for allowing an application function to perform a recovery function for a corresponding process by itself. The configuration of functions that are related to each other to perform the present invention includes a process to monitor the process information according to whether process monitoring is required at the start of the function. Application function that registers as management information through the system and performs regeneration by receiving notification from the monitoring function in the event of a failure, a library that registers and monitors process information that is called by the application function and manages information, and an operating system when a process failure occurs That information from By using the information managed in the library to perform the monitoring function and executing the monitoring function to deliver the processor information to be restored to the corresponding application function, the effect of the function abnormality caused by the process failure is minimized and the overall operation of the processor is affected. When the regeneration is repeated due to its own function error, the regeneration can be suppressed if necessary according to the operator's confirmation procedure. Therefore, the process monitoring function can be flexibly utilized. This can be applied to various types of resources that can be applied.

Description

Process Monitoring and Failure Recovery

The present invention relates to a process failure recovery method using a library.

Until now, in the process recovery method of this structure for the exchange system, upon receiving notification of the failure from the operating system, the process monitoring function checks whether the process is a monitored process, and then, through the system startup management function, the entire execution module to which the corresponding process belongs. Restart method was used.

In this way, by restarting other processes in normal operation at the same time, it could not only affect the system but also cause inconsistencies such as data used in common with other interoperable functions.

As another method, the process of regenerating the process by an automatic procedure through the operating system is performed through the operating system by registering the regeneration in advance when creating a process in the application function.

In this method, since the regeneration is performed by the operating system independently of the application functions, the recovery procedure due to a failure during the process cannot be performed normally, and in the case of a faulty process, the regeneration process is repeated to operate the entire processor function. It has some disadvantages that can adversely affect it.

In order to solve the above shortcomings, the present invention provides a processor-specific function by interworking with various application functions such as call processing, operation management, and maintenance, and an operation system for performing and controlling them in one processor as in an exchange system. If any process fails and is destroyed in the structure to be executed, it is necessary to create the process at initial processor start-up and regenerate it by directly judging from the function that manages the application function as a whole. The goal is to provide a way to restore functionality with minimal impact.

Therefore, in the present invention, in order to solve the deficiency of the failure recovery function by restarting the entire process and the automatic process regeneration process by the operating system of the application function belonging to the failure process represented by the problem in the prior art, Process recovery procedures according to the

To do this, the process information for monitoring the process is registered in advance through the library so that it can be notified when a process failure occurs by an automatic procedure.In the application function that is notified of the failure, the recovery of data and resources managed by the process After synchronization with the function, it is restored through the process of regeneration.

In addition, when repeated regeneration is performed due to a defect in the process itself, a method for suppressing regeneration is controlled to have no influence on the processor.

According to the method for recovering from a process failure according to the present invention for achieving the above object, if the process to be monitored by creating a process to be performed in the application function started from the processor operating system and checking the process to be monitored, management information on the process Read the first step to determine whether to recover; A second step of transitioning to a signal reception waiting state for the next performance if the recovery is in the inhibited state after confirming the recovery; A third step of storing a process instance definition of the processor if the recovery is allowed, performing a library for process monitoring, and then transitioning to a signal waiting state for normal function processing; A fourth step of performing an associated data recovery and initialization process upon receiving process failure occurrence information from a process monitoring function in the signal reception standby state; After performing the process of data recovery and initialization related to process failure, regenerate the process and save the process instance definition to register process monitoring information through the library, increase the number of recovery, and then transition to the standby state to receive a signal. It is made up of.

In addition, according to the process process failure monitoring method according to the present invention, after performing the registration information initialization process for the monitoring buffer in the library for the process monitoring, and transitioning to the standby state for receiving the process failure; A second step of performing a recovery control process on a process for an operator request upon receiving a signal for process recovery control by an operator request in a standby state; A third step of processing the process monitoring information output for the operator request upon receiving a signal for outputting the process monitoring information by the operator request; A fourth step of receiving processor loss information through an operating system when a process failure occurs when performing an application function in the standby state; After receiving the process loss information, the first value of the monitored process property managed by the library is read to check whether the received process failure information is the destination process process instance definition, so that the failed process is the watchdog buffer in the case of the destination process instance definition. A fifth step of requesting to output a process failure message after the corresponding information is deleted from the corresponding information and transitioning to a standby state for the next processing; If it is not the definition of the destination process instance, check the definition number of the monitored process and if it is different, check whether the processing for all registered processes is completed. If not, specify the next serial number and perform the repetition process. With six steps; A seventh step of requesting to output a failure message for the corresponding process if the processing of the entire registered information is completed and transitioning to a waiting state for the next processing; If it is the same as the monitored process definition number, check whether it matches the monitored processor process instance definition and if it is different, check whether the processing for all registered processes is completed.If not, specify the next serial number and then repeat the process. Performing an eighth step; A ninth step of requesting to output a failure message for the corresponding process if the processing of the entire registered information is completed and transitioning to a waiting state for the next processing; If it is the same as the monitored process instance definition, it sends the process failure information signal to the registered destination, deletes the already registered process monitored process instance definition information, requests the failure message for the process, and then executes the next processing. The tenth step is to transition to the standby state.

1 is a flowchart of a related function configuration and processing according to the present invention;

2 is a process monitoring information registration and recovery process flow chart according to the present invention;

3 is a library processing flowchart for registering process monitoring information according to the present invention;

4 is a process failure monitoring function processing flowchart to which the present invention is applied;

5 is a flowchart illustrating a process recovery control process by an operator request according to the present invention;

6 is a process monitoring process output flow chart according to an operator's request according to the present invention;

* Explanation of symbols for main parts of the drawings

100: processor (OMP / CCCP) 10, 20, 30, 40: application functions A, B, C, D

50: process monitoring information library

60: process monitoring function 70: operating system

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

The present invention registers the process information to be monitored to the exchange system, and the application function to be notified of the failure occurrence information and regenerate, and the monitoring function and the monitored process information to determine whether to recover the process by receiving the process information from the operating system when a process failure occurs It consists of library function to manage registration.

Figure 1 shows the interaction between the related functions constituting the present invention, when the application functions A, B, C, D (10, 20, 30, 40) is started at the time the processor is started for the monitored process The library 50 is called to register process information (S1, S2, S3, S4).

When a process failure occurs in the operation of a process belonging to an application function, the operating system 70 detects (S5, S6, S7, S8) and notifies the process monitoring function unit 60 (S9).

The process monitoring function unit 60 reads the process information managed by the library based on the reported process information (S10) and checks whether the process is a process to be monitored, and delivers it to the application function only in the case of a process requiring repair (S11). , S12, S13, S14).

Application functions notified of process failures perform the relevant recovery procedure and then recreate the process.

Table 1 is a configuration information for the process to be monitored using the library can be used to determine the size of the buffer in units of processors using a serial number.

In this table, only 100 are specified from 0 to 99, but this can be used by expanding the number according to the operating characteristics of the processor.

In addition, the process definition number represents a unique number for functionally identifying the process, which is determined at the stage in which an execution module for the corresponding application function is produced.

Process Instance Definition (hereinafter referred to as Pid) to be monitored corresponds to a unique number assigned by the operating system at the time a corresponding process is created, and this information is changed when it is regenerated due to a process failure.

The receiving process Pid indicates the unique number of the process to receive the information when the registered monitoring target process fails. If the value is not correct, the receiving process Pid cannot receive the information through the signal.

The number of process recovery is the recovery information indicating how many times the process has been regenerated due to a process failure for the process definition number. By checking this information, it is indirect to determine whether the process contains a serious error and is continuously regenerated. It can be determined that the result of this process is the information that can determine that the performance of the processor function.

The recovery information indicates whether a recovery state (TRUE) indicating a state in which a recovery for the process can be performed by the application function itself or an operator's request or a suppressed state (FALSE) requiring recovery.

FIG. 2 illustrates a process for a process flow of registering information for process monitoring in an application function and recovering when a failure occurs. The application function is started by a processor operating system (S30), and a process to be performed in an application function is created. After confirming (S31) whether or not the process to be monitored (S32), and if the process to be monitored (S33) to read the management information for the process (S33) to check whether or not to recover.

At this time, if the recovery is allowed (TRUE) state, the processor Pid is stored (S35), the library for process monitoring is performed (S36), and the state transitions to the signal waiting state for normal function processing (S37), and the recovery is inhibited (FALSE). ), The state transitions to the signal reception waiting state for normal function processing (S37).

In step S37, if the process failure information is received from the process monitoring function (S38), after performing the relevant data recovery and initialization process (S39), the process is regenerated (S40) and the process Pid is stored. (S41), the process monitoring information through the library is re-registered (S42), and the recovery count is increased by one (S43), and then the state transitions to the standby state for receiving the next signal (S44).

3 is a process of a library for registering and managing process monitoring information, and corresponds to a procedure called when a process is created during the process of the application function described in FIG. do.

When the process monitoring information is called from the application function for registration (S36a), it is checked whether it is already registered for the corresponding process information (S36b), and if it is not registered, the procedure for checking the monitored target Pid is performed (S36c). If it is already registered, an error message for the abnormal registration request is requested (S36h) and ends (S36g).

If the check result of the monitoring target Pid is normal, a procedure for checking the destination Pid is performed (S36d). If the monitoring target Pid is in an abnormal state, an error message for this is output (S36h) and terminated.

If the check result of the destination Pid is normal, a procedure for checking the free state of the monitoring buffer is performed (S36e). If the destination Pid is in an abnormal state, an error message for this is output (S36h) and the process is terminated. S36g).

If the monitoring buffer can be stored, the requested process information is registered in the monitoring buffer (S36f) and then terminated (S36g).

4 illustrates a process of a process failure monitoring function. When the execution of a function is started, the process proceeds to the standby state for accepting a process failure after performing a property initialization process for a monitoring buffer in a library for process monitoring (S45). (S46).

When the signal for the process recovery control by the operator request is received in the standby state (S47), the process of FIG. 5 is performed, and when the signal for the process monitoring information output by the operator request is received (S48), the process of FIG. Do this.

If a process failure occurs while the application function is executed in the standby state, the processor loses information about the processor loss through the operating system (S49).

After reading the first value of the supervised process property managed by the library (S50), it is checked whether the received process failure information is the destination process Pid (S51). In this case, if the failed process is the destination Pid, a failure occurs. Since there is no process to receive the information, the corresponding information is deleted from the monitoring buffer (S52), and a process failure message is output (S53), and the state transitions to a waiting state for the next processing (S54).

If it is not the destination Pid, check the definition number of the process to be monitored (S55), otherwise check whether the processing for all registered processes is completed (S56), and if not, specify the next serial number ( S57) Perform an iterative process.

If the processing for the entire registered information is completed, a failure message for the corresponding process is output (S53), and the state transitions to the waiting state for the next processing (S54).

If it is the same as the supervised process definition number, it checks whether it matches the supervised process Pid (S58), and if it is different, checks whether the processing for all registered processes is completed (S56). After designation (S57), an iterative process is performed.

If the processing for the entire registered information is completed, a failure message for the corresponding process is output (S53), and the state transitions to the waiting state for the next processing (S54).

If it is the same as the monitored process Pid, the process failure occurrence information signal is transmitted to the registered destination (S59), and the previously registered process monitored process Pid information is deleted (S60), and a failure message for the process is requested. (S53), the process goes to the standby state for the next processing (S54).

The monitored process information about this is re-registered at the time of regenerating the process in the application function.

FIG. 5 illustrates a process for controlling process recovery by an operator's request. After reading the monitored process information (S47a) and confirming the existence of the requested process definition number (S47b), if not present, an abnormal processing result message is shown. The output is requested (S47c), and the state transitions to the standby state for the next processing (S47d).

In the case of an existing process, the requested control type is checked (S47e). In case of a release process, it is checked whether it is already released (S47f). If it is released, an abnormal processing result message is output (S47c). Transition to state (S47d).

If not, the recovery release (TRUE) information for the corresponding process is registered (S47g), a normal processing result message is outputted (S47h), and the state transitions to the waiting state for the next processing (S47d).

If the control mode is suppressed, it is checked whether it is already in the inhibited state (S47i). If the control mode is suppressed, an abnormal processing result message is output (S47c), and the state transitions to the standby state for the next process (S47d).

If it is not in the suppressed state, the recovery suppression (FALSE) information for the process is registered (S47j), the normal processing result message is outputted (S47h), and the state transitions to the standby state for the next process (S47d).

6 illustrates a process for outputting process monitoring information by an operator request. After checking the requested form (S48a), if the output request is made for an individual process, the monitoring process information is read (S48b). The presence or absence of the number is checked (S48c). If it is not present, an abnormal processing result message is output (S48d), and the state is transferred to the standby state for the next processing (S48e).

If a process definition number exists, a result message is configured (S48f), then a normal processing result message is requested for output (S48g), and the state transitions to the standby state for the next process (S48e).

In the case of request result checking result, the first monitored process information is read (S48h), and the result message is checked (S48i). After checking whether the whole process is completed (S48j), the next process location information is designated. After that (S48k) it repeats the process of reading the process to be monitored and constructing the result message.

When the entire process is completed, the normal process result message is requested for output (S48l), and the state transitions to the standby state for the next process (S48e).

As described above, the method for recovering by monitoring a process failure occurrence according to the present invention can register monitoring information in a function operating state according to the necessity of an application function, thereby providing convenience of use, and creating an application. By performing the procedure of notifying and regenerating the failure information from the function, the abnormal state caused by the process failure is the safest and the minimum range in that the recovery is performed in the function that can know the functional characteristics of the process most accurately. The recovery process can be done at.

Therefore, it has the advantage of minimizing the effects of functional abnormalities due to process failures and minimizing the effect on the overall operation of the processor.

In the case of repetitive regeneration due to its own function error, the regeneration can be suppressed if necessary according to the operator's confirmation procedure, so the process monitoring function can be flexibly utilized.

In addition, this type of function configuration can be applied to various types of resources that can be monitored as well as process monitoring functions.

Claims (5)

In the case of a process failure to recover it, Creating a process to be performed by an application function started from a processor operating system and checking whether the process is a process to be monitored, and if the process is a process to be monitored, reading management information about the process and checking whether to recover the process; A second step of transitioning to a signal reception waiting state for the next performance if the recovery is in the inhibited state after confirming the recovery; A third step of storing a process instance definition of the processor if the recovery is allowed, performing a library for process monitoring, and then transitioning to a signal waiting state for normal function processing; A fourth step of performing an associated data recovery and initialization process upon receiving process failure information from a process monitoring function in the signal reception standby state; After performing the process of data recovery and initialization related to process failure, regenerate the process and save the process instance definition to register process monitoring information through the library, increase the number of recovery, and then transition to the standby state to receive a signal. Process failure recovery method characterized in that consisting of. The method of claim 1, The fifth step, A first sub-step, when called from an application function, checking whether the corresponding process information is already registered and checking the monitored process instance definition if it is not registered; A second sub-step of requesting and terminating the error message output for the abnormal registration request if already registered; A third substep of performing a procedure for checking a destination process instance definition if it is normal that the monitored process instance definition is confirmed, and requesting and terminating a failure message output in case of abnormality; A fourth sub-step for checking a free state of the monitoring buffer if the check result of the destination process instance definition is normal, and outputting a failure message for the case where the destination process instance definition is abnormal, and ending the request; And a fifth substep of registering the requested process information in the monitoring buffer and terminating if the monitoring buffer can be stored. In the process failure monitoring method, A first step of performing a registration information initialization process for a monitoring buffer in a library for monitoring the process and transitioning to a standby state for receiving a process failure; A second step of performing a recovery control process on a process for an operator request upon receiving a signal for process recovery control by an operator request in a standby state; A third step of processing the process monitoring information output for the operator request upon receiving a signal for outputting the process monitoring information by the operator request; A fourth step of receiving processor loss information through an operating system when a process failure occurs when performing an application function in the standby state; After receiving the process loss information, the first value of the monitored process property managed by the library is read to check whether the received process failure information is the destination process process instance definition, so that the failed process is the watchdog buffer in the case of the destination process instance definition. A fifth step of requesting to output a process failure message after the corresponding information is deleted from the corresponding information and transitioning to a standby state for the next processing; If it is not the definition of the destination process instance, check the definition number of the monitored process and if it is different, check whether the processing for all registered processes is completed. If not, specify the next serial number and perform the repetition process. With six steps; A seventh step of requesting to output a failure message for the corresponding process if the processing of the entire registered information is completed and transitioning to a waiting state for the next processing; If it is the same as the monitored process definition number, check whether it matches the monitored processor process instance definition and if it is different, check whether the processing for all registered processes is completed.If not, specify the next serial number and then repeat the process. Performing an eighth step; A ninth step of requesting to output a failure message for the corresponding process if the processing of the entire registered information is completed and transitioning to a waiting state for the next processing; If it is the same as the monitored process instance definition, it sends the process failure information signal to the registered destination, deletes the already registered process monitored process instance definition information, requests the failure message for the process, and then executes the next processing. Process failure monitoring method comprising the tenth step of transition to the standby state for. The method of claim 3, wherein The second step, A first sub-step of reading a monitored process and checking whether a requested process definition number exists and if it does not exist, requesting an output of an abnormal processing result message and transitioning to a standby state for the next processing; A second sub-step, if present after checking, to check the requested control type and checking whether it is already released, and if it is released, requesting output of an abnormal processing result message and transitioning to a waiting state for the next processing; A third substep of registering the recovery release information for the corresponding process if it is not in the released state, requesting output of a normal processing result message, and transitioning to a standby state for the next processing; A fourth sub-step of checking whether the control mode is already in the inhibited state and requesting the output of the abnormal processing result message if the control state is in the inhibited state and transitioning to the standby state for the next processing; And a fifth substep of restoring recovery suppression information for the process, requesting output of a normal processing result message, and transitioning to a standby state for the next process. The method of claim 3, wherein The third step, After checking the requested form, if it is an output request for an individual process, it reads the monitored process information and checks the existence of the requested process definition number. If it does not exist, it requests an abnormal processing result message and transitions to the waiting state for the next processing. A first substep of doing; A second substep of constructing a result message if the process definition number exists, requesting output of a normal process result message, and transitioning to a standby state for the next process; In case of request result checking result, read first monitoring process information and compose result message. After checking whether complete processing is completed, if not completed, specify next process location information and read monitoring process and result message. A third substep of repeating the configuring process; And a fourth substep of requesting output of a normal processing result message and transitioning to a standby state for the next processing when the whole processing is completed.