JPH09179754A - Task monitoring device and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow a task itself to recognize its stop time, to visually grasp the stop state of the task and to automatically cope with the execution delay of the task. SOLUTION: An operating system(OS) part 2 for attaining a multi-task execution format is provided with a task stop allowable time setting means 2a, a stop time accumulating means 2b, a stop state provision means 2c, and a stop time excess reporting means 2d, and the OS part is also provided with a means for managing the stop time of a user defintion task to be executed by a computer, and when the stop time exceeds allowable time, informing the task of the event. A system managing monitor 3 to be driven as a task on the computer is provided with a task stop record collecting means 3a, a stop state displaying means 2b, a bottleneck detecting means 3c, a fault informing means 3d, a task state transition displaying means 3e, and a task inputting means 3f. The monitor 3 enters the task stop state recorded by the OS part 2 by the use of an application interface and manages the performance and operation of a computer system.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention manages a task stop time in a multi-task execution system computer and enables the task to recognize a processing delay, and the load status of the computer system. It enables the monitoring of. Furthermore, the present invention relates to a technique that enables performance management and operation management of a computer system.

[0002]

2. Description of the Related Art In a conventional multi-task execution type computer, various evaluation tools for evaluating the performance of the computer have been prepared and provided. For example, in the UNIX ps command, management information by the OS is displayed for each execution unit of a process which is an individual environment of a program executed under the control of an operating system (hereinafter referred to as OS). Similarly, in UNIX, vms is used as an evaluation tool to understand the usage status of system resources on computers.
A tat command is provided. The vmstat command displays data for detecting a response deterioration due to excessive use of system resources. In the multitasking execution environment, a transaction processing monitor is used to guarantee the response of the processing in the transaction processing.

In the ps command, event information indicating the cause of the process (program) stopping is displayed as software address information in the WCHAN column in the command output shown in FIG. 13, for example. Where WCHAN
Processes that do not have an address in the column are those that are in a runnable state or are actually programs that are part of the OS. Display data such as address information displayed in the WCHAN column of the ps command output is managed by the OS in a table allocated for each process. The ps command uses an application interface (hereinafter, API) for obtaining the data in the table and displays the obtained data. However, this information could not be judged by the task itself.

Further, in the vmstat command, as shown in FIG.
You will get an output like In the output, the items of procs are waiting for CPU, the items of memory are memory usage status, the items of page are paging processing status, the items of fault are interrupt status, and cpu.
In the item, the usage status of the CPU is shown. These pieces of information are managed by the OS as data inside the OS. Also in the vmstat command, the API is executed to obtain the data inside these OSs, and the obtained data is displayed.

Further, in the online transaction processing system, a means for improving the throughput of the system is adopted by organizing the transactions input to the computer system by using the transaction processing monitor. This transaction processing monitor has a queuing function of storing execution requests in a queue and then passing them to a database application, and a scheduling function of sequentially processing according to priority.
This function guarantees the response time of the processing in the computer system using the transaction processing monitor.

However, all of the above information can be judged only by the OS itself, and the task itself cannot use this information. Even if an interface for taking in this information was provided, it was not possible to judge the information.

[0007]

In executing a task in a computer, a task execution delay may occur due to disturbance. Disturbance here refers to all events that make it impossible to continue executing a task. This event includes, for example, waiting for completion of various I / Os, waiting for computer resources such as memory and CPU, communication with other tasks under multitask execution,
There are communication with tasks on other computers, and stoppages set by the tasks themselves. Therefore, the task on the computer that does not manage the stop time cannot know the execution delay due to the disturbance.

Further, in the conventional system evaluation tool, O
Since S does not manage the stop time of the task, it is difficult to immediately understand the cause of the response deterioration or system hang caused by the disturbance. For example, when I / O retries occur frequently and the response of the system deteriorates, an abnormality is not detected by the time spent for one I / O, but the number of I / O retries or I / O service is performed. It was necessary to detect the abnormality by the log of the software that provides the.

Further, in the conventional transaction processing monitor for managing the queue of execution requests, it is necessary to grasp how much one execution request causes a delay in the task due to the overhead in the processing of the OS. Was difficult.

The present invention has been made to solve the above problems, and it is an object of the present invention to obtain the following functions in a multitasking computer. (1) In an OS that realizes multitask execution, processing for notifying a task of execution delay is performed.
By enabling the task itself to recognize the execution delay, it is possible to automate the recovery process in the task itself. (2) Realize the system management monitor as one task, obtain the stop time distribution for each cause that the task being executed in the system is stopped at regular intervals, and facilitate the bottleneck analysis of the computer system. . (3) Further, the system management monitor makes it possible to obtain the stop time distribution for each stop cause in an arbitrary task. (4) Further, the system management monitor monitors the stop time in an arbitrary task, and when the stop time exceeds the allowable value, the action defined by the user is automatically executed. (5) Also, the system management monitor provides task state transition information focusing on the cause and duration of the task. (6) Further, the system management monitor determines the load status based on the stop time of the task being executed, and controls the task to be input to the computer system according to the load status.

[0011]

A task monitoring apparatus according to the present invention is a computer in which a plurality of tasks are executed in parallel, and an operating system for controlling the execution of the tasks is provided with the following elements. It is a thing. (A) Permissible stop time setting means for setting the permissible value of the stop time for each task, (b) Every time the operating state of the task transits, the task is suspended for each task according to the transition state Stop time accumulating means for recording the cause of the stop, the stop time, the stop time, and the stop status of the stop cause and the cumulative value of the stop time according to the cause of the stop of the task, and (c) the task recorded by the stop time accumulating means. Comparison means for comparing the stop time with the tolerance value corresponding to the task,
(D) Stop time excess reporting means for reporting to the task that the stop time exceeds the allowable value when the comparing means determines that the task stop time exceeds the allowable value.

Further, in a computer in which a plurality of tasks are executed in parallel, the following elements are provided in an operating system which controls the execution of the tasks, and (a) each time the operating state of the task transits, its transition Depending on the state, the cause of stopping the task for each task, the stop time, the stop time, and the stop time accumulating means for recording the stop status such as the stop cause and the cumulative value of the stop time according to the cause of the stop of the task,
(B) The following elements are provided in the stop situation providing means for providing the stop situation of the task recorded by the stop time accumulating means and the system management monitor operating at a predetermined interval. (A) stop status collecting means for collecting the stop status provided by the stop status providing means, (b) stop for displaying the task load status in a frequency distribution based on the stop status of the task collected by the stop status collecting means Status display means.

Further, the stop status display means displays the load status of the task at a predetermined interval.

Further, the stop status display means displays the accumulated task load status.

Further, the stop status providing means provides the stop status including the information on the stop status.

Further, in a computer in which a plurality of tasks are executed in parallel, an operating system for controlling the execution of the tasks is provided with the following elements, and (a) each time the operating state of the task transits, its transition Depending on the state, the cause of stopping the task for each task, the stop time, the stop time, and the stop time accumulating means for recording the stop status such as the stop cause and the cumulative value of the stop time according to the cause of the stop of the task,
(B) The following elements are provided in the stop situation providing means for providing the stop situation of the task recorded by the stop time accumulating means and the system management monitor operating at a predetermined interval. (A) Stop status collection means for collecting the stop status provided by the stop status providing means, (b) Abnormal status definition information defining the task abnormal status for each task, (c) The stop status and the abnormal status definition An abnormal state detecting means for detecting an abnormal state of the task based on the information; and (d) an abnormal state notifying means for notifying the task of the abnormal state when the abnormal state detecting means determines the abnormal state.

Further, in a computer in which a plurality of tasks are executed in parallel, the following elements are provided in an operating system for controlling the execution of the tasks, and (a) each time the operating state of the task transits, the transition Depending on the state, the cause of stopping the task for each task, the stop time, the stop time, and the stop time accumulating means for recording the stop status such as the stop cause and the cumulative value of the stop time according to the cause of the stop of the task,
(B) The following elements are provided in the stop situation providing means for providing the stop situation of the task recorded by the stop time accumulating means and the system management monitor operating at a predetermined interval. (A) Stop status collection means for collecting the stop status provided by the stop status providing means, (b) Abnormal status definition information defining the task abnormal status for each task, (c) The stop status and the abnormal status definition An abnormal state detecting means for detecting an abnormal state of the task based on the information, and (d) a task inputting means for activating a task for processing the abnormal state when the abnormal state detecting means determines the abnormality.

Further, a system management monitor interface means for updating the abnormal state definition information from a user task is provided.

Further, the system management monitor is provided with a task state transition display means for displaying the transition state of the task based on the stop status of the task.

Further, in a computer in which a plurality of tasks are executed in parallel, the operating system for controlling the execution of the tasks is provided with the following elements, and (a) each time the operating state of the task transits, its transition Depending on the state, the cause of stopping the task for each task, the stop time, the stop time, and the stop time accumulating means for recording the stop status such as the stop cause and the cumulative value of the stop time according to the cause of the stop of the task,
(B) A stop status providing means for providing the stop status of the task recorded by the stop time accumulating means, a system management monitor interface means for holding the input information of the task set by the user task are provided, and the operation is performed at a predetermined interval. The system management monitor has the following elements. (A) Stop status collection means for collecting the stop status provided by the stop status providing means, (b) Judgment possibility for judging whether the task can be inputted based on the stop status of the task collected by the stop status collection means And (c) task input means for inputting a task based on the task input information obtained from the system management monitor interface means when the task input availability determination means determines that the task can be input.

The task monitoring method according to the present invention has the following steps. (A) setting a permissible value of the stop time of the task for each task, (b) recording the stop time and cause of the stop when the task is stopped, (c) recovering the task from the stopped state When, the step of recording the stop time and the cause of the stop and the cumulative value of the stop time, (d) the comparing step of comparing the stop time with the allowable value, and (e) the stop time exceeding the allowable value in the comparing step If it is determined that the task has been stopped for more than the allowed time, then the step of reporting.

Further, it has the following steps. (A) When the task is stopped, a step of recording the stop time and the cause of the stop, (b) When the task is recovered from the stopped state, the stop status of the task such as the stop time and the cause of the stop and the cumulative value of the stop times And (c) displaying the task load status in a frequency distribution based on the stop status.

Further, it has the following steps. (A) a step of defining an abnormal state of each task, (b) when the task is stopped, its stop time,
Recording the cause of the stop, (c) recording the stop status of the task such as the stop time and the cumulative value of the stop cause and the stop time when the task recovers from the stopped state, (d) based on the stop information An abnormal state detecting step of detecting the occurrence of the abnormal state defined in 1., and (e) a step of processing the abnormal state when an abnormality is detected in the abnormal state detecting step.

Further, the step of processing the abnormal state is a step of reporting to the task corresponding to this abnormal state.

Further, the step of processing the abnormal state is a step of activating a task for processing the abnormal state.

[0026]

1 is a diagram showing the overall configuration of an embodiment of a task monitoring apparatus according to the present invention. In the figure, 1 indicates the overall configuration of software that constitutes the task monitoring device, and 2 indicates an operating system (hereinafter,
It is called OS). The OS 2 has an allowable time setting means 2a for setting a stop time that may stop the task, a stop time and stop cause accumulating means 2b for the stopped task, a task stop status providing means 2c, and a stop time excess reporting means 2d. doing. A system management monitor 3 operates at a predetermined cycle, and is realized as one task executed on the OS 2. The system management monitor 3 includes a task stop status collection unit 3a, a stop status display unit 3b, a task abnormal state detection unit 3c, an abnormal state notification unit 3d, a task state transition display unit 3e, and a task input unit 3f for activating a task. Have Reference numeral 4 denotes a user-defined task (hereinafter referred to as a task) executed under the control of the OS 2, and 5 denotes a system management monitor interface means (hereinafter, system management monitor) for interfacing the user-defined task 4 with the system management monitor 3. (I / F means).

Embodiment 1 In the first embodiment, O
An example in which S2 manages the stop time of the task 4 and notifies the task 4 of the event when the stop time is exceeded will be described. In the task 4, it is possible to set the permissible value of the task-specific stop time for the OS 2 by the permissible stop time setting means 2a, and when the stop time of the task 4 exceeds this permissible value, the OS 2 The event notification can be received by the means 4d that the task 4 can detect by the stop time excess reporting means 2d. An example of a method of notifying an event to a task is a means such as a signal such as UNIX.

FIG. 2 is a diagram for explaining the cumulative processing of the task stop time and the reporting time of the stop time in the first embodiment, which will be described with reference to this figure. The graph 10 shown in FIG. 2 represents the state transition of one task in time series. The vertical axis represents task status, and the horizontal axis represents time flow. In the state of 11a, the task 4 being noted is CP
It is in a state of being executed by U (hereinafter also referred to as a processor). The transition to the state of 12a at the time in the state of 11b represents a stopped state due to the task 4 requesting resources in the program. At the time of the state of 11c, the task 4 changes to the state of 13a by the CPU.
Represents the case where the use of is restricted. 12a to 12
The period of b represents the time during which the task 4 stopped until the task 4 acquired the requested resource. Also, 13
The period from a to 13b represents the time when the task 4 is stopped until the task 4 has the execution right by the CPU. 1
From 1d, the task 4 is being executed again.

FIG. 3 shows a task management table 20 managed by the OS 2 for each task 4 and a trace buffer 30 for tracing the stop status of tasks. The task management table 20 is a task management table assigned by the OS 2 for each task 4, and the OS 2 stores management data relating to the task 4. That is, the time 21 at which the task 4 was last stopped, the cause 22 that was stopped last, the allowable stop time 23, the total stop time 24, the allowable stop time 25 for each stop cause, the cumulative number of times 26 for each stop cause, and each stop cause The task stop status such as cumulative time 27 is stored.

Also, the task stop status is set to O in the trace buffer 30 provided in a part of the data area of OS2.
S2 writes in. Reference numeral 40 denotes an API for extracting data from the task management table 20 and the trace buffer 30 in which the OS 2 records the task stop status, and can be executed by a task program (task). The system management monitor 3 receives various data from the OS 2 using this API 40.

In order to notify the task 4 of an event, the OS 2 accumulates and manages the number of stop times and the stop time for each stop cause for each task. The task stop event management is performed by a computer in the context saving process (processes 12a and 13a in FIG. 2) and the context recovery process (processes 12b and 13b in FIG. 2) when the OS 2 switches the task 4. This is done using a timer that records the time that has elapsed since the startup. At the same time, the OS 2 writes the trace data such as the time stamp 31, the identifier 32 of the task 4 that has been subjected to the recovery process, the stop time 33, the stop cause 34, etc., to the trace buffer 30 secured in the storage device. Writing to the trace buffer 30 is performed recursively so that the system management monitor 3 can obtain data in time series. If the stop time in task 4 exceeds the preset allowable value, the OS
2 is task 4 at the timing of 11d in FIG.
Using the method that can be detected, that is, the stop time excess reporting means 2d notifies the event.

The details of the processing for managing the stop time of the task 4 in the first embodiment will be described below with reference to the flow chart shown in FIG. In step S101, it is determined whether or not the cause of the task 4 being stopped is waiting for an event, and the processing differs depending on the result. The event wait here is a factor that causes the execution of task 4 to be stopped, that is, C
It is something other than waiting for PU usage. When the cause of the stop is not the event wait, the process of step S120 saves the context and does not perform the process related to the event wait.

In step S102, processing for saving the execution environment of the task 4 which has been executed up to now is performed on the premise of task switching. In step S103, a process of recording the last stop time 21 (time information when the task started waiting for an event) and the last stop cause 22 in the task management table 20 managed by the OS 2 is performed. Further, in the trace buffer 30, the task stop status such as time information (time stamp) 31 at which the task 4 starts waiting for an event, the task identifier 32, and the stop cause 34 is written. In step S104, it is indicated that the task 4 is stopped by waiting for an event. After that, the OS 2 of the multitask execution format processes the task 4 in which the stop event has been resolved.

In step S105, it is determined whether or not there is a task 4 whose event waiting has been resolved, and the processing differs depending on the result. It should be noted that this step S105 also comes into the end interrupt processing waiting for an event (flow from B in FIG. 5).

In step S106, processing for selecting the task 4 whose event waiting has been resolved is performed. Next, in step S107, a logging process is performed when the cause of the stop is resolved and the execution is restarted. Here, the total stop time 24, the cumulative number of times 25 by stop cause, and the cumulative time 27 by stop cause of the task management table 20 managed by the OS 2 are updated. In the trace buffer 30, time information (time stamp) 31 when the event waiting in the task 4 is resolved,
Task identifier 32, stop time 33 (time stamp 31
Is calculated by subtracting the last stop time 21 from
Write down the cause of stop 34.

Next, in step S108, logging processing is performed on the stop information that the task 4 has started waiting for the CPU. The process here is the same as the process in step S103, but the information on the cause of the stop in the log is different. Next, in step S109, a process of queuing the corresponding task 4 for waiting for the CPU is performed.
Next, in step S110, it is determined whether or not there is a task 4 in which the event that has stopped the task 4 is resolved, and the processing differs depending on the result. If there is a task 4 whose event waiting is resolved, the processing from step S105 is executed, and if not, step S11 is executed.
Move to 1. In step S111, it is determined whether or not there is a task 4 waiting for the CPU, and the processing differs depending on the result. If there is no task 4 waiting for the CPU, the CPU goes into the idle state. This idle state is resolved when various interrupts are received, and at that time, the process is restarted from B (step S105: event waiting solution) or C (step S111: scheduling interrupt after a fixed time).

If there is a task 4 waiting for the CPU, the process proceeds to step S112 to perform a process of selecting an execution task from the CPU waiting queue. Next, step S
At 113, a logging process for resolving the cause of the stop and resuming the execution is performed. The process here is the same as the process of step S107, but the information on the cause of the stop in the log is different. Next, in step S114, the CPU
The task execution environment for the task 4 which has obtained the execution right is restored. After that, in step S115, it is determined whether or not the stop time of the task 4 exceeds the allowable time. If the stop time exceeds the allowable value, the OS 2 notifies the task 4 of the event in the process of step S130.

As described above, in the first embodiment, in the stop time management of the task 4 in the OS 2,
Information about the cause of the stop and the stop time is recorded before and after the task 4 is stopped. Furthermore, if the stop time of the task 4 exceeds the allowable value, the OS 2 notifies the task of an event, so that the task itself can recognize that the stop time has been exceeded.

Embodiment 2 In the second embodiment, the load status display on the system management monitor 3 in the task monitoring device according to the present invention will be described. In the present invention, the system management monitor 3, which is realized and operates as one task on the computer, provides the following information regarding the stop event of all the tasks 4 executed on the computer. (A) Frequency distribution of stop time for each stop cause in a specific task. (B) Frequency distribution of the stop time according to the stop cause in all tasks in the computer alone. The cause of stopping the task depends on the OS implementation method. For example, UNIX has the following causes.・ Waiting for completion of various input / output devices. Example) Magnetic disk device, terminal device, network device, etc.・ Wait due to exhaustion of work area in OS data. Example) Data exchange between main memory and secondary memory, management of various buffer memories.・ Waiting for event of communication (inside / outside computer) between processes (tasks). Example) Semaphore, message communication between tasks, communication between computers.・ Waiting due to CPU usage restrictions. Example) Waiting for execution depending on the priority of the process (task) or stopping execution due to time limit.・ Spontaneous waiting. Example) Stop the program for a certain period of time using the timer function.

In the system management monitor 3 according to the present invention, the frequency distribution of the stop time for each stop cause is provided at a constant interval in the task 4 which is specified and operates in the entire system. FIG. 5 is a schematic diagram of the frequency distribution of the stop time according to the stop cause provided by the second embodiment. In FIG. 5, the vertical axis represents the number of times task 4 is stopped within the interval, and the horizontal axis represents the stop time per stop. In the second embodiment, the stop time is 10 m.
Rounded up in seconds and reported. The example shown in FIG. 5 shows that most of the noted causes of termination are completed within 10 to 20 msec, but occasionally the duration of the suspension is long.

FIG. 6 is a flow chart showing the flow of processing until the frequency distribution display of the stop time by task stop cause in the second embodiment is displayed. The second embodiment will be described with reference to FIG. 6 below. The operation will be described. Step S
In 201 (corresponding to stop status acquisition means and stop status providing means), all tasks 4 running on the computer system
With respect to, the process of obtaining the current stop status is performed. In addition, at this time, the time when the stop situation is obtained is also recorded at the same time.
It should be noted that the acquisition of the stop status is an atomic process by executing the API provided by the OS2, and the process inside the OS2 will be described later in a fourth embodiment.

Next, in step S202, the system management monitor 3 processes the snapshots collected in step S201 as data for displaying the load status. Next, in step S203, the first embodiment
Step S1 of the flowchart of FIG. 4 described in the processing
It is determined whether or not to process the log written in 03, S107, S108, and S113, and the processing differs depending on the result. If there is no log, no processing is done. However, even when the log is not processed, it is possible to obtain a snapshot display of the status of tasks in the computer system.

If there is a log, the process proceeds to step S204 to read one record of the unprocessed log. Next, the process proceeds to step S205, and it is determined in step S205 whether or not the read record (log) is suitable for processing as data within the interval, based on the time information obtained in the process of step S201. Next, in step S206, the record (log) read in step S204 is processed as load status display data. Next, in step S207, it is determined whether or not there is an unprocessed log, and if any unprocessed log remains, the process returns to step S204. If there is no unprocessed log, the process proceeds to step S208 and step S20.
The data processed in 2 and step S206 is displayed as a frequency distribution chart of the stop time for each stop cause.

As described above, according to the second embodiment, the system management monitor displays the stop time for each cause of the task in a frequency distribution based on the stop status of the task at regular intervals. Since it is set, it is possible to visually grasp the cause of the task stop.

Embodiment 3 In the system management monitor 3 of the present invention, it is possible to obtain the frequency distribution display regarding the stop time for each stop cause of the task 4 described in the second embodiment within the interval and also in the cumulative period. In the third embodiment, in the system management monitor 3 according to the present invention, for the task 4 operating in the specific or the entire system, provision of the frequency distribution of the stop time for each stop cause in the cumulative period will be described.

FIG. 7 is a flow chart showing the flow of the processing for displaying the frequency distribution of the stop time for each stop cause in the accumulation period in the third embodiment. Hereinafter, with reference to the flow chart shown in FIG. explain. The processing in the third embodiment is almost the same as the processing flow in FIG. 6 used for the description in the second embodiment, and only the processing method at the start of processing is different. That is, since the data processed in step S202 shown in FIG. 6 in the second embodiment and used for the frequency distribution display is valid only within the measurement interval (the operation interval of the system management monitor 3), the data at the next measurement interval is used. It is necessary to invalidate this data in the process of accumulating. This is step S300,
This invalidation process is executed at the start of the process. Further, it is necessary to display the frequency distribution so that the data can be invalidated in the processing of step S300 and to superpose the stop information obtained at the new measurement interval with the information obtained at the previous measurement interval. This process is performed in step S308 (corresponding to the process of step S208 in the second embodiment).

As described above, according to the third embodiment,
Since the system management monitor displays the frequency distribution of the stop time for each cause of stopping the task in the cumulative period, it is possible to confirm the load status of the task in a somewhat long range.

Embodiment 4 In the fourth embodiment, the OS
2 is an AP that provides the cause of stopping task 4 and its stopping time
The I means will be described. With the API according to the present invention,
It not only provides the cause of the stop and the stop time in the task 4 that resumed execution, but also provides the stop time from the start of the stop in the task 4 that is currently stopped. By using the means described in the fourth embodiment, it becomes possible to detect the task 4 that has been stopped for a long time.

FIG. 8 shows OS2 in the fourth embodiment.
Is a flow chart showing the flow of processing of means for providing the stop status of task 4, which will be described below with reference to this figure. First, in step S401, processing for prohibiting interrupt processing for the processor is performed. This is a procedure for atomically performing the series of processing shown in FIG. 8 in a short time.

Next, in step S402, a process of prohibiting access to the task management table 20 is performed.
This process is a process for preventing the task management table 20 from being rewritten by another processor by the process when the computer system is a tightly coupled multiprocessor system.

Next, in step S403, time data when this process is executed is collected, and in step S40
In step 4, the time data collected in step S403 and the snapshot information of task 4 are passed to the task (system management monitor 3) requesting this processing. The snapshot status of task 4 is
It is time information from the stop cause and the start of the stop of the task 4 being stopped, based on the task state 28, the time 21 at which the task was stopped last, and the cause 22 that was stopped last.

Next, in step S405, a process of releasing the access prohibition to the task management table 20 locked (prohibited) in step S402 is performed. Finally, in step S406, a process of permitting the interrupt to the processor prohibited in step S401 is performed.

As described above, according to the fourth embodiment, when the system management monitor collects the task stop status, it is possible to acquire the stop status of the task that is being stopped. It can detect tasks inside.

Embodiment 5 FIG. In the fifth embodiment, means for detecting an abnormal state of a computer system and detecting a bottleneck based on the detection result in the system management monitor 3 according to the present invention will be described. In the present invention, in the system management monitor 3, it is possible to set a filter defining an abnormal situation by at least one of the following (1) to (3) regarding the measured task stop time. (1) One stop time exceeds the allowable value. (2) The stop time for each cause of stop exceeds the allowable value. (3) The cumulative downtime exceeds the allowable value. Further, after it is determined that the status is abnormal due to the filter setting, the system management monitor 3 displays the user-defined task 4 desired by the user.
By inputting, it is possible to automate processing such as recording, reporting, avoiding, and recovering abnormal conditions.

FIG. 9 is a flow chart showing a processing flow of the abnormal state detecting means of the system management monitor 3 in the fifth embodiment, which will be described below with reference to this figure. The system management monitor 3 executes the processing shown in FIG. 9 at regular intervals. First, step S
At 501, it is determined whether the definition of the abnormal state has been changed. If the result of determination is that the definition of the abnormal state has been changed, the flow moves to step S510 and the system management monitor I / F means sets the abnormal state defined by the user.

If the definition of the abnormal state has not been changed, the process proceeds to step S502, and the stop status (load status of the computer system collected by the stop status collection means 3a described in the second or third embodiment By referring to the data indicating the (condition), a process of diagnosing whether or not the situation defined as the abnormal state is performed is performed. Next, step S50
In step 3, the result of the diagnosis of the computer system is judged in step S502, and if there is a state judged as abnormal, in step S520 the abnormal state notifying means 3d reports to the task 4 defined by the user. .
Alternatively, the user-defined task 4 for coping with the abnormal state that has occurred is activated by the task input means 3f to solve the problem.

As described above, according to the fifth embodiment, the system management monitor detects an abnormal state of the computer system and notifies the task itself of the action to be taken from the abnormal state, or an appropriate task. Since it is activated, it is possible to take appropriate measures against an abnormal condition. Further, for example, by investigating the cause of the occurrence of the abnormal state, it is possible to know where the bottleneck of the system is.

Embodiment 6 FIG. In the sixth embodiment, a means for providing trace information capable of grasping the state transition of the task 4 in the system management monitor 3 according to the present invention will be described. 10 and 11 are diagrams for explaining the sixth embodiment, and FIG. 10 shows the sixth embodiment.
Transition state diagram (trace diagram) of tasks provided by
FIG. 11 shows the system management monitor 3 in the sixth embodiment.
3 is a flowchart showing the flow of the processing of FIG.

In the system management monitor 3 according to the present invention,
Task 4 based on the data collected by the stop situation collection means 3a
The task state transition display means for providing the state transition of is provided as a trace diagram (transition state diagram) as shown in FIG. The vertical axis of FIG. 10 represents the task state, and in the present invention,
There are three states: execution state, CPU wait state, and stop state.
The horizontal axis represents the time axis. The system management monitor 3
The graph shown in FIG. 10 is created based on the data in the log processing described in step S206 in the second or third embodiment. The original data of this graph is the time of state transition, the states before and after the transition, and the task identifier. At the transition time, draw an arrow from the state before the transition to the state after the transition. Below the arrow, the cause of the stop before and after the transition is described.

As a supplementary explanation of FIG. 10, [-] indicates "execution state" and [CPU] indicates "CP".
"Waiting state", [ipc: internal pro
[cess communication] indicates a “stopped state” due to inter-task communication due to one cause of stoppage. For example, the first state of the time series shown in FIG. 10 is "
The "task 1" transitions from the "execution state" to the "CPU wait state", and the next state indicates that the "task 2" transitions from the "stop state" to the "CPU wait state".

Next, description will be made with reference to the flowchart of FIG. 11 showing the flow of processing in the system management monitor 3 in the sixth embodiment. Step S204 to Step S20
The processing up to 7 is the same as the processing with the same step number in the second or third embodiment. The process peculiar to the sixth embodiment is that the process (state transition display means) in step S601 is newly added and a trace diagram as shown in FIG. 10 is displayed.

As described above, according to the sixth embodiment, the system management monitor is adapted to provide the trace information capable of grasping the state transition of the task. Therefore, in the multitask execution type computer, It becomes easy to understand the dependency relationships between tasks.

Embodiment 7 FIG. In the seventh embodiment, a means for controlling the task 4 input to the computer system in the system management monitor 3 according to the present invention to improve the throughput of the task 4 will be described. The system management monitor according to the present invention comprises 3 and task input means 3f. The user-defined task 4 uses the system management monitor I / F means 5 according to the present invention to acquire task information to be input, and the task input means 3f inputs the task 4 to the computer system. By doing so, the system management monitor 3 can input the task 4 to the computer system based on the task stop situation collected by the task stop situation collecting means 3a, that is, the load status of the computer. Therefore, multiple tasks 4 at a time
Is not input to the computer system, the overhead of the OS 2 is reduced, that is, the throughput of the task 4 is improved.

FIGS. 12A and 12B are flow charts showing the processing flow of this task input means.
12B shows the processing in the user task 4, and FIG. 12B shows the processing in the system management monitor 3. In the seventh embodiment, the load status based on the stop time of the task 4 is determined, and the result determines whether or not the task should be input. Hereinafter, description will be made with reference to the flowchart shown in FIG.

In the user-defined task 4, in step S701, the system management monitor I / F means 5 is passed in order to pass the information of the task to be input to the system management monitor 3.
Then, set that information. The system management monitor 3 is responsible for the actual task input process. In this case, the means of information transmission between different tasks is
There are various methods known to those skilled in the art, which will not be described here.

Steps S710 to S713 shown in FIG. 12B show the task input procedure on the system management monitor 3 side. The system management monitor 3 performs this series of processing because it is necessary to input a new task 4 to the computer system when step S701 is executed by the task input side or after a certain time has passed from the previous processing. That is the case.

In step S710 (task input availability determination means), it is determined whether or not task 4 should be input to the computer system. For example, when the overhead of the OS 2 is increasing due to a long time waiting for the CPU or the like, it is determined that it is not suitable for inputting a new task 4, and the process is terminated. The determination of whether the task input timing is good or bad is made by the same method as the diagnosis routine described in the fifth embodiment. If it is determined that the task can be input, in step S711, step S701 is performed.
The processing for obtaining the information about the task to be delivered, which is passed by the execution of the above, from the system management monitor I / F means 5. Next, in step S712, processing for inputting a task to the computer system is performed. Next, in step S713,
It is determined whether there is another pending task input process, and if there is a pending task input process, the processes from step S710 are repeated for the pending task input request.

As described above, according to the seventh embodiment, since the task is input through the system management monitor, it is possible to limit the input task according to the load status of the computer system. It becomes possible to improve the throughput of the task.

[0069]

As described above, according to the present invention, means for setting the allowable value of the task stop time is provided, and
Since the OS manages the cause and the stop time of the task for each task, when the stop time of the task exceeds the allowable value, the OS can notify the task of the event. Therefore, the task itself can recognize that a delay has occurred in the execution of the task, and thus the task itself can automate the recovery processing such as the processing for dealing with the processing delay.

Since a system management monitor that operates at a predetermined cycle is provided and the system management monitor displays the stop status of the task, that is, the load status in a frequency distribution, the load status of the task can be visually grasped. be able to.

Further, since the stop status of the task is displayed according to the cause of the stop at predetermined intervals, it is possible to grasp the cause of the stop which suddenly occurs.

Further, since the task stop status is displayed by the cumulative value according to the cause of the stop, it is possible to grasp the overall stop status of the task.

Further, since the information of the task in the stopped state can be acquired, it becomes possible to detect the task which has been stopped for a long time, and it is possible to perform the process for the cause of the stop.

Further, the system management monitor is provided with a means for defining an abnormal state of a task, and when the defined abnormal state occurs, the task is reported.
The task itself can take further action.

When the abnormal condition occurs, the system management monitor automatically activates the task for dealing with the abnormal condition, so that the abnormal process can be automatically performed.

Since the means for updating the definition of the abnormal state is provided, the system can be operated flexibly.

Since the system management monitor is provided with means for displaying task state transitions, it is possible to grasp the dependency relationship between tasks, and by analyzing this dependency relationship, the optimum throughput for the system can be obtained. It is possible to operate the task so that

While grasping the load status of the task,
Since the task is input, the system throughput can be improved.

In addition, the permissible value of the task stop time is set in advance, the task stop time is measured, and if the stop time exceeds the permissible value, the task itself is reported. It is possible to provide a computer system capable of taking measures.

Further, since the stop status of the task is displayed in the frequency distribution, it is possible to construct a computer system capable of intuitively grasping the load status of the task.

Further, the abnormal state of the task is defined, and when the defined abnormal state occurs, the abnormal state is dealt with, so that it is possible to provide a computer system for automatically performing abnormal processing.

Further, since the task itself performs the processing for the abnormal state, it is possible to take the optimum measure.

Further, since the processing for the abnormal state is associated with each other in advance and is performed by the task, it is possible to take an automatic countermeasure.

[Brief description of the drawings]

FIG. 1 is a diagram showing an overall configuration of software of a task management device according to the present invention.

FIG. 2 is a diagram showing a state transition of tasks in the first embodiment in time series.

FIG. 3 is a diagram showing details of a task management table and a trace buffer according to the first embodiment.

FIG. 4 is a flowchart showing a processing flow of task down time management in the first embodiment.

FIG. 5 is a diagram showing an example of frequency distribution for each interval displayed by the system management monitor according to the second embodiment.

FIG. 6 is a flowchart showing a flow of processing performed by a system management monitor according to the second embodiment.

FIG. 7 is a flowchart showing the flow of processing performed by the system management monitor according to the third embodiment.

FIG. 8 is a flowchart showing a flow of processing performed by API means in the fourth embodiment.

FIG. 9 is a flowchart showing a flow of processing performed by a bottleneck detection unit included in the system management monitor according to the fifth embodiment.

FIG. 10 is a diagram showing an example of a state transition diagram displayed by the system management monitor according to the sixth embodiment.

FIG. 11 is a flowchart showing a flow of processing performed by a system management monitor according to the sixth embodiment.

FIG. 12 is a flowchart showing the flow of processing performed by a task input means included in the system management monitor according to the seventh embodiment.

FIG. 13 is a diagram showing an output format of a ps command in UNIX.

FIG. 14 is a diagram showing an output format of a vmstat command in UNIX.

[Explanation of symbols]

1 software configuration diagram, 2 operating system, 2a stop allowable time setting means, 2b stop time accumulating means, 2c stop status providing means, 2d stop time excess reporting means, 3 system management monitor, 3a stop status collecting means, 3b stop status display Means, 3c Abnormal state detecting means, 3d Abnormal state notifying means, 3e Task state transition display means, 3f Task input means, 4 User-defined tasks, 5 System management monitor I / F means, 20 Task management table, 30 Trace buffer.

Claims (15)

[Claims] 1. A task monitoring device comprising: a computer in which a plurality of tasks are executed in parallel, wherein an operating system for controlling the execution of the tasks is provided with the following elements. Permissible stop time setting means for setting the permissible value of the stop time of the task, (b) Every time the operating state of the task transits, the cause of the suspension of the task, the suspension time, the suspension for each task according to the transition state Stop time accumulating means for recording the stop cause and the stop status such as the cumulative value of the stop time according to the time and the cause of stopping the task, and (c) corresponding to the stop time of the task and the task recorded by the stop time accumulating means. Comparing means for comparing the allowable value with
(D) Stop time excess reporting means for reporting to the task that the stop time exceeds the allowable value when the comparing means determines that the task stop time exceeds the allowable value. 2. A computer in which a plurality of tasks are executed in parallel, the operating system having the following elements for controlling the execution of the tasks, and (a) each time the operating state of the task transits, its transition: Depending on the status, for each task, the cause of the task's stoppage, stop time, stop time, and
A stop time accumulating means for recording the stop cause for each task stop cause and the stop status such as a cumulative value of the stop time, (b)
A stop status providing means for providing the stop status of the task recorded by the stop time accumulating means, a system management monitor having the following elements and operating at a predetermined interval, and (a) a stop status provided by the stop status providing means. And a stop status display unit for displaying the load status of the task in a frequency distribution based on the stop status of the task collected by the stop status collection unit. Monitoring equipment. 3. The task monitoring device according to claim 2, wherein the stop situation display means displays a task load situation at a predetermined interval. 4. The task monitoring device according to claim 2, wherein the stop status display means displays the load status of accumulated tasks. 5. The task monitoring device according to claim 2, wherein the stop status providing unit provides the stop status including information on the stop status. 6. A computer in which a plurality of tasks are executed in parallel, the operating system having the following elements for controlling the execution of the tasks, and (a) each time the operating state of the task transits, its transition: Depending on the status, for each task, the cause of the task's stoppage, stop time, stop time, and
A stop time accumulating means for recording the stop cause for each task stop cause and the stop status such as a cumulative value of the stop time, (b)
A stop status providing means for providing the stop status of the task recorded by the stop time accumulating means, a system management monitor having the following elements and operating at a predetermined interval, and (a) a stop status provided by the stop status providing means. Stop status collecting means for collecting, (b) abnormal state definition information defining the task abnormal state for each task, (c) an abnormal state for detecting an abnormal state of a task based on the stop state and the abnormal state definition information Detection means,
(D) A task monitoring device comprising: an abnormality notifying unit for notifying that the task is abnormal when the abnormal state detecting unit determines that the task is abnormal. 7. A computer in which a plurality of tasks are executed in parallel, the operating system having the following elements for controlling the execution of the tasks, and (a) each time the operating state of the task transits, its transition Depending on the status, for each task, the cause of the task's stoppage, stop time, stop time, and
A stop time accumulating means for recording the stop factor and the stop status such as the cumulative value of the stop time according to the cause of the stop of the task, (b)
A stop status providing means for providing the stop status of the task recorded by the stop time accumulating means, a system management monitor having the following elements and operating at a predetermined interval, and (a) a stop status provided by the stop status providing means. Stop status collecting means for collecting, (b) abnormal state definition information defining the task abnormal state for each task, (c) an abnormal state for detecting an abnormal state of a task based on the stop state and the abnormal state definition information Detection means,
(D) A task monitoring device comprising: a task inputting means for activating a task for processing the abnormal state when the abnormal state detecting means determines that the abnormal state is present. 8. The task monitoring device according to claim 6 or 7, further comprising a system management monitor interface unit for updating the abnormal state definition information from a user task. 9. The system status monitor according to claim 6, further comprising task status transition display means for displaying a task transition status based on the task suspension status. Task monitoring device. 10. A computer in which a plurality of tasks are executed in parallel, the operating system having the following elements for controlling the execution of the tasks, and (a) each time the operating state of the task transits, its transition: A stop time accumulating means for recording the stop status of each task, the stop time, the stop time, and the stop factor of each task and the cumulative value of the stop time according to the state,
(B) A stop status providing means for providing the stop status of the task recorded by the stop time accumulating means, a system management monitor interface means for holding the input information of the task set by the user task, and the following elements: A system management monitor operating at intervals of (a) a stop status collecting means for collecting the stop status provided by the stop status providing means, and (b) a task status based on the stop status of the task collected by the stop status collecting means. An input availability determination means for determining whether the input is possible,
(C) a task input means for inputting a task based on the task input information obtained from the system management monitor interface means when the task input availability determination means determines that the task can be input apparatus. 11. A task monitoring method including the following steps: (a) a step of setting a permissible value of a stop time of the task for each task, and (b) recording the stop time and the cause of the stop when the task is stopped. A step of: (c) recording a stop time, a cause of the stop, and a cumulative value of the stop time when the task recovers from the stopped state; (d) a comparing step of comparing the stop time with the allowable value; e) A step of reporting to the task that a stop exceeding the allowable time has occurred when it is judged in the comparing step that the stop time exceeds the allowable value. 12. A task monitoring method comprising the steps of: (a) recording a stop time and a cause of the stop when the task is stopped; and (b) a stop time and a stop when the task is recovered from a stop state. A step of recording the task stop status such as the cause and the cumulative value of the stop time, and (c) a step of displaying the task load status in a frequency distribution based on the stop status. 13. A task monitoring method comprising the following steps: (a) a step of defining an abnormal state of each task, (b) when the task is stopped, its stop time,
Recording the cause of the stop, (c) recording the stop status of the task such as the stop time and the cumulative value of the stop cause and the stop time when the task recovers from the stopped state, (d) based on the stop information An abnormal state detecting step of detecting the occurrence of the abnormal state defined in 1., and (e) a step of processing the abnormal state when an abnormality is detected in the abnormal state detecting step. 14. The task monitoring method according to claim 14, wherein the step of processing the abnormal state is a step of reporting to a task corresponding to the abnormal state. 15. The task monitoring method according to claim 14, wherein the step of processing the abnormal state is a step of activating a task for processing the abnormal state.