JPH02201672A - Monitoring device for fault of terminal equipment - Google Patents

Abstract

PURPOSE:To promptly detect the no-response fault of terminal equipment by sampling a necessary time when control data are inputted/outputted between a CPU and the terminal equipment normally, and determining a monitoring time. CONSTITUTION:An input/output control detecting part 11, an input/output necessary time counting part 12, and the monitoring time setting part 13 are provided. At the time of the input/output repeated between the CPU and the terminal equipment in a first mode, the maximum value of the input/output necessary time in a certain range is obtained as a monitoring time maximum value, in a second mode at the time of deciding the normality of the terminal equipment, the necessary time is fetched as sampling data, the monitoring time optimum value is determined with the use of the newest constant value, and it is used for the monitor of the terminal equipment fault. Only when the necessary time exceeds the monitoring time optimum value, the monitoring time maximum value obtained in the first mode is used for the monitoring time instead of the optimum value, and the terminal equipment is finally decided. Thus the terminal equipment can be efficiently decided.

Description

[Detailed Description of the Invention] [Purpose of the Invention (Industrial Application Field) This invention provides a terminal device in various computer systems;
The present invention further relates to a terminal device failure monitoring device for monitoring non-response failures in terminal devices connected to a computer system via a communication line.

(Prior Art) Conventionally, monitoring of non-response failures in terminal devices has relied entirely on software processing of a CPU, which forms the center of a computer system. In other words, conventionally, for all the devices intervening between the CPU and the terminal device, the delay time due to the operation of these devices should be taken into consideration = the delay monitoring time was set in advance on the software, and the response was made in this state. Time was measured using software and compared with the delay monitoring time described above, and if a delay exceeding a set value occurred, it was determined to be a non-response failure and a countermeasure was taken. Here, software-L
The delay monitoring time set in advance must be one that takes into account the load status of the computer system, the operating status of devices intervening between the CPU and terminal equipment, and the operating time specific to the terminal equipment. However, since each maximum value is set with margins taken into consideration, optimization is difficult, and values often end up being larger than necessary. In this case, it takes a great deal of time to detect a non-response failure in the terminal device, making it impossible to take prompt action against the failure. Further, the delay monitoring time has to be reset each time there is an improvement in the hardware or a change in the software. Furthermore, the monitoring time is set based on experience or trial and error, which is complicated and inefficient.

(Problems to be Solved by the Invention) As mentioned above, in the past, the delay monitoring time applied to non-response monitoring of terminal devices was based on the delay time due to the operation of various devices interposed between the CPU and the terminal devices. In consideration, the settings are set with more leeway than necessary through experience and trial and error.
The problem is that optimization is difficult, it takes a lot of time to monitor unresponsive failures in terminal devices, and it is not possible to take prompt measures against failures.Furthermore, with improvements in hardware and changes in software, etc. There is a problem in that the monitoring time must be reset, which is cumbersome. Furthermore, in the past, since non-response monitoring was performed or software processing was performed, there was a problem in that the load on the CPU increased, especially when a large number of terminal devices were connected.

Therefore, the problem to be solved by this invention is that it is possible to automatically set the monitoring time applied to the non-response failure monitoring of a terminal device connected to a computer system, and also to optimize the same time. - To enable non-response failure monitoring to be performed without software processing.

[Structure of the Invention] (Means for Solving the Problems) The present invention provides input/output control detection means for detecting the start and end of control data input/output between a CPU forming the center of a computer system and a terminal device; Input/output required time counting means for counting the time required from the start of the input/output to the end γ (i.e., terminal device response time) according to the detection result of the output control detecting means, and monitoring that serves as a reference for detecting failures in the terminal device. A time is set, and this monitoring time is compared with the time required from the start of input/output counted by the input/output control time counting means,
A determining means for determining a terminal device failure according to the comparison result, and a monitoring time setting means for setting a monitoring time used for determination by this determining means, which retains a maximum value of the monitoring time and monitoring time setting means for internally retaining the required time as sampling data for determining the optimum monitoring time each time the terminal device is determined to be normal by the determining means; In this case, when the terminal device is determined to be normal by the determining means and the sampling data (required time) at that time is larger than the maximum value of the monitoring time held by itself, the monitoring time setting means sets the same data as a new monitoring time. During the second mode for normal terminal device monitoring, the monitoring time setting means sets the latest predetermined number of times each time the start of input/output is detected by the input/output control detection means. The optimum value of the monitoring time determined based on the sampling data (required time) of The internally held maximum monitoring time value is set in the determination means instead of the value, and the determination means makes a final determination that a failure has occurred in the terminal device when the time required from the start of input/output exceeds the maximum monitoring time value. It is characterized by the fact that it is made to do so.

(Function) According to the above configuration, in the input/output between the CPU and the terminal device that is repeatedly performed in the first mode, the maximum time within a certain range is determined as the maximum monitoring time value, and the second mode, each time the terminal device is determined to be normal,
At that time, the required time of step 1 is imported as sampling data, and the latest predetermined number of the imported data is used to determine the optimum monitoring time value, which is used for terminal device failure monitoring, and the required time from the start of input/output is determined. Only when exceeds this optimum monitoring time value, the maximum monitoring time obtained in the first mode is used as the monitoring time instead of this optimum value, and the final determination of a terminal device failure is made. The terminal device fault determination can be performed more efficiently than in the past, in which it was necessary to use a value even larger than the above-mentioned maximum value as the monitoring time from the beginning.

(Embodiment) FIG. 1 is a block diagram of a terminal device failure monitoring device according to an embodiment of the present invention. In the figure, 10 is a terminal device failure monitoring device for monitoring a non-response failure of a specific terminal device (not shown), and 30 is a plug (for example, control data) between a CPU (not shown) and the terminal device. This is an input/output bus 30 of the CPU used for input/output of the CPU. The terminal device failure monitoring device 10 is connected to an input/output bus 30 for use.

In the terminal device failure monitoring device 10, 11 detects that input/output of control data is being performed between the CPU and a specific terminal device on the input/output bus 30, and in particular detects the start and end of control input/output. 12 is an input/output required time counter that counts the corresponding input/output time according to the detection result of the input/output control detector 11; 13 is a delay monitoring time applied to monitoring non-response failures; This is a monitoring time setting section for setting the optimum value of . This monitoring time setting section 13 is
The current optimum value of the monitoring time, the past maximum value, and data on the required time for multiple times sampled to statistically determine the maximum value (required time sampling data) are held.

14 compares the required input/output time of the terminal device counted by the required input/output time counting section 12 with the monitoring time set by the monitoring time setting section 13, and determines whether the terminal device (the required input/output time) is normal. A determining unit 15 determines whether or not the determining unit 14 starts sampling the maximum monitoring time.
A switch 16 for notifying the end is a switch for instructing the monitoring time setting unit 13 to set initial values (initial optimum value and initial maximum value) of the monitoring time. 1
7, the CPU sets the initial value of the monitoring time (
an instruction line 18 connected to the input/output bus 30 for instructing to set the initial optimum value and the initial maximum value;
is a signal input line from the input/output bus 30 to the input/output control detection section 11. 19 is a line for transmitting the detection result of the input/output control detection section 11 to the input/output time counting section 12; 20 is a line for transmitting the detection result of the input/output control detection section 11 to the determination section 14;
This is the line for transmitting information to. 21 is a line for transmitting the counting result of the input/output required time counting section 12 to the monitoring time setting section 13; 22 is a line for transmitting the counting result of the input/output required time counting section 12 to the determining section 14. 23
24 is a line for transmitting the monitoring time setting value in the monitoring time setting section 13 to the determining section 14; 24 is a line for instructing the monitoring time setting section 13 to update the monitoring time setting value from the determining section 14; Reference numeral 25 is a line for transmitting the judgment result of the judgment unit 14 to the input/output bus 30 and notifying the CPU.

Next, sampling of the maximum monitoring time in the configuration shown in FIG. 1 will be explained.

First, when the switch 15 is turned on by a user operation or the like, the determination unit 14 is notified that sampling of the maximum monitoring time is being performed (maximum value sampling mode). In this state, for example, by turning on the switch 16 by a user operation, the monitoring time setting section 13 is instructed to set the initial value of the monitoring time, and the initial optimal value (temporary optimal value) and initial maximum value ( (temporary maximum value) is set and held inside the monitoring time setting section 13 via a path not shown. Here, a target value is appropriately selected as the initial maximum value, and a value smaller than the target value is appropriately selected as the initial optimum value. This initial value setting can also be instructed from the CPU via the input/output bus 30 and line 17, in which case the set value is given to the monitoring time setting section 13 via the input/output bus 30.

In the above state, when an input/output request (not control data) from the CPU to a specific terminal device appears on the input/output bus 30, the input/output control detection unit 11 detects the state of the signal input line. 18. In/Out/J
When the control detection unit 11 detects the occurrence of an output request from the CPU to the terminal device N, that is, the start of input/output, the control detection unit 11 detects the start of input/output from the CPU to the terminal device N, and then outputs the required input/output time counting unit 1'2 (for x=rL). Start counting the time required for input/output of the controller controller.
It instructs the determination unit 14 to start determining whether the required time for outputting the control data is normal or not through line 2o.

When the determination unit 14 receives an instruction to start determination from the input/output control detection unit 11 when the gargle is on, the determination unit 14 first sets the monitoring time setting unit 13 by 2 seconds and sets the initial optimal value as the monitoring time. request via line 24 to set the .

As a result, the monitoring time setting section 13 sets the initial optimal value as the monitoring time to the determination section 14 via the ζ line 23.The input/output station history/hourly number section 12 sets the initial optimum value as the monitoring time to the input/output control detection section 11.
When instructed to start counting, it starts counting the time required for inputting and outputting control data, and outputs the counting result (required time) to the monitoring time setting unit 1 knee I via line 21 at fixed time intervals, for example. At the same time, it is output to the determination unit 14 via the input 22. The determining unit 14 constantly compares the required time outputted from the input/output required time counting unit 12 and the monitoring time (here, the initial optimum value) set by the monitoring time setting unit 13. When the control data input/output ends in this state and the input/output control detection section 11 detects this, it notifies the input/output time counting section 12 and the determination section 14 of the end of the input/output. The required input/output time counter 12 stops counting the required input/output time upon receiving the input/output end notification from the input/output control detection unit 11. Further, when the determination unit 14 is notified of input/output completion from the input/output control detection unit 11, if the required time≦monitoring time, the determination unit 14 terminates the determination as normal termination, and
Sampling of the current required time (input/output time required until normal termination) is sent to the monitoring time setting unit 13 on the line 24.
Direct through. As a result, the monitoring time setting section 13
The required time (current required time) given by the input/output required time counting section 12 is stored in an internal ring buffer (not shown), for example. This ring buffer can store (data) for a maximum of n required ports and spaces, and its area is used in a ring shape (cyclically).

On the other hand, if the required time>monitoring time is reached before the input/output completion is detected by the input/output control detection unit 11 (before the control data input/output is completed), the determination unit 14 determines that the monitoring delay state has entered. Judgment and indicate that on line 25,
The CPU is notified via the input/output bus 30. The determination unit 14 also requests the monitoring time setting unit 13 via line 24 to set the maximum value as the monitoring time setting value. As a result, the monitoring time setting unit 13 sets the internally held maximum monitoring time value (in this case, the initial maximum value, that is, 1).
The determination unit 14 via the line 23 uses the target value) as the monitoring time.
Set to .

When the monitoring time is set by the monitoring time setting unit 13, the determining unit 14 uses the newly set monitoring time (maximum value) to compare it with the required time given from the input/output required time counting unit 12. continue. After that, when the control data input/output is normally completed, the determining section 14 stops the determination and instructs the monitoring time setting section 13 to sample the current required time via the line 24. On the other hand, if the newly set monitoring time (maximum value) becomes the required time>monitoring time, the determination unit 14
A notification is sent to the PU via the line 25 and the input/output bus 30 to continue the comparison/judgment process (monitoring).

The determination unit 14 also requests the monitoring time setting unit 13 to provisionally set a new maximum value as the monitoring time setting value (setting a value several times the maximum value held by the monitoring time setting unit 13). do. Thereby, the monitoring time setting unit 13 sets the requested value (a value several times the maximum value) as the monitoring time in the determining unit 14 via the line 23.

When the monitoring time is set by the monitoring time setting unit 13, the determining unit 14 uses the newly set monitoring time (a value that is several times the maximum value) and is given from the input/output time counting unit 12. Continue to compare the required time. After that, if the control data input/output is completed normally, the determination unit 14 stops the determination, and also sends the monitoring time setting unit 13 a sample of the current required time and the information held in the monitoring time setting unit 13. Instructs to update the maximum monitoring time to the current required time. As a result, the monitoring time setting unit 13 stores the current required time in the ring buffer, and
The current required time is held as the new maximum monitoring time in place of the previous maximum monitoring time. On the other hand, if the required time>monitoring time is reached even with the provisional setting value (a value several times the maximum value) as a reference, the judgment unit 4 sends the line 25 and input/output bus 30 to the CPU. A terminal device non-response failure is notified via the terminal device.

The above operations are repeated every time input/output control is performed. Then, when a series of input/output controls are completed, the switch 15
When it is turned off, the determination section 14 is notified that sampling of the maximum monitoring time has ended, and the maximum value held in the monitoring time setting section 13 at that point becomes the maximum sampled monitoring time.

Next, normal operation after sampling the maximum monitoring time will be explained.

First, in normal operation, the operations of the input/output control detecting section 11 and the input/output required time counting section 12 are the same as in the case of maximum value sampling described above, or the operations of the monitoring time setting section 13 and the determining section 14 are as follows. Different like. That is, determination section 1
4 causes the monitoring time setting unit 13 to set the optimum monitoring time value when the input/output control detection unit 11 instructs the start of determination via the line 20 while the switch 15 is off (normal mode). request via line 24. As a result, the monitoring time setting unit 13 statistically predicts the current optimal monitoring time value based on the sampling data of the latest n times of required time stored in the ring buffer, and monitors this predicted optimal value. The time is set in the determination unit 14. In this state, the determination unit 14 uses the monitoring time set by the monitoring time setting unit 13 (in this case, the optimum value predicted by the monitoring time setting unit 13) and the input/output time counting unit 12.
Compare and judge the required time given by .

If the control data input/output ends normally and the required time≦monitoring time, the determining unit 14 cancels the determination and instructs the monitoring time setting unit 13 to sample the current required time. , nothing is notified to the CPU. On the other hand, if the required time>monitoring time is reached, the determining unit 14 notifies the CPU that a non-response failure has occurred in the terminal device and that the monitoring delay state has been entered, and further notifies the monitoring time setting unit 13 A request is made via line 24 to set the maximum monitoring time.

The monitoring time setting unit 13 sets the current maximum value (the maximum value obtained by maximum value sampling) to the determining unit 14 via the line 23 when there is a request from the determining unit 14 to set the maximum monitoring time value. The determining unit 14 continues to compare the required time given by the input/output required time counting unit 12 using this maximum monitoring time. After that, if the control data input/output ends normally, the determination unit 14 stops the determination and instructs the monitoring time setting unit 13 to sample the current required time,
Cancel failure notification to CPU. On the other hand, if the required time>monitoring time is reached even with the newly set maximum value as a reference, the determination unit 14 notifies the CPU of the occurrence of a failure again.

Note that since the terminal device failure monitoring device 1o in FIG. (l may be provided corresponding to each terminal device. Also, the input/output control detecting section if identifies the terminal device that is performing input/output with the CPU, and the terminal device name is sent to the determining section 14 or the monitor In addition to providing a function to notify the time setting unit I3, the monitoring time setting unit 13 retains data for failure monitoring (optimal value, maximum value, sampling data for statistics) for each terminal device, and among the retained data, By providing the determination unit 14 with a function of setting data corresponding to the terminal device currently being input/output, it is possible to monitor a plurality of terminal devices for failure with one terminal device failure monitoring device.

[Effects of the Invention] As described in detail below, according to the present invention, the following effects can be achieved.

■ Since the monitoring time is determined by sampling the time required when control data input/output is normally performed between the CPU and the terminal device, there is no data between the CPU and the terminal device. It is possible to optimize the monitoring time specific to a terminal device without being aware of the delay time of the device or the operation time specific to the terminal device, and it is possible to quickly detect a non-response failure of the terminal device.

(2) The load on the CPU in monitoring failures in terminal devices is reduced, and wasteful response monitoring processing by software processing can be eliminated.

■ Monitoring time can be easily reset due to hardware improvements and software changes.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a terminal device failure monitoring device according to an embodiment of the present invention. 10... Terminal device failure monitoring device, 11... Input/output control detection section, 12... Input/output required time counting section, 13...
Monitoring time setting section, 14...judgment section, 15. Te...
・Switch, 30...Input/output bus. Applicant's agent Patent attorney Takehiko Suzue

Claims (1)

[Scope of Claims] Input/output control detection means for detecting the start and end of control data input/output between a CPU forming the center of a computer system and a terminal device, and according to the detection result of the input/output control detection means,
An input/output time counting means for counting the time required from the start to the end of the input/output, and a monitoring time serving as a reference for detecting a non-response failure of the terminal device are set, and this monitoring time and the input/output control time counting means are set. determining means for comparing the time required from the start of input/output counted by the means and determining a failure of the terminal device according to the comparison result; and setting the monitoring time used for the determination by the determining means. monitoring time setting means for holding a maximum monitoring time value, and sampling the required time for determining the optimum monitoring time each time the terminal device is determined to be normal by the determining means; It is equipped with a monitoring time setting means that is internally held as data, and a mode setting means for switching and setting between a first mode for determining the maximum monitoring time value and a second mode for performing normal terminal device monitoring. When the first mode is set by the mode setting means, the monitoring time setting means determines that the terminal device is normal by the determining means, and the required time is the sampling data at that time. is larger than the maximum monitoring time held by itself, the same required time is held as a new maximum monitoring time, and if the second mode is set by the mode setting means, the monitoring time setting is set to the second mode by the mode setting means. The means sets, in the determination means, an optimum value of the monitoring time determined based on the latest sampling data for a predetermined number of times, each time the start of the input/output is detected by the input/output control detection means; If the determination means determines that the required time from the start of input/output counted by the input/output time counting means exceeds this optimal value for monitoring time, the optimal value for monitoring time is internally held instead of the optimal value for monitoring time. The determination means sets the maximum value of the monitoring time set by the monitoring time setting means to the maximum value of the monitoring time set by the monitoring time setting means. A terminal device failure monitoring device characterized in that, when a maximum value is exceeded, a final determination is made as to whether a failure has occurred in the terminal device.