JPS63231538A - Fault detecting system for input/output device - Google Patents

Abstract

PURPOSE:To attain the early detection of a trouble without giving an excessive load to a system by activating automatically a task for testing the access only when an access frequency to an input/output device is lower than a reference value. CONSTITUTION:At a central processing device 1, a use frequency measuring means 6 to measure the use frequency of an input/output device 3 at an act side and a comparing means 7 to compare the measured use frequency with a reference value, and only when the measured use frequency is lower than the reference value, input/output devices 3 and 3a are operated by a trouble detecting means 5, the data are transferred, and the presence and absence of the fault are detected. Since a fault detecting means 3 is activated and an input/output action is executed only when the use frequency of the input/output device 3 is lower than the reference value, the access for the trouble detection is executed considering the time when a general processing request except the fault detection is little. Thus, the time from when the trouble occurs at the input/output device at a standby side until the fault is detected can be shortened without increasing the burden of the system.

Description

[Detailed Description of the Invention] [Summary] This is an input/output device failure detection method that detects failures in the standby input/output device due to reasons such as infrequent access or uneven access to parts. To solve the problem of delays, by automatically starting the access test task only when the access frequency to the input/output device is below a standard value, failures can be detected early without placing an extra load on the system. What made it possible.

[Industrial application field]

The present invention relates to a failure detection method for input/output devices that can be randomly accessed.

In general, in computer systems, especially in the field of electronic exchange, all devices are duplicated (standby) in order to achieve 24-hour non-stop operation, but while the standby side devices are important, they are accessed less frequently. Therefore, there is a problem that failure detection is delayed.

For this reason, early detection of failures in devices or parts within devices that are accessed less frequently is required.

[Conventional technology]

FIG. 6 shows an example of a conventional fault detection method.

In FIG. 6, the task scheduler 61 controls the T10 access routine 62 at a period of, for example, 8 msec.
Start it up. The purpose of periodic activation is to maintain real-time performance in exchange processing.

At each cycle startup, if there is a processing request in the waiting queue 63, access is made to the duplicated random access input/output devices 64 and 64a such as external storage devices, and data is written in accordance with the processing request. Or reading is performed.

Apart from the access by the I10 access routine 62,
In order to periodically check whether there are any failures in the devices on the active side and standby side, the task scheduler 61, for example, once a day at a certain time when the load on the system is thought to be light, such as in the middle of the night or early morning. , activates the failure detection access request queue 65 to input access requests for all areas of the input/output devices on the active side and the standby side to the waiting queue 63. The I10 access routine 62 detects access requests to all the areas mentioned above by its periodic activation, and detects whether there is a failure in the active side or standby side devices.

Note that 66 is a general task that generates a processing request that is input to the waiting queue 63.

[Problem that the invention seeks to solve]

According to the above-mentioned conventional method, the access tasker 6 for failure detection
The activation of 5 is limited to a predetermined time, and the number of times per day is also limited. For this reason, even if a failure occurs in the standby device, it takes time to detect it, and if a failure occurs in the active device during this time and the switch is switched to the standby device, the standby device There is a problem in that a situation may occur in which the device is also at fault.

To avoid this, 1 of access task 65 for failure detection
There is also the problem that increasing the number of activations per day increases the load on the system and limits processing requests from the general task 65.

In addition, the startup time of the access tasker 65 for failure detection is fixed at night or early in the morning when the load on the system is considered to be low, but even at night or early in the morning, general processing requests may be concentrated due to emergency situations, etc. However, there is a problem in that the above-mentioned fixed time is not always a time when the load on the system is light.

[Means for solving problems]

FIG. 1 is a block diagram of the principle of the present invention.

In FIG. 1, the early failure detection method for input/output devices according to the present invention is such that a central processing unit 1, a main storage device 2, and a duplex input/output device 3゜3a capable of random access are connected via a bus 4. The central processing unit 1 is connected to the input/output device 3.
In the system, the central processing unit 1 includes a usage frequency measuring means 6 that measures the frequency of use of the input/output device 3 on the act side, and uses the measured usage frequency as a reference value. A comparison means 7 is provided for comparison, and only when the measured frequency of use is lower than a reference value, the failure detection means 5 operates the input/output device 3゜3a to transfer data and detect the presence or absence of a failure. That's what I do.

[For production]

Since the failure detection means 3 is activated to perform input/output operations only when the frequency of use of the input/output device 3 is lower than the standard value, access for failure detection is performed in a time when general processing requests other than failure detection are low. Therefore, the time from when a failure occurs in the standby input/output device to when the failure is detected can be shortened without increasing the burden on the system.

[Embodiment] FIG. 2 is a block diagram showing an early failure detection method for an input/output device according to an embodiment □ of the present invention.

In FIG. 2, 2I is a time scheduler, 22 is an I10 access routine, 23 is a waiting queue, 24 is an act side random access input/output device, 24a is a standby side random access input/output device, and 25 is an access task for failure detection. , 26 are general tasks, which are the same as those in the conventional system shown in FIG. 27 is a processing request counter, 28 is a periodic activation table, and 29 is an access address information table.

FIG. 3 is a diagram showing an example of the contents of the periodic activation table 28.

In FIG. 3, the horizontal axis indicates the type of task, with 1.0.0 in the top row. ...are tasks A and B.

In FIG. 0, which is a flag indicating the presence or absence of a startup request for C1..., the flag for task A is "1", indicating that there is a startup request for task A. The vertical axis indicates time; for example, the activation cycle of task A is 1001. That is, when the activation flag of task A becomes 11'', task A is activated at time jl+t4.Similarly, the activation cycle of task B is 0101-...
, the activation cycle of task C is 1010...

FIG. 4 is a flowchart showing the operation of the I10 access routine 22, and FIG. 5 is a flowchart showing the updating operation of the periodic activation table 28.

The operation of the embodiment of FIG. 2 will now be described with reference to FIGS. 3-5.

For example, the task scheduler 21 has a period of 8 msec,
The I10 access routine 22 is activated. When starting each cycle, if there is an access request from the general task 26 in the waiting queue 23, the I10 access routine 22 takes out the processing request and sends it to the random access input/output device 24, 24a such as an external storage device. Data is written or read according to processing requests. When a processing request is executed, the processing request counter 27 is incremented (see steps 41 to 43 in FIG. 4).

On the other hand, as shown in FIG. 5, the access routine 22 extracts the value of the processing request counter 27 at regular intervals (step 51), and determines whether this value is less than the reference value (step 52). . If it is below the reference value, counter 2
7 in step 53), and then the fault detection access task in the periodic activation table 28, for example A
The start flag of is set to "l" (step 54).

When the task scheduler 21 detects that the activation flag of the failure detection access task A in the periodic activation table 28 is turned on, it activates the failure detection access task A 25 according to the activation cycle of the task A in the periodic activation table 28 . In this case, each time the fault detection access task 25 is activated, the addresses from the access/address information table 29 are sequentially changed to access the entire area of the input/output device W24/24a and check for the presence or absence of a fault. To detect.

According to this embodiment, failure detection processing is executed only when the load on the system is light, so that failures in input/output devices can be detected early without increasing the load on the system.

〔Effect of the invention〕

As is clear from the above description, according to the present invention, the fault detection means is operated only when the usage amount of the input/output device is less than the reference value, so that the fault detection of the input/output device is performed by the system. It is carried out immediately when the load is light,
High reliability of the system is thus achieved.

[Brief explanation of drawings]

FIG. 1 is a block diagram of the principle of the present invention. FIG. 2 is a block diagram showing an early failure detection method for an input/output device according to an embodiment of the present invention. FIG. 3 is an example of the contents of the periodic activation table in FIG. 2. 4 is a flowchart showing the operation of the I10 access routine, FIG. 5 is a flowchart showing the updating operation of the periodic activation table, and FIG. 6 is a block diagram showing an example of a conventional failure detection method. 1...Central processing unit, 2...Main storage device, 3...
- Input/output device, 4... Bus, 5... Fault detection means, 6... Usage frequency measurement means, 7... Comparison means.

Claims (1)

[Claims] A central processing unit (1), a main storage device (2), and a duplex input/output device (3, 3a) capable of random access.
are connected via a bus (4), and the central processing unit (1) is equipped with a failure detection means (5) for detecting a failure of the input/output device (3), wherein the central processing unit (1), the input/output device (3, 3a
) and a comparison means (7) to compare the measured frequency of use with a standard value, and when the measured frequency of use is below the standard value. The failure detection means (5) detects the input/output device (3) only when
, 3a) to transfer data and detect the presence or absence of a failure.