JPH0433035A - Error monitoring system - Google Patents

Abstract

PURPOSE:To easily correct a fault point after separating it by providing a means to discriminate a sub system, for which an error generating device belongs, based on a system configuration definition table, error recovery process code and threshold value control table for the unit of the sub system, and a means to control the number of times for generating the faults for the unit of the doubtful point. CONSTITUTION:In the form of tables 1 and 2, the doubtful point weighted corresponding to whether the device is in a high-order or low-order, and a threshold value to instruct the exchange of the doubt full point are provided for each ERP code generated and outputted to a hardware when the error is generated during an input / output operation. On the other hand, the number of times for generating the error for each doubtful point is counted and controlled as a statistical information table 3, and when the error is generated, the doubtful device can be specified based on the outputted ERP code and the address of the error generating device. Further, when it is recognized that the number of times for generating the error in the doubtful device up to the moment exceeds the threshold value, it is judged to exchange the device, and it is automatically instruct to exchange the doubtful device. Thus, the fault point can be easily separated and the fault can be corrected without fail.

Description

[Detailed Description of the Invention] [Summary] This invention relates to a method for monitoring errors that occur in a computer system equipped with an error recovery procedure (ERP) mechanism that attempts to analyze and recover from device errors that occur during input/output operations. , The failure point can be easily isolated without human intervention, and the computer system can fail.

The aim is to provide an error monitoring method that can accurately repair failures, and weights suspect locations based on the system configuration definition table and the error recovery procedure code (ERP code) created by the computer system hardware. The dimension with the larger weight is treated as the upper dimension, and the dimension with the smaller weight is managed as the lower dimension. In addition, multiple error monitoring times are set corresponding to the weighted suspect locations, and the set multiple error monitoring times are A threshold control table that sets the number of error occurrences to perform threshold control in accordance with the error monitoring time; and a statistical information table that stores the number of error occurrences for the number of suspect points weighted above for each weight. When an error occurs during an input/output operation, the error recovery procedure (ERP) mechanism is generated.

The error recovery procedure (ERP) mechanism generated and outputted by the error recovery procedure (ERP) mechanism identifies the error generation device based on the output error recovery procedure (ERP) code, the error generation device address, and the system configuration definition table. E.R.
P) The above threshold control table using the code and the error generating device address as keys.

Update the number of errors in the corresponding part of the statistical information table by referring to the statistical information table, and if the updated number of errors exceeds the threshold set in the threshold control table, Configure for error notification.

[Industrial application field]

The present invention relates to a method for monitoring errors that occur in a computer system that includes an error recovery procedure (ERP) mechanism that attempts to analyze and recover from errors in devices that occur during input/output operations.

In recent years, as the need for highly reliable computer systems has increased, it has become increasingly important to prevent system failures and failures, that is, to ensure that recovery is possible if the total number of intermittent failures occurs exceeds a specified number of times. However, it is becoming increasingly important to perform maintenance before catastrophic failure occurs.

Further, by notifying maintenance personnel of the suspected component (unit, device, etc.) based on the error event that has occurred, the time required to repair the fault can be shortened.

For this reason, there is a need for an error monitoring system that can effectively notify maintenance personnel of suspect parts when errors related to input/output operations occur.

[Prior art and problems to be solved by the invention] FIG. 3 is a diagram illustrating a conventional error monitoring system.

Traditionally, during O3 (operating system) operation,
When a hardware error related to input/output operations occurs, maintenance personnel can start a test program, analyze the error list, visually isolate the fault by looking at the output log edit list, and decide whether to replace the part. was going on.

Therefore, there is a ratio of 1:n between the cause of the error and the suspected location.

That is, when a plurality of suspected locations are considered for one error occurrence factor, it is difficult to identify the suspected location.

In addition, errors can be corrected by software recovery processing, and the frequency of occurrence of errors that requires monitoring the number of occurrences varies, such as cases in which many errors occur in a short period of time or cases in which they occur at regular intervals. at regular intervals (for example, 3
It has been difficult to perform so-called threshold control, in which the number of error occurrences exceeding a predetermined threshold is detected during the 0 minute period, and the suspected location is identified.

In view of the above-mentioned conventional drawbacks, the present invention provides a manual system for errors that occur in a computer system equipped with an error recovery procedure (ERP) mechanism that attempts to analyze and recover from errors in devices that occur during input/output operations. It is an object of the present invention to provide an error monitoring method that can easily isolate the location of a fault without going through the steps, and can accurately repair failures and failures in a computer system.

[Means to solve the problem]

FIG. 1 is a diagram explaining the principle of the present invention, in which (a) shows a diagram of the basic configuration, and (b) shows an outline of an error monitoring process flow.

The above problems are solved by an error monitoring system configured as follows.

This is a system for monitoring errors that occur in a computer system that is equipped with an error recovery procedure (ERP) mechanism that attempts to analyze and recover from device errors that occur during input/output operations, the system configuration definition table l.

For each error recovery procedure code (ERP code) created by the hardware of the computer system, the suspected locations are weighted, the one with the larger weight is treated as the upper dimension, and the one with the smaller weight is managed as the lower dimension. ,
A threshold control table 2 in which a plurality of error monitoring times are set corresponding to the weighted suspect locations, and a number of error occurrences for performing threshold control is set in accordance with the plurality of set error monitoring times. .

Statistical information tables 3, 3a that store the number of error occurrences corresponding to the number of weighted suspect locations, according to weighting.
+3b is provided, so that when an error occurs during input/output operation, the error recovery procedure (ERP) generated and output by the above error recovery procedure (ERP) mechanism is
An error recovery procedure (ERP) code generated and output by the error recovery procedure (ERP) mechanism that identifies the error generation device based on the error generation device address, the error generation device address, and the system configuration definition table 1; Using the error generating device address as a key, refer to the HM control table 2 and the statistical information tables 3.3a and 3b, and update the number of errors in the corresponding parts of the statistical information tables 3, 3a and 3b, If the updated number of errors exceeds the threshold set in the threshold control table 2, the system is configured to issue an error notification.

[Operation] That is, according to the present invention, as shown in FIG. 1(a),
An error recovery procedure (ERP) that attempts to analyze and recover from a device error that occurs during input/output operations.A system configuration definition table is used to monitor errors that occur in a computer system equipped with an error recovery procedure (ERP).

Error recovery procedure (ERP) code (hereinafter referred to as ERP code) and 9. For example, a subsystem in which suspect points are weighted depending on whether they are upper or lower devices and thresholds for multiple monitoring time units are defined. means for determining to which subsystem the error generating device belongs based on a unit threshold value control table;

Means for creating statistical information tables for the number of weights (dimensions) mentioned above and managing the number of failure occurrences, etc. for each suspect location;
A means for checking the frequency of error occurrence within the specified monitoring time is provided, and error monitoring is performed as follows.

That is, as shown in Figure (b), first, when a hardware failure occurs during O3 operation, O3 assembles error log information and records it in an error logging file.

Taking this event as a trigger, the control unit calls the log analysis unit using the ERP code and the error generating device address as arguments.

The log analysis unit determines which subsystem this device belongs to based on the system configuration definition table, and reads statistical information (multidimensional) from the statistical file onto the memory.

Using the ERP code and error generating device address handed over from the control unit as keys, time monitoring control and weighting of suspect locations are performed based on the data of the threshold control table and statistical information table (multidimensional) of the relevant subsystem. Perform threshold control and update the statistical information table.
Write this to the statistics file.

If the number of errors for each ERP code that occurs in the device exceeds the threshold defined in the threshold control table above, a message will be output and the operator will be notified of the suspected location (device).
Request a replacement. Also, for "unconditional replacement" errors that do not require threshold control, a message from the suspect device is output in the same way, and if the threshold control table indicates multiple suspect locations, has the next highest weight (i.e., ) depending on the device address generated by the hardware.

It operates to find the suspected location (secondary).

That is, in the present invention, when an error occurs during an input/output operation, for each ERP code generated and output by the hardware, a suspect weight is assigned depending on whether the device is located at a higher level or lower level. The location and the threshold value (corresponding to the monitoring time) indicating that it is better to replace the suspect location are prepared in the form of a table, and the number of error occurrences for each suspect location (corresponding to the above monitoring time) is prepared. It is counted and managed as a statistical information table, so when an error occurs, the suspect device can be identified based on the output ERP code and the address of the device where the error occurred, and the suspect device's past history can be identified. When it is recognized that the number of error occurrences exceeds a threshold, it is determined that it is appropriate to replace the suspect device and automatically instructs maintenance personnel to replace the suspect device, thereby preventing system failure.・It has the effect of being able to accurately repair failures in the form of preventive maintenance.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings.

The above-mentioned FIG. 1 is a diagram for explaining the present invention in detail, and FIG.
The figure shows an example of the present invention, in which (a) shows an example of the configuration of a threshold control table, (bl) to (b3) show examples of the configuration of a statistical information table, (C1), (C
2) shows a detailed processing flow of threshold control in the magnetic tape (MT) subsystem.

In the present invention, a system configuration definition table 1,
Threshold control table 2 that specifies threshold conditions for each ERP code and suspect locations weighted according to upper and lower devices; Statistical information tables 3, 3a, and 3b are provided to record the number of times, and when an error occurs during input/output operation, the above system configuration definition table 1 is provided.
With reference to
3b, the number of error occurrences recorded in the statistical information tables 3.3a and 3b exceeds the threshold specified by the threshold control table 2, or the suspected unconditional exchange. A means for instructing replacement of a part is a necessary means for carrying out the present invention. Note that the same reference numerals indicate the same objects throughout the figures.

Hereinafter, the error monitoring system of the present invention will be explained with reference to FIG. 2 while referring to FIG.

In this embodiment, a magnetic tape (MT) subsystem is taken as an example, but it goes without saying that the present invention is not limited to this.

First, the threshold control table 2 shown in FIG.
For each code, a threshold condition and, as shown in the figure, suspect locations weighted in correspondence with upper-level equipment (MTU) => lower-level equipment (TAPE) are shown in the form of a table.

The threshold conditions include the case of unconditional exchange and the monitoring period, for example, defining a threshold of the number of errors counted at 30 minute intervals or monthly intervals, and if there is an error exceeding the second threshold. , it is recommended that the suspected part be replaced.

(bl) to (b3) Statistical information tables 3, 3 shown in the figures
a.

3b is configured to add the current error in accordance with the suspect location weighted in the threshold control table 2 using threshold control, which will be described later.

First, as shown in the schematic operation flow in Figure 1(b), when a hardware failure occurs during O3 operation, the O3
8 assembles error log information and records it in an error logging file. Taking this event as an opportunity, the control unit started the ERP
Call the log analysis section with the code and error generating device address as arguments.

The log analysis unit determines which subsystem this device belongs to based on the system configuration definition table 1, and extracts statistical information (multidimensional) tables 3.3a, 3 from the statistical file.
The contents of b are read into a memory (not shown).

Then, the subsystem (
In this embodiment, threshold control of the MT subsystem), table 2, and statistical information table (multidimensional) 3.3a, 3
Time monitoring control based on the data of b.

Then, threshold control is performed, characterized by weighting of suspected locations, the number of errors in the statistical information tables 3.3a and 3b is updated, and this is written in the statistical file.

If the number of errors recorded in the statistical information tables 3, 3a, and 3b exceeds the threshold set in the threshold control table 2, a message is output to request the operator to replace the device.

The detailed operation of the threshold value control will be explained below using the operation flows shown in FIGS. 2(cl), (C2), and (C3).

When first referring to the −-dimensional statistical information table 3 using the ERP code and error generating device address output from the control unit as keys, the corresponding E
It is checked whether there is an item that matches the RP code and the address after 9 weighting. If not, a new corresponding item is created, but if there is, the threshold control table 2 is referred to for the corresponding item and the threshold value Check the presence or absence of the parameter (threshold value) for control, and if there is none, that is, "unconditional exchange"
If specified, the maintenance person (operator) is notified of the suspected location (3) specified by the threshold control table 2. (Steps 1O111, 12, 2 in Figure 2 (cl)
0) If the above threshold parameter ■ is specified, the monitoring start time (previously set monitoring start time) is set to the check range time (for example, 30 minutes) for the time when the error occurred. , January, etc.).
It is checked whether the time obtained by adding the check range time to (2) has passed.

Here, if the error occurrence time does not exceed the sum of the check range time, the error is recognized as a regular error, and the current number of errors in the statistical information table 3 is added by 1", but If the time exceeds the sum of the check range time, the above error is treated as a temporary error, and the error count corresponding to the check range time that has been counted up to that point is cleared, and the time is Set the monitoring start time ■, and use that time as the monitoring start time ■ to perform new periodic error monitoring. (Refer to steps 13, 14, and 15 in Figure 2 (C1).) The updated number of errors in the statistical information table 3 is compared with the threshold specified in the threshold control table 2, and if it exceeds the threshold, the suspected location is determined to have regular errors. It has been determined that this is a location that may go down someday, and the maintenance personnel (operator)
and notify them to replace the suspect part. (Figure 2 (
cl) step 16.21) Add two-dimensional table 3 to the statistical information table 3 of the - dimension above.
If it is specified that there is a, the same process as for the above-mentioned -dimensional table 3 is performed on the two-dimensional table 3a using the same procedure as above.

Similarly, if it is specified that the two-dimensional table 3a has a three-dimensional table 3b, the above-mentioned -
Execute the same process as for dimension table 3. (Figure 2 (C2
), (C3)) As described above, the present invention includes the system configuration definition table 1, the threshold conditions for each ERP code, and the threshold control table that indicates suspected locations that are weighted according to the order of upper device → lower device. 2. A multi-dimensional statistical information table that records the number of errors occurring at the suspect location for each ERP code and for each of the above threshold conditions.3.
3a and 3b, and when an error occurs during input/output operation, refer to the system configuration definition table 1 above,
Identify the suspected location, refer to the threshold control table 2 and the statistical information tables 3.3a and 3b, update the number of errors recorded on the statistical information tables 3, 3a and 3b, and A feature of this system is that the operator is instructed to replace suspected locations where the number of error occurrences exceeds the threshold specified by the threshold control table 2.

〔Effect of the invention〕

As described above in detail, the error monitoring method of the present invention is applicable to a computer system equipped with an error recovery procedure (ERP) mechanism that attempts to analyze and recover from errors in devices that occur during input/output operations. Suspect points are weighted for each error recovery procedure code (1!RP code) created by the system configuration definition table and the hardware of the computer system, and the one with the larger weight is considered the upper dimension, and the one with the smaller weight is managed as a lower dimension, multiple error monitoring times are set corresponding to the weighted suspect locations, and error threshold control is performed in response to the set multiple error monitoring times. A threshold@control table that sets the number of occurrences and a statistical information table that stores the number of error occurrences for the number of suspected points weighted above according to weighting are provided, so that when an error occurs during input/output operation, the above error The error recovery procedure (ERP) mechanism identifies the error generation device based on the error recovery procedure (ERP) code generated and output by the error recovery procedure (ERP) mechanism, the error generation device address, and the system configuration definition table, and executes the error recovery procedure (ERP) mechanism. Using the error recovery procedure (ERP) code generated and output by The number of errors is updated, and if the updated number of errors exceeds the threshold set in the threshold control table, an error notification is issued. In contrast, even when multiple suspect locations are considered, the fault location can be easily isolated without human intervention, and system failures and
This has the effect of accurately repairing failures in the form of preventive maintenance.

[Brief explanation of drawings]

FIG. 1 is a diagram explaining the principle of the present invention. FIG. 2 is a diagram showing an embodiment of the present invention. FIG. 3 is a diagram explaining a conventional error monitoring method. It is. In the drawing, 1 is the system configuration definition table. 2 is a threshold control table. 3.3a and 3b are statistical information tables. 10-17.20.21 are processing steps. are shown respectively. The first circle (part 2) (b2) (b3) A diagram showing an embodiment of the present invention No. figure (Part 2) (bl) A diagram showing an embodiment of the present invention No. figure (Part 1) No. figure (Part 3) No. figure (Part 4) No. figure (Part 5)

Claims (1)

[Claims] A method for monitoring errors occurring in a computer system equipped with an error recovery procedure (ERP) mechanism that attempts to analyze and recover from errors in devices occurring during input/output operations, the system comprising: For each configuration definition table (1) and the error recovery procedure code (ERP code) created by the hardware of the computer system, weight the suspect locations, and consider the one with the higher weight as the upper dimension, and the one with the lower weight. In addition to managing it as a lower dimension,
A threshold control table (2) is created in which a plurality of error monitoring times are set corresponding to the weighted suspect locations, and the number of error occurrences for performing threshold control is set corresponding to the plurality of set error monitoring times. ), and a statistical information table (3, 3
a, 3b), and when an error occurs during input/output operation, the error recovery procedure (ERP) generated and output by the above error recovery procedure (ERP) mechanism is
The error recovery procedure (ERP) code generated and output by the error recovery procedure (ERP) mechanism identifies the error generation device based on the error generation device address, the error generation device address, and the system configuration definition table (1) above. and the error generating device address as keys, the threshold control table (2),
Update the number of errors in the corresponding part of the statistical information table (3, 3a, 3b) by referring to the statistical information table (3, 3a, 3b), and the updated error number is applied to the threshold control table. (2
), an error monitoring method is characterized in that an error notification is issued if the threshold value set in the error monitoring method is exceeded.