JP3060039B2 - Automatic Scheduling Method for Online Diagnosis of Disk Unit - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a magnetic disk device connected to a computer system, for example, for preventing a job abend (abnormal termination of a job) due to the use of a magnetic disk device and a system down. The present invention relates to an automatic scheduling method of online automatic diagnosis for a computer.

With the recent increase in the scale of computer systems, the number of peripheral devices (particularly, magnetic disk devices) has increased remarkably, and there has been a demand for prevention of failures and early measures when failures occur.

For example, in the field of maintaining peripheral devices of an electronic computer, the system is in operation to realize one of the problems on the maintenance side: “early confirmation / early action of an abnormal device, and prevention of a serious failure”. The maintenance (CE) cylinder, surface and
Analyze (SA) cylinder ｛Home address of all trucks
When (HA) is stored and a track cannot be read,
Reading the surface analyze (SA) cylinder, recognizing the home address (HA) of the track, cylinder 各 for reading and processing each record, all user data areas, for example, perform a full read test, It is necessary to perform a diagnosis (especially a medium surface inspection) so as not to affect the user job in the designated period.

[0004] The maintenance technician can narrow down the magnetic disk devices to be checked mainly based on the result. In addition, when a critical failure is detected during online diagnosis, it can be automatically reported to the customer engineer (CE) department.

[0005]

2. Description of the Related Art FIGS. 7 and 8 are diagrams for explaining a conventional online diagnosis method for a magnetic disk drive. FIG. 7 shows a configuration example of a computer system, and FIG. Shows a format example of a test definition table for use.

Conventionally, the computer system shown in FIG.
As a test definition table for performing a diagnostic test on an online file storage device (DASD), for example, a magnetic disk device (DASD01, 02) 4, a keyboard (KB)
From 30, the device number and test name in the format shown in Fig. 8
(Test module name) and I / O instruction issue interval (DL
Value) is manually set and expanded in a predetermined area on the main storage device (MS) 2.

A central processing unit (CPU) 1 is a main storage device (MS)
When a diagnostic program 20 that has been developed in advance on the main memory (MS) 2 is executed, the test definition table on the main storage device (MS) 2 is referred to, and the issue interval (DL value ), An input / output instruction is issued, and the entire area of the magnetic disk device (DASD01, 02) 4 is read, for example, to diagnose the disk drive.

[0008]

Normally, various types of magnetic disk devices connected to a computer system have different storage capacities and command systems corresponding to the types of the devices. However, it is necessary to specify a predetermined test module name in advance, and recently, when the system is scaled up, it is necessary to manually specify the diagnosis (test module name) for each device number. Has the problem that it takes a lot of man-hours.

Also, once the input / output instruction issuance interval is set, it cannot be flexibly changed. Therefore, the magnetic disk device (DASD01, 02) 4 in the busy state is also accessed at a fixed I / O instruction issuance interval (DL value).
It had an adverse effect on user jobs and user systems. Furthermore, a magnetic disk device (DASD01,
02) For 4 above, the above-mentioned fixed input / output instruction issue interval
(The issuing interval includes a waiting time), and it took a long time to make a diagnosis.

The present invention has been made in view of the above-mentioned conventional drawbacks, and for example,
An automatic scheduling method for online diagnosis of a magnetic disk drive is provided, which aims to provide a scheduling method capable of issuing an input / output instruction for performing a diagnosis without affecting a user job and a user system. is there.

[0011]

FIG. 1 to FIG. 3 are explanatory diagrams of the principle of the present invention. FIG. 1 shows the whole structure, and FIG.
FIGS. 2A, 2B, 2C, and 2D show a method of setting an L value (input / output issue interval) and a setting example.
3A and 3B show a case where a DL value (input / output issuance interval) is set based on accumulated information of (MB). FIGS. 3A and 3B show the case where the DL value (input / output issuance interval) is set based on a unit control block (UCB). Output issuance interval). The above problem can be solved by an automatic scheduling method for online diagnosis of a magnetic disk drive, for example, configured as follows.

(1) An automatic scheduling method for diagnosing all online devices connected to the computer system, for example, by performing full access (for example, full read) to the magnetic disk device 4, With reference to a control information block of each disk device 4 connected to the computer system, for example, a measurement block (MB) or a unit control block (UCB), the busy time of the disk device 4 for a specified period is determined. From the status, the issue interval of I / O instructions issued to each disk device (D
L value) is dynamically changed and set, the test module name corresponding to each disk unit, and the issue interval of the I / O instruction
A test schedule table in which (DL value) is set is generated, and the disk device is diagnosed based on the generated test schedule table.

(2) An automatic scheduling method for online diagnosis of a disk device according to item 1, wherein the determined issue interval (DL value) of the input / output instruction and each disk device connected to the computer system are provided. 4 different track numbers,
Further, based on the average track access (read) time for each disk device 4, the completion date and time of full access (read) to each disk device 4 connected online to the computer system is predicted, and for example, If the maintenance time overlaps, the I / O instruction issue interval (DL
Is corrected, for example, to correct the predicted full access completion date and time.

[0014]

According to the automatic scheduling method for online diagnosis of a disk device of the present invention, the configuration definition of the input / output device under the system, that is, the device identification information table is automatically read, and the online file storage device (DASD) is read. Using a unit control block (hereinafter, referred to as UCB) or a measurement block (hereinafter, referred to as MB), the busy state of the file storage device (DSAD) to be diagnosed is checked, and an input / output issue interval ( While dynamically changing the DL value, online diagnosis of the file storage device (DSAD) is automatically performed so as not to adversely affect the system.

(1) The device identification code (DID: CLASS / TY) is referred to by referring to the device ID table under the system.
PE), a file storage device (DASD) is picked up, its name and device number are taken out, and a test name to be executed for that device number is determined from a device name registered in advance and a test module for diagnosis. / Create a device correspondence table. The setting of the correspondence table between the device name and the test module for diagnosis is, as described above, because the storage capacity and the command system are different for each device.
This is because a different test method is required for each device.

Also, for example, UCB corresponding to the device number
And checks whether or not the UCB is online. The issuance interval (DL value) of the input / output command is obtained based on the busy state of the device. There are two ways to determine the busy state of the device. Basically, according to the method of 1), when the MB is not found, the issue interval (DL value) of the input / output instruction is found by the method of 2).

1) A busy time MB (α) at a certain time is obtained based on the address of the UCB. Specifically, the following information is stored in the MB. Cumulative value of device connection time [cumulative value of channel and device connection time] Cumulative value of device hold time [cumulative value of device end pending] Cumulative value of device disconnection time [cumulative value of time when channel and device are disconnected] Therefore, based on the above three accumulated values, the busy time MB
When (α) is determined, for example, a weight is added to each of the cumulative values, and a weighted average is obtained, whereby the cumulative busy time MB (α) at a certain time can be determined.

Next, M after a specified time interval (Δt)
B (β) is obtained, and β−α = γ. γ when γ ≧ 0
The DL parameter value is determined according to / Δt (gradient value).
If γ <0, it is considered that the counter area has made one round, so the MB between the two points is measured again. The relationship between MB (α) and MB (β) at this time is shown in FIG.
This is shown in (d). FIG. 2D shows an example in which γ <0 and the measurement is performed again.

For example, if γ <0, the measurement is repeated up to three times until γ ≧ 0, and if that is not possible, the value of Δt is shortened and repeated three times, and a message to that effect is output. The above is repeated until γ ≧ 0.

FIG. 2A shows the correspondence for deriving the DL value (input / output issue interval) from the γ / Δt (gradient value). That is, when the γ / Δt (gradient value) is smaller than “1”, it means that the idle time is long. Therefore, the DL value (input / output issuance interval) is also changed as shown in FIG. ,shorten.

Similarly, the γ / Δt (gradient value) is from “1” to
In the case of "2", in the case of "2" to "3", and in the case of "3" or more, the method of obtaining the DL value (input / output issue interval) is determined as shown in FIG.

Therefore, when the minimum value of the DL value (input / output issue interval) specified by the operator is 600 msec and the maximum value is 1000 msec, an example of setting the actual DL value (input / output issue interval) is shown in FIG. As shown in b).

2) If the MB is not obtained,
The following method is used. That is, the UCB address is obtained from the device number of the online file storage device (DASD), the specified time interval is divided into n, and the number of busy flags on the UCB that are "on" for n times at equal intervals is counted. (M), and using the following formula, the busy rate of the device (b)
Ask for.

B = m / n × 100 [%] FIG. 3 (a) shows a method of setting a DL value from the UCB busy rate, and FIG.
The minimum value of the L value (input / output issue interval) is set to 600 ms
c, a setting example of an actual DL value (input / output issue interval) when the maximum value is 1000 msec.

Device model number; uuuu test module name; testmod input / output issuance interval (DL value); test definition when time is assumed; ON = uuuu / test
mod / DL = time is obtained by the procedure shown in FIG.

That is, the identification code of the device is obtained by referring to the device ID table under the system, and the file storage device connected to the system is obtained from the device identification code.
(DASD), extract its name and machine number,
A test module name is obtained from a predetermined test module name / device correspondence table, a UCB address corresponding to the device is recognized, and, for example, an MB total value is obtained via the UCB extracted from the UCB address. A DL value (interval of issuing input / output instructions) for the device is calculated from the accumulated MB value, and the test schedule table of the present invention shown in FIG. 1 and conventionally created manually is automatically generated.

(2) Next, the input / output issuance interval (DL value)
Based on the above, a method for obtaining the predicted date and time of completion of the entire read from the file storage device (DASD) will be described below. When the specified date is input, the online file storage device (DASD) is read from the device ID table under the system.
Then, the type of the file storage device (DASD) is classified, the entire read time of all the file storage devices (DASD) is obtained by the following formula T, and the read completion prediction date and time are presented.

If the user is dissatisfied with the date and time, for example, there is a problem such as being unable to keep up with the next regular maintenance by the customer engineer (CE), if the minimum value of the input / output issue interval is shortened (slight burden on the user JOB), , Present what is possible.

Here, when it is understood that the minimum value of the input / output issuance interval may be shortened depending on the agreement with the operator, specifically, the current use status of the file storage device (DASD), the test schedule table Change the I / O issuance interval time (DL value).

In accordance with the test schedule table generated as described above, the entire read test is started. Calculation formula: total number of cylinders of DASD type (A) x number of tracks per cylinder of DASD type (A) x average track read time of DASD type (A) + total number of cylinders of DASD type (B) x DASD type (B) Number of tracks per cylinder x DASD type (B) average track read time + + Total number of cylinders of DASD type (n) x Number of tracks per cylinder of DASD type (n) × Average track read time of DASD type (n) ─────────
────T After that, perform online diagnosis (for example, read the entire file storage device (DASD) in the retry suppression mode) by the due date, output the diagnosis result, and send a customer engineer (C
E) Notify department automatically.

When the full read test is started, the input / output issue interval (DL value) is changed to minimize the influence on the user job. ) To determine the busy state of each device, and dynamically change the input / output issue interval (DL value).

Further, it displays and prints out how far the lead has been read each day, and automatically reports to the customer engineer (CE) section how much it remains and how long it will be finished.

If there is a correctable error at the time of reading, the sense information and its position (CCHHR: cylinder number,
Track number) in the memo data set. When the collectable error is concentrated on a specific cylinder head, a flag code is generated when the threshold value exceeds a preset threshold value, and the flag code is automatically reported to a customer engineer (CE) department, for example.

When a stop request is input to this subsystem, the read processing ends at a sharp end,
The extent to which the currently read device number has been read (CCHH) is recorded and held in the memo data set of the test schedule. When a major failure (such as detection of an unreadable track) is detected during reading, a flag code is immediately generated and automatically reported to the customer engineer (CE) section.

Thereafter, the file storage device (D
When the SAD is activated, the memo data set is referred to, the read channel command word (CCW) is scheduled again from the next CCHH, and the read processing is restarted.

A dual volume control mechanism (DVC
In the case of the magnetic disk device controlled by F), read processing is performed on two physical volumes that are physically equivalent to one logical volume.

As described above, all the tests in the test schedule table are performed, and when all the file storage devices (DASD) under the system are read, a notification is automatically sent to a customer engineer (CE). The customer engineer (CE) in charge prepares for periodic maintenance based on the results.

Therefore, it is possible to reduce the maintenance work of the maintenance engineer, and also to obtain an effect useful for early detection of a faulty device and prevention of a serious fault.

[0039]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. FIGS. 1 to 3 are explanatory diagrams of the principle of the present invention.
4 to 6 show one embodiment of the present invention.
Shows a configuration example of a computer system to which the present invention is applied, and FIGS. 5 and 6 show configuration examples of each table generated corresponding to the computer system shown in FIG.

In the present invention, each magnetic disk device (DASD01, 02) is referred to by referring to the measurement block (MB) or unit control block (UCB) of each magnetic disk device 4 connected to the user's computer system.
In response to the busy status of 4, each magnetic disk device (DASD0
(02) The means for dynamically changing the period (DL value) for issuing the I / O command for each 4, and the period (DL value) for issuing the I / O command Disk device (DASD01, 02) 4 Means for predicting the overall read completion date and time based on the number of different tracks and the average track read time for each device is a means necessary for implementing the present invention. Note that the same reference numerals indicate the same object throughout the drawings.

Hereinafter, referring to FIGS. 1 to 3, FIGS.
Referring to FIG. 6, an automatic scheduling method in online diagnosis of a disk device according to the present invention will be described. In the case of the device configuration of the computer system shown in FIG. 4, the device identification information table (device ID table) is as shown in FIG. In FIG. 4, all devices are not shown for convenience of explanation, but the total number of registered devices = 25 (decimal), and the device number “0002” includes a printer device (device name: PT0).
01) are connected, and the device numbers “0140” and “01”
41 "is a magnetic disk device (device name:
DASD 01,02) 4 is connected.

The unit control block (UC) of each magnetic disk device (DASD 01, 02) 4 used in this embodiment
As described above, the following information is stored in the MB (measurement block) retrieved from the address B). Cumulative value of device connection time [cumulative value of channel and device connection time] Cumulative value of device hold time [cumulative value of device end pending] Cumulative value of device disconnection time [cumulative value of time when channel and device are disconnected] Therefore, as described above, this is also defined as a busy time at a certain time of the magnetic disk device (DASD01, 02) 4 by weighting each accumulated value and taking, for example, a weighted average. .

As described above, each magnetic disk drive (DASD0
Since the storage capacity and the command system are different, the test module names registered in advance for each magnetic disk device (DASD01,02) 4 (for example, DASD01,02, Test module name corresponding to (testm
od 1, 2, 3,...), a test module / device correspondence table is generated as shown in FIG.

A unit control block (UCB) corresponding to the magnetic disk device (DAAD01, 02) 4 (FIG. 6)
(a) In reference (1), the device number and the device are in a usable state, "ON" when the device is incorporated in the computer system, and the device is in an unusable state. If the unit control block (UCB) is not incorporated in the unit control block (UCB)
By referring to, the magnetic disk device 4 connected to the computer system can be specified.

The magnetic disk device 4 actually connected to the computer system obtained as described above is extracted and retrieved from its unit control block (UCB). Recognizing the busy state of the magnetic disk drive 4, calculating the DL value (interval of input / output instructions) for the magnetic disk drive 4, generating the test schedule table shown in FIG. The program determines the DL value indicated by the test schedule table (issue interval of input / output instructions)
Based on the memo data set, an input / output instruction is issued based on the above, and to which cylinder head (CCHHR) has been read, or which magnetic disk device (DAAD01, 02) 4 has completed the entire read, Remember.

Specifically, 1) Referring to the unit control block (UCB),
Device number of online magnetic disk unit 4 is 14
Since it is recognized that the numbers are 0 and 141, "002" is set in the column of the number of machine numbers in the test schedule table as shown in the figure.

2) The test E module name to be scheduled for the magnetic disk device 4 of the device number “140” is testmod 1, and the issue interval (DL value) of the input / output instruction during the schedule is 800 msec (X “0320”). ), The minimum input / output issuance interval of this device number is 600 msec (X
“0258”), the maximum I / O issue interval is 1000 ms
c (X "03E8"), cylinder head read completed
(CCHHR) is 0101000A0C, and the maximum cylinder head (CCHHR) in the unit number is 0A680.
00F00, the flag under test {bit 15 of the control bit shown in FIG. 1} is turned on.

3) Similarly, for the device number "141", the test module (testmod 2) is set as shown, and the control field is set to "0000" indicating that the test has not been performed. Set.

4) When the issuance interval (DL value) of the input / output command for each magnetic disk device 4 is scheduled as described above, as described in detail in the operation section,
The total read time for each magnetic disk device 4 can be obtained according to the formula T, and the full read predicted completion date and time of the online diagnosis can be obtained. The calculated total read completion date and time and the customer engineer (CE) maintenance time can be calculated. From the relationship,
The issue interval (DL value) of the input / output instruction is corrected as needed.

In the above embodiment, the magnetic disk device has been described as an example. However, the present invention is not limited to the magnetic disk device, and other disk devices such as an optical disk device and a magneto-optical disk device may be used. The good thing is, of course.

[0051]

As described above in detail, according to the automatic scheduling method for on-line diagnosis of a disk device of the present invention, the diagnosis program is controlled by a unit control block (UCB) of a magnetic disk device under the computer system. ), For each magnetic disk device connected online, the issue interval (DL
Value) and the issue interval (DL
Value) is dynamically changed in accordance with the busy state of the magnetic disk drive, a test schedule table is generated, and an input / output instruction is issued based on the generated test schedule table. When all the tests on the table are performed and all the file storage devices (DASD) under the system are read, the customer engineer (CE) department is automatically notified, and the customer engineer (CE) in charge sees the result, This is to prepare for regular maintenance.

Therefore, it is possible to reduce the maintenance work of the maintenance engineer, and to obtain an effect useful for early detection of a faulty device and prevention of a serious fault.

[Brief description of the drawings]

FIG. 1 illustrates the principle of the present invention (part 1).

FIG. 2 is a diagram illustrating the principle of the present invention (part 2).

FIG. 3 is a diagram illustrating the principle of the present invention (part 3).

FIG. 4 shows an embodiment of the present invention (part 1).

FIG. 5 shows an embodiment of the present invention (part 2).

FIG. 6 shows an embodiment of the present invention (part 3).

FIG. 7 is a view for explaining a conventional magnetic disk drive online diagnosis method (part 1);

FIG. 8 is a diagram for explaining a conventional magnetic disk drive online diagnosis method (part 2);

[Explanation of symbols]

1 Central processing unit (CPU) 2 Main storage unit (MSU) 3 Operation console 30 Keyboard (KB) 4 File storage unit (DASD01,02) or magnetic disk unit (DASD01,02) Unit control block (UCB) Measurement block (MB) Equipment identification information table (Equipment ID table) Test module / Equipment correspondence table Test schedule table Test definition table Input / output instruction issue interval (DL value)

 ────────────────────────────────────────────────── ─── Continuation of the front page (72) Keiko Mizokawa, Inventor Fujitsu Limited, 1015 Kamikodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture (56) References JP-A-4-305864 (JP, A) JP-A-5-74059 (JP, A)

Claims (2)

(57) [Claims] An automatic scheduling method for diagnosing by accessing all disk units (4) in an online state connected to a computer system, wherein each disk device (4) is connected to the computer system. Disk unit (4)
With reference to the control information block (or) of the disk device (4), from the busy state of the specified period,
The issue interval () of the I / O instruction issued to each disk device is dynamically changed and set, and the test module name and the issue interval () of the I / O instruction corresponding to each disk device (4) are set. Test schedule table with ()
And the generated test schedule table
(D) diagnosing the disk device (4) based on (1). 2. The automatic scheduling method for online diagnosis of a disk device according to claim 1, wherein: the issuance interval of the obtained input / output command; and each disk device connected to the computer system. ) Based on a different number of tracks and a different average track access time for each disk device (4), it predicts the date and time of complete access to each disk device (4) connected online to the computer system. Automatic scheduling method in online diagnosis of a disk device.