JP2000029638A - Diagnostic system for mirror disk - Google Patents

Abstract

PROBLEM TO BE SOLVED: To diagnose the presence/absence of the read access fault on a hard disk device in a computer system to which a multiplexed hard disk device is mounted. SOLUTION: In the computer system to which a mirror disk device 111 is mounted to the hard disk device on a slave side for performing only write access at the time of starting the system a hard disk diagnostic program 106 for reading all areas including an unused area while being on-line is activated and the presence/absence of the read access fault is diagnosed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to high reliability of a multiplexed hard disk drive in a computer system.

[0002]

2. Description of the Related Art In a computer system, the failure rate of a hard disk drive is higher than that of other components. Therefore,
Japanese Patent Application Laid-Open
158628 publication. This is to perform reading diagnosis when the disk device is idle, but since it is realized by hardware, the disk device must be replaced in order to apply to an existing system. Further, there is no means for tuning the read size in one read operation according to the load on the system and the disk capacity.

On the other hand, the reliability of the entire computer system is improved by duplicating a hard disk device and continuing data input / output processing with another hard disk device even if one of the hard disk devices constituting the hard disk device fails. ,
A so-called mirror disk device is known.

In the mirror disk device, data is written to both of the constituent hard disk devices.
As for reading, there are a method of executing only from the master disk, and a method of reading from both hard disks from a hard disk device which can read quickly.

[0005] The former method has the advantage that the number of accesses to the disk is small and the failure rate can be reduced because the hard disk that does not read only performs writing. However, whether or not the written data can be read is determined only after a disk access failure has occurred on the master disk and the read destination disk has been changed. It will be down. There is a method of reading written data at the time of writing to a hard disk and verifying the writing. However, when this method is executed, there is a problem that an access time becomes longer. Further, in this method, even if reading at the time of writing can be confirmed, it cannot be guaranteed whether the reading is possible at the next reading.

In the latter method, writing and reading are performed on both disks, so that an unreadable failure of the hard disk can be detected earlier than in the former method. However, since it is determined that the data cannot be read at the time of the access request, a data read failure cannot be detected for data that is not accessed.

Further, even if a failure occurs in an unused area, these two methods cannot detect the failure before use.

[0008]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a computer system having a multiplexed hard disk drive which is capable of preventing an unreadable failure of some or all of the hard disk drives constituting the multiplexed hard disk drive. It is an object of the present invention to provide a method of diagnosing a mirror disk, which can be detected at an early stage before an access by a mirror disk.

[0009]

The present invention provides a means for accessing individual hard disk devices constituting a multiplexed hard disk device to read data in a designated area, and a parameter input means for adjusting a load of a hard disk by a diagnostic program. I do.

[0010]

FIG. 1 shows a configuration of an embodiment of the present invention.

The computer system comprises a memory 101, a processor 102, an I / O adapter 103, and a mirror disk device 111. Mirror disk unit 111
Is composed of a hard disk device A104 and a hard disk device B105. On the memory 101, a hard disk diagnostic program 106 for diagnosing the soundness of the hard disk device and a hard disk diagnostic program management table 107 for controlling the operation of the program are provided.

The operating system 108 incorporates a mirror disk driver 110 so that a user can access the mirror disk device 111 in the same operation as a normal single disk. The mirror disk driver 110 is provided with a mirror disk management table 112 for storing definition information indicating which combination of hard disks constitutes a mirror disk, and a mirror disk state. As the mirror disk status, the master /
There is a slave / block. Further, a single disk driver 109 is also incorporated so that the hard disk diagnostic program 106 individually accesses the hard disk devices constituting the mirror disk device 111.

FIG. 2 illustrates write access of a mirror disk device by a general process.

When the general process 201 writes data to the mirror disk device 111, the mirror disk management table 1 provided in the mirror disk driver 110 is used.
The hard disk drive A 104 and the hard disk drive B are identified by referring to the
The writing is performed to both of 105. Mirror disk device 1
The write access to 11 ends when the writing to both of the two hard disks constituting it has been completed.

FIG. 3 illustrates the read access of the mirror disk device by the general process.

The general process 201 uses the mirror disk driver 110 to read from the mirror disk device 111, but reads from a predetermined hard disk device and does not perform read access to the other hard disk device. This is because when reading from the mirror disk device 111, the mirror disk management table 1 provided in the mirror disk driver 110 is used.
The hard disk device A104 registered as a hard disk (master) that performs both writing and reading is read with reference to the mirror disk state of No. 12, and the other hard disk device B105 registered as a write-only hard disk slave is read. Implemented by not reading.

Hereinafter, a hard disk device that performs writing and reading is referred to as a master disk, and a hard disk device that performs writing but does not perform reading is referred to as a slave disk. The initial state of the master disk / slave disk is defined when the mirror disk is constructed, and is stored in the mirror disk state of the mirror disk management table 112.

FIG. 4 illustrates the operation of the mirror disk device when a failure occurs in the master disk.

The mirror disk device 111 comprises a master disk which performs both writing and reading, here a hard disk device A 104, and a slave disk which performs only writing, here a hard disk B 105 402. At this time, if a failure occurs in the hard disk device A 104 which is the master disk, the mirror disk is closed to make the hard disk device A inaccessible, and the hard disk device B is set as the master disk and the mirror disk status in the mirror disk management table is changed. 403 is stored. In this way, access to the hard disk device A104 is stopped, the hard disk device B105 of the slave disk is switched to the master disk, and thereafter, write and read access to the hard disk device B105 is performed.

The problem here is that there is no guarantee that read access is possible because the hard disk device switched as the master disk has only performed write access. This is because a write access succeeds but a read access becomes impossible due to wear of the hard disk drive (for example, wear of a bearing of a head drive unit or a spindle motor, scratches on a magnetic disk, etc.). is there.

That is, since only write access is performed to the slave disk, it is not possible to detect even if a failure that disables read access has occurred, and the failure is detected by read access after switching to the master disk. Become. In addition, since the failure is detected on the master disk at this time, it is conceivable that the hard disk device is not in a multiplexed configuration. Access to the disk device will not be possible.

FIG. 5 illustrates the read access to the slave disk by the hard disk diagnostic program.

The hard disk diagnostic program 106 uses the single disk driver 109 to individually access the hard disk devices constituting the mirror disk device 111 during online operation. This figure shows a read access to the hard disk device B105 as a slave disk.

When a write request 502 is issued to the mirror disk device 111 while accessing the hard disk device B 105 using the single driver 109, the hard disk device A10 constituting the mirror disk
4. Attempt to access the hard disk device B105.
The hard disk device A105 performs a write process because no other access request is processed.

However, since the hard disk drive B105 is in the process of being read by the single driver 109, it waits until the read request 501 of the slave disk is completed, and executes the writing after the completion. Conversely, when a read request for the slave disk is issued by the single disk driver 109 during the write process by the mirror disk driver 110, the slave disk read process waits until the mirror disk write is completed.

FIG. 6 illustrates a method of read access to the slave disk.

Using the hard disk diagnostic program, read access to the slave disk 105 is performed for all areas, and this is repeated thereafter. When the power of the computer system is turned on again, the diagnostic program is restarted from the address where the diagnosis has been completed up to the previous time. By this read access to the entire slave disk, it is possible to detect an unreadable failure of the slave disk before switching the master disk, and it is also possible to prevent both master / slave failure of the mirror disk device.

FIG. 7 shows a hard disk diagnostic program management table for controlling the hard disk diagnostic program.

This table is created by the system administrator 702 using the diagnostic program setting / reference interface 701, and is read by the hard disk diagnostic program when the system is started. The operating status of the diagnostic program is written in this table, and can be confirmed using the diagnostic program setting / reference interface 701.

This table is composed of common information 703 and diagnostic program individual information 704 set for each hard disk drive to be diagnosed. In the common information 703, an “interval time” which is an activation interval of the hard disk diagnostic program is set. The diagnostic program individual information 704 includes “information on the hard disk to be diagnosed”, “diagnosis status”,
“Read size per start”, “last read address”, and “last read time” are set.
A device ID or the like is set in the “information of a hard disk to be diagnosed”.

The “diagnosis status” includes “normal” when the diagnostic program is being executed, “abnormal” when a fault is detected and the diagnostic program is skipped,
If the diagnostic program is temporarily interrupted by 702, “interruption” is set.

In the "read size per start", the size at the time of read access is set in KB units. In the “last reading address” and “last reading time”, the reading end address of the previous execution and the time at that time are set by the hard disk diagnostic program.

This table is stored in a file when information is updated, and is read into the memory by the hard disk diagnostic program when the system is restarted.

FIG. 8 shows the flow of the hard disk diagnostic program.

The hard disk diagnostic program is started when the system is started.

First, an "interval time" set in the common information area of the hard disk diagnostic program management table is read (801). Next, the presence / absence of a diagnostic target hard disk is determined based on the presence / absence of diagnostic program individual information 802. If there is individual information, 803 reads “read size per start”, “last read address”, and “last read time”. Then, "diagnosis status" is read from the individual information, and it is determined whether or not this is "normal" 804. If "normal", data of a designated size is read from the disk to be diagnosed 805.

If the diagnosis status is other than "normal", the diagnosis of the hard disk device is skipped and the processing of the next hard disk to be diagnosed is performed. When the reading 805 of the diagnosis target hard disk is successful, the “last read address” and the “last read time” of the diagnostic program individual information are updated. Read the hard disk to be diagnosed 8
If the step 05 fails, notification of the failure and logging of the failure content are performed 806, and the "diagnosis status" of the individual information is set to "abnormal", so that the hard disk device is skipped in the subsequent diagnosis program. When the diagnosis of all the hard disks to be diagnosed is completed, the computer sleeps 808 for a time designated as the “interval time”, and repeats the diagnosis again after the elapse of the time.

[0038]

According to the present invention, in a computer system having a multiplexed hard disk drive, a program for diagnosing soundness is executed on some of the hard disk drives constituting the multiplexed hard disk drive. It is possible to detect an unreadable failure of the hard disk device at an early stage during online operation.

Further, according to the system load and the size of the hard disk to be diagnosed, the diagnostic program start cycle and the disk size to be diagnosed by one start can be adjusted. Further, since the diagnostic program is realized as a user-level task, it can be incorporated into an existing system without replacing hardware.

[Brief description of the drawings]

FIG. 1 is an overall configuration diagram of a multiplexed hard disk drive according to an embodiment of the present invention.

FIG. 2 is a configuration diagram showing normal write access in FIG. 1;

FIG. 3 is a configuration diagram showing normal read access in FIG. 1;

FIG. 4 is a view for explaining the operation of the mirror disk device of the present invention when a failure occurs.

FIG. 5 is a diagram showing read access to a slave disk by the diagnostic program of the present invention.

FIG. 6 is a view for explaining a method of reading from a slave disk according to the present invention.

FIG. 7 is a view for explaining a hard disk diagnostic program management table of the present invention.

FIG. 8 is a flowchart illustrating a hard disk diagnostic program according to the present invention.

[Explanation of symbols]

101: memory, 102: processor, 103: I / O
Adapter 104, hard disk device A, 105 hard disk device B, 106 hard disk diagnostic program, 107 hard disk diagnostic program management table, 108 operating system, 109
Single disk driver, 110: mirror disk driver, 111: mirror disk device, 112: mirror disk management table, 201: general process, 202
... flow of write processing, 301 ... flow of read processing,
Reference numeral 401 denotes a writable and unreadable data area; 402 a mirror disk management table before a failure occurs; 403 a mirror disk management table updated after a failure occurs; 501 a flow of reading processing from a slave disk; and 502 a mirror. 701: diagnostic program setting / reference interface, 702: system administrator, 703: common information,
704: diagnosis program individual information, 801 to 808: flowchart of a hard disk diagnosis program.

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Ryoyoshi Yoshizawa 5-2-1 Omikacho, Hitachi City, Ibaraki Prefecture Inside the Omika Plant of Hitachi, Ltd. (72) Inventor Saburo Iijima 5-chome Omikacho, Hitachi City, Ibaraki Prefecture No. 1 Hitachi Process Computer Engineering Co., Ltd. (72) Inventor Teruyuki Yasunaga 5-2-1 Omikacho, Hitachi City, Ibaraki Prefecture Hitachi Process Computer Engineering Co., Ltd. (72) Inventor Hiroto Tsukahara Omika, Hitachi City, Ibaraki Prefecture 5-2-1, Machi-machi Process Computer Engineering Co., Ltd. (72) Inventor: Shinichi Takeda 5-2-1, Omika-cho, Hitachi City, Ibaraki Prefecture F-term (reference) 5B065 BA01 CC08 EA02 EA31 EA40

Claims (2)

[Claims] A hard disk drive is duplicated, writing is performed on both hard disks, and reading is performed only from a predetermined hard disk device (hereinafter, referred to as a master disk), and a failure occurs in the master disk. The other hard disk drive (hereafter,
In a computer system that reads from a slave disk), a hard disk diagnostic program that periodically reads the entire area of the slave disk and a hard disk diagnostic program management table are provided.
Diagnose the health of the slave disk while online,
A mirror disk diagnostic method for improving the reliability of a mirror disk device. 2. The computer system according to claim 1, wherein a cycle time and a read capacity per one access are specified for each hard disk device in the hard disk diagnostic program for confirming the soundness of the slave disk, thereby executing said step. A diagnostic method for a mirror disk, wherein the access load of the slave disk can be adjusted.