JPH09259047A - Recovery device for duplexed disk - Google Patents

Abstract

PROBLEM TO BE SOLVED: To speed up the restoring process of a duplexed disk device which is performed after the unexpected restart-up of the system. SOLUTION: A program executed by a CPU 100 can access a nonvolatile memory 300 and disk devices 401-40n through a system bus 200. On the nonvolatile memory 300, there are disk device write information storage areas 301-30n corresponding to the disk devices 401-40n. Those areas are updated when requests to write the corresponding disk devices are made. For the restoring process of duplexed disk devices, the CPU 100 refers to the areas and restores only matched blocks. Consequently, the actual I/O to and from the disk devices for the restoring process is limited to the matched blocks, so the process is speeded up.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dual disk in a computer system in which reliability is improved by processing two disk devices as one logical disk and giving redundancy to the disk device. Related to recovery equipment.

[0002]

2. Description of the Related Art In this type of conventional recovery device for a duplicated disk, there is a possibility that the data between two disks may be inconsistent when the system goes down and the system is restarted. , All data is copied from one disk to the other in order to make the data consistent.

Further, in Japanese Patent Laid-Open No. 04-369710, at the time of input / output processing of a duplicated disk, there is a mismatch during input / output to / from the duplicated disk, and it is judged at the time of completion whether or not data is consistent and managed by a flag. By doing so, it is determined whether or not it is necessary to execute the duplication restoration work (data consistency) after the system goes down, and the duplication restoration work is made efficient.

[0004]

In the above-described conventional duplex disk recovery device, when the system goes down unexpectedly due to a failure or the like, there is no information as to which block is being accessed, so the system is restarted. Later, there is the problem that a copy of all the data on the disks is required to make the data consistent between the two disks of the duplex disk.

An object of the present invention is to identify a block in which data inconsistency occurs between two disk devices, and perform a process for obtaining consistency only in that block, thereby achieving high-speed data consistency. It is to provide a recovery device for a redundant disk capable of taking advantage of the above.

[0006]

The apparatus of the present invention causes a higher-level program to recognize two disk devices in which the same data is stored in the same block as one logical disk, and By issuing a write request to the two disk devices, if a failure occurs in one of the disk devices that make up the logical disk, the recovery device for the redundant disk in the system that accesses the other disk device, By storing the information of the block position to be written in each disk device at the time of the write processing to the two disk devices configuring the logical disk, and storing the information in the memory that does not lose the memory due to the power shutdown, After unexpected system restart, 2 from the block position in the memory
It is characterized in that a block in which data inconsistency may occur between two disks is identified and data consistency between the two disks is taken.

According to the present invention, when a write request is issued to a disk device, for example, the write block position is stored in the disk device write block information storage area corresponding to the corresponding disk device existing in the non-volatile memory. Then, when the write request is completed, the write block information is initialized. Therefore, if the write request to the disk device is completed or is being executed, the block position can be known by referring to the write block information storage area. Since the disk device write block information storage area exists in the non-volatile memory, it retains the information even after the system is restarted. Therefore, even if the data inconsistency of the two disk devices forming the logical disk occurs due to an unexpected system down, it is possible to know the state of the write request to the two disk devices immediately before the system down, The data can be matched in a short time by recovering the data only for the block in which the write is being executed.

[0008]

Next, the present invention will be described with reference to the drawings.

FIG. 1 is a hardware configuration diagram of an embodiment of the present invention. The CPU 100 can access the nonvolatile memory 300 and the n disk devices 401, 402 ... 40n via the system bus 200.
Any two of the disk devices 401 to 40n can be targets for duplication. On the non-volatile memory 300, disk device write block information storage areas 301, 302, ... 30n are secured in a one-to-one correspondence with the mounted disks, and write data for the disk device 1, the disk device 2 ,. It is updated by the program executed by the CPU 100 during processing.

FIG. 2 is a block diagram of a program executed by the CPU 10. Logical disk control program 20
Provides a disk input / output interface for the upper software 10. When the logical disk control program 20 receives an input / output request from the higher-level software 10, the logical disk input / output control unit 21 sends the same input / output request to the disk device control drivers 31 and 32 that control the disk devices having the same data one by one. To issue.

Disk device control driver 31 (32)
Is a disk device control driver input / output control unit 311 (3
21) issues an input / output request to the disk device, and if the input / output request is a write request, the write block information update unit 312 (322) writes the corresponding write request to the disk device write block information storage area. The block position to be written is written, and as a result, the block is considered to be being processed. When the disk device control driver input / output control unit 311 (321) receives the completion interrupt from the disk device, it notifies the logical disk input / output control unit 21 of the completion and the write block information update unit 312 (32).
By 2), the contents of the disk device write block information storage area are cleared and the processing is completed. The logical disk input / output control unit waits for completion notifications from the two disk device control driver input / output control units 311 and 322, and then processes the next request. A flowchart of the above write processing is shown in FIG.

On the other hand, in the disk devices 401 to 40n, when a situation that causes data mismatch between the duplicated disk devices occurs, the information is inter-disk data matched via the disk device control driver input / output control unit 311. It is transmitted to the processing unit 22 and the recovery processing shown in FIG. 4 is performed. That is, the inter-disk data matching processing unit 22
Refers to the disk device write block information storage area corresponding to the two disk devices, identifies the block that is being processed as the block in which the mismatch occurs, and controls the two disk devices respectively. Data matching is achieved by issuing an input / output request for matching the data to the device control drivers 311 and 321.

Next, according to the flow charts shown in FIG. 3 to FIG.
The write processing and recovery processing from will be described in more detail. (1) Write processing from higher-level software (see FIGS. 2 and 3) The upper software 10 issues a write request to the logical disk input / output control unit 21 of the logical disk control program 20 (STEP1-1).

2. The logical disk input / output control unit 21 that has received the request issues a write request to the disk device control driver input / output control units 311 and 321 in the disk device control drivers 31 and 32 that control the respective subordinate disk devices ( STEP 1-2-1, 1-2-2).

3. The disk device control driver input / output control units 311 and 312 perform [write execution processing] (STE).
P1-3-1, 1-3-2).

4. The logical disk input / output control unit 21 waits for completion notification of the write processing from both the disk device control driver input / output control units 311 and 321 that control the respective subordinate disk devices (STEP1-4).

5. The logical disk input / output control 21 receives the completion notification from the disk device control driver input / output control units 311 and 312, and then notifies the upper software of the completion of the write processing (STEP 1-5).

The "write execution process" will be described later with reference to the flowchart shown in FIG. (2) Flow of recovery processing (see FIGS. 2 and 4) The flow of processing (recovery processing) for ensuring data consistency between two disks forming a logical disk is shown below.

1. The disk-to-disk data matching processing unit 22 in the logical disk control program 20 uses the system bus 200.
The information in the write information storage area in the non-volatile memory 300 corresponding to the subordinate disk device is referred to via the
EP2-1).

2. Processing is divided as follows depending on the result of reference.

When the information of the corresponding block exists in either of the write information storage areas, [recovery read / write processing] is performed under the following conditions (STEP 2-2).

[Recovery read / write processing] will be described later with reference to the flowchart shown in FIG.

When the information of the corresponding block exists only in the write information storage area on one side, [recovery read / write processing] is performed under the following conditions (STEP 2-3).

When the information of the corresponding block does not exist in either of the write information storage areas: [Recovery read / write processing] is not performed because the matching of the corresponding blocks has been achieved.

3. It is checked whether or not all blocks of the target disk have been processed (STEP2-4). If not, the processing of the next block is started (STEP2-).
5). The recovery process ends when all blocks are completed.

Next, the [write execution process], [read execution process] and [recovery read / write process] used as parts in the above-described software write process and recovery process will be described in detail.

Flow of write execution processing (see FIGS. 2 and 5) Disk device control driver 3 that received the write request
Disk device control driver input / output control unit 3 in 1, 32
Reference numerals 11, 321 denote write block information updating units 312, 3
An information update request is issued to 22 (STEP 3-1).

2. Upon receiving the update request, the write block information update units 312 and 322 add the information of the corresponding block to the corresponding write information storage area in the nonvolatile memory 300 via the system bus 200 (STEP2-
2).

3. The disk device control driver input / output control units 311 and 321 issue a write request to the disk device via the system bus 200 (STEP3-
3).

4. The disk device control driver input / output control units 311 and 321 wait for a write processing completion notification from the disk device (STEP 3-4).

5. After receiving the completion notification from the disk device, the disk device control driver input / output control units 311 and 3
The write block information updating units 312 and 322, which have received the notification of completion from 21, delete the information of the block from the corresponding write information storage area in the nonvolatile memory via the system bus 200 (STEP 3-5).

6. The disk device control driver input / output control units 311 and 321 issue a write processing completion notification to the logical disk input / output control unit 21 (STEP 3-6).

Flow of read execution processing (see FIGS. 2 and 6) Disk device control driver input / output control units 311 and 3
21 issues a read request to each disk device via the system bus 200 (STEP4-1).

2. The disk device control driver input / output control units 311 and 321 wait for a read processing completion notification from the disk device (STEP 4-2).

3. The disk device control driver input / output control unit issues a read processing completion notification to the logical disk control program (the logical disk input / output control unit 21 or the inter-disk data matching processing unit 22 that issued the read request) (STEP 4). -3).

Flow of recovery read / write processing (see FIGS. 2 and 7) The inter-disk data matching processing unit 22 in the logical disk control program 20 issues a read request to the disk device control driver input / output control unit in the recovery source disk device control driver (STEP 5-1).

2. The disk device control driver input / output control unit that receives the request performs [read execution processing] (STEP
5-2).

3. The inter-disk data matching processing section 22 waits for a read processing completion notification from the disk device control driver input / output control section on the side issuing the read request (STEP 5-3).

4. The inter-disk data matching processing unit 22 issues a write request to the recovery destination disk device control driver input / output control unit (STEP 5-4).

5. The disk device control driver input / output control unit that has received the request performs a [write execution process] (STEP
5-5).

6. The inter-disk data success processing unit 22 waits for a write processing completion notification from the disk device control driver input / output control unit that issued the write space (STEP 5-6).

Although the disk recovery is performed in block units in the above-described embodiments, it may be performed in track units, and the disk device write block information storage area is held in a non-volatile memory. However, it may be held in a RAM backed up by a battery, for example.

[0043]

According to the present invention, even when the system goes down, it is held in the memory at the block position of the write to the disk device at that time.
By referring to the information on the memory and checking if the data do not match, if there is a mismatch, I / O can be performed only for that block, so that it is possible to achieve consistency in a short time. become.

[Brief description of drawings]

FIG. 1 is a hardware configuration diagram of an embodiment of the present invention.

FIG. 2 is a software configuration diagram according to an embodiment of the present invention.

FIG. 3 is a flowchart of a write process from higher level software according to an embodiment of the present invention.

FIG. 4 is a flowchart of a recovery process according to an embodiment of the present invention.

5 is a flowchart of a write execution process used in the flowcharts shown in FIGS. 3 and 7. FIG.

6 is a flowchart of a read execution process used in the flowchart shown in FIG.

FIG. 7 is a read / write processing flowchart of recovery processing used in the flowchart shown in FIG. 4;

[Explanation of symbols]

100 CPU 200 System Bus 300 Nonvolatile Memory 301 to 30n Disk Device Write Block Information Storage Area 401 to 40n Disk Device 10 Upper Software 20 Logical Disk Control Program 21 Logical Disk Input / Output Control Unit 22 Inter-Disk Data Matching Processing Unit 31, 32 Disks Device control driver 311, 321 Disk device control driver input / output control unit 312, 322 Write block information updating unit.

Claims (3)

[Claims] 1. A two-disk device in which the same data is stored in the same block is recognized by a higher-level program as a logical one disk, and a write request from the higher-level program is made to the two-disk device. When a failure occurs in one of the disk devices that make up the logical disk by issuing it to the other disk device, in the redundant disk recovery device in the system that accesses the other disk device, the two disks that make up the logical disk At the time of write processing to the device, the information of the block position to be written is secured for each disk device and is stored in the memory that does not lose the memory due to the power cutoff, so that the above-mentioned Data inconsistency may occur between two disks from the block position in memory Identify is block, two redundant disk recovery device characterized by taking the consistency of data between disks. 2. A disk device write block for storing the information of the block position to be written, which is prepared by the number equal to or larger than the maximum mountable number of disks of the system on the memory accessible from the CPU via the system bus. An information storage area and a write information update unit (312, 32) for updating the disk device write block information storage area during write processing.
2) and, when the data consistency between the two disks forming the logical disk is to be obtained, the data mismatch block is identified by referring to the disk device write block information storage area corresponding to each disk, and the data mismatch occurs. The recovery apparatus for a duplicated disk according to claim 1, further comprising: an inter-disk data matching processing section that inputs and outputs only to blocks. 3. The dual disk recovery apparatus according to claim 1, wherein the memory is a non-volatile memory.