JP3015631B2 - Disk failure automatic recovery processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk failure automatic recovery processing method for recovering a failure of a multiplexed disk device.

[0002]

2. Description of the Related Art As a countermeasure against a failure of a disk device, for example, a magnetic disk device, a method of improving reliability by making a magnetic disk device redundant and writing and multiplexing the same data to a plurality of magnetic disk devices. Is generally adopted. Such a multiplexed magnetic disk device is generally called a mirrored disk device, or a duplicated disk device (in the case where it is composed of two disk devices).

Conventionally, when a failure (failure) occurs in one magnetic disk device in a system having such a multiplexed disk device, failure recovery processing has been performed in the following procedure.

First, a failed magnetic disk device is logically disconnected from the system (on-line state). This disconnection is performed automatically. Next, after the hardware maintenance personnel reformats the entire area of the disk medium of the failed magnetic disk apparatus or replaces the apparatus itself, the entire area from the normal disk (master disk) in the multiplexed disk apparatus is changed. Data (master disk image). Then, when the copying of the entire area is completed, the corresponding device is logically incorporated into the system to restore the online state (ONLINE state).

[0005]

As described above, in the conventional failure recovery processing method when a failure occurs in a multiplexed disk device, manual intervention is required until the online state is restored after the failure occurs. In addition, since it is necessary to copy the entire area of the large-capacity disk drive, it takes a very long time. During this time, the multiplexed disk drive is forced to perform a single operation, and the reliability is extremely reduced.

In addition, there is another problem that the cost is increased because manual work is required. SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and has as its object to provide a disk failure automatic recovery processing method that can automatically and quickly recover from a failure when a failure occurs in a multiplexed disk device. It is in.

[0007]

SUMMARY OF THE INVENTION According to the present invention, when a failure occurs in a multiplexed disk device, it is determined whether or not the failure is caused by a disk medium based on the failure information. Partial formatting is performed in a minimum unit including a failed portion of a disk medium, and after the normal completion of the partial formatting, only data of a corresponding portion is copied from another normal disk device and restored to an online state. .

Further, according to the present invention, in response to a normal disk write request generated during the above-mentioned partial formatting, multiplex writing to a multiplexed disk device including a failed disk device is performed, and Is characterized in that exclusive processing is performed with the normal disk writing processing that occurs during the processing.

[0009]

Generally, when a failure occurs in a multiplexed disk device, failure information including the details of the failure occurrence and the location of the failure is reported from the channel device to the multiplexed disk access mechanism.

Therefore, in the above configuration, it is determined based on the failure information notified from the channel device to the multiplexing disk access mechanism whether or not the failure of the generated disk device is caused by the disk medium. However, if the error is caused by the disk medium, it is assumed that the failure can be recovered by reformatting the disk medium.
Partial formatting is performed in a minimum unit including a failed portion of a disk medium of a failed disk device.

As a result of this partial formatting, if the content written in the relevant portion can be read without error,
Assuming that the partial formatting was completed successfully,
Data corresponding to the part destroyed by the partial formatting is copied from the master disk.

After the copy is completed, the disk drive is installed online assuming that the disk drive in which the failure occurred has been recovered from the failure. As described above, the present invention focuses on the fact that when a failure in a disk device is caused by a disk medium, the failure can be recovered by reformatting. Partial formatting is performed in the minimum unit, and when the formatting is completed normally, only the data of the corresponding part is copied from another normal disk device, so the failure of the disk device can be recovered without human intervention ,
It can be automatically restored to the online state, and since it is the minimum unit of failure recovery including the failed part by partial formatting, the amount of copying from the master disk is small and single operation with multiplexed disk unit Time can be minimized.

In addition, in response to a normal disk write request generated during the above-mentioned partial formatting, multiplex writing to a multiplexed disk device including a disk device in which a failure has occurred is performed. If the exclusion is made from the normal disk write processing, the disk write processing can be continued during a series of processing for disk failure recovery. For this reason, even during the disk failure recovery processing period, the operation is performed as a multiplexed disk device, so that the reliability is further improved particularly in the case of writing a disk to an area other than the part to be partially formatted. Also, even in the case of writing a disk to the part to be subjected to partial formatting, since the exclusion is performed with the copy processing, there is no possibility that data will not match between the disk devices.

[0014]

FIG. 1 is a block diagram showing an embodiment of an information processing system provided with a dual magnetic disk drive to which the present invention is applied. In FIG. 1, reference numeral 10 denotes a CPU (central processing unit) which forms the center of the system, reference numeral 20 denotes a duplicated magnetic disk drive configured using two magnetic disk drives 21 and 22, and reference numerals 31 and 32 denote magnetic disk drives 2
This is a channel device for connecting the CPUs 1 and 22 to the CPU 10. The channel devices 31 and 32 receive a request from the CPU 10 and control data input / output between the magnetic disk devices 21 and 22 and the CPU 10.

The CPU 10 includes a multiplexed disk access unit 11, a file system 12, a data processing unit 13,
It has a magnetic disk failure notification mechanism 14 and a magnetic disk automatic recovery processing mechanism 15.

The multiplexed disk access mechanism 11 is realized by a processing function of the CPU 10 according to multiplexed disk software, and in accordance with a disk write request given from the file system 12, transfers the requested data through the channel devices 31 and 32. Magnetic disk drive 2
It is responsible for the process of writing data twice to 1 and 22. When a magnetic disk failure occurs, the multiplexed disk access mechanism 11 determines whether or not the failure can be automatically recovered, and if the automatic recovery is possible, notifies the magnetic disk failure notification mechanism 14 of the fact.

The file system 12 provides the data processing mechanism 13 with a file access interface. The data processing mechanism 13 is realized by a processing function of the CPU 10 according to a general application program, and is a request source of a magnetic disk access.

The magnetic disk failure notification mechanism 14 multiplexes the failure information of the magnetic disk device with the multiplexed disk access mechanism 1.
1 to notify the magnetic disk automatic recovery processing mechanism 15.

The magnetic disk automatic recovery processing mechanism 15 is realized by a processing function of the CPU 10 according to a magnetic disk automatic recovery program, and receives a failure occurrence notification from the magnetic disk failure notification mechanism 14 to automatically recover a magnetic disk failure. Perform processing.

Next, the operation of the system shown in FIG. 1 will be described with reference to the flow charts shown in FIGS. 2 to 4, taking a case where a failure occurs in the magnetic disk device 21 in the duplicated magnetic disk device 20 as an example. 2 is a flowchart showing a processing procedure when a magnetic disk failure occurs in the multiplex disk access mechanism 11, FIG. 3 is a flowchart showing an automatic recovery processing procedure in the magnetic disk failure notification mechanism 14, and FIG. 6 is a flowchart illustrating a procedure of a disk access process in the disk access mechanism 11.

First, the occurrence of a failure in the magnetic disk device 21 is detected by the channel device 31. The channel device 31 notifies the multiplexed disk access mechanism 11 of the failure of the magnetic disk device 21 using the failure information.

The multiplexed disk access mechanism 11 recognizes the occurrence of a failure in the magnetic disk device 21 based on the failure information from the channel device 31. This fault information includes
Information such as the content of the fault (of the magnetic disk device 21) and the location of the fault is included.

The multiplexed disk access mechanism 11 uses the failure information from the channel device 31 to
When the occurrence of the failure 1 is recognized, first, based on the information, a failure caused by the disk medium (a defective error which is an uncorrectable error with respect to the disk read request).
Is checked to determine whether or not the failure can be automatically recovered by reformatting the disk medium (step S1 in FIG. 2).

If the failure of the magnetic disk device 21 is caused by the disk medium and it is determined that the automatic recovery is possible, the multiplex disk access mechanism 11
Is a state in which only a disk read request from the data processing mechanism 13 is prohibited (this state is called a COPY state, which is an intermediate state for restoring to the ONLINE state, and a write request is not prohibited) (step S2 in FIG. 2).

Next, the multiplexed disk access mechanism 11
The occurrence of a failure is notified to the magnetic disk failure notification mechanism 14 based on the failure information received from the channel device 31 (step S3 in FIG. 2). Then, the magnetic disk failure notification mechanism 14 notifies the magnetic disk automatic recovery processing mechanism 15 of the occurrence of the failure based on the failure information, and requests the magnetic disk device 21 in which the failure has occurred to perform automatic recovery processing.

If it is determined in the step S1 that automatic recovery is impossible, the multiplexed disk access mechanism 11 disconnects the failed magnetic disk device 21 from the system and sets it in the OFFLINE state (FIG. 4). 2 steps 4). In this case, the subsequent processing is the same as the conventional processing.

Now, the magnetic disk automatic recovery processing mechanism 15
Receives the failure occurrence notification from the magnetic disk failure notification mechanism 14, determines the failure location (error occurrence location) based on the failure information, and
, An attempt is made to perform partial formatting on the corresponding disk medium of the failed magnetic disk device 21 in the minimum unit including the failed part (step S11 in FIG. 3). In this partial formatting, a sector replacement that replaces an error sector with another normal sector, a track replacement that replaces a track containing a defective sector with another normal track, or a cylinder that contains a defective track with another normal cylinder Cylinder replacement is performed. Here, if the number of error sectors in the same track is smaller than a predetermined number, sector replacement is performed, and if the number exceeds the predetermined number, track replacement is performed. If the number of tracks requiring track replacement exceeds a predetermined number, cylinder replacement is performed.

In step S11, the magnetic disk automatic recovery processing mechanism 15 performs partial formatting on the relevant disk medium of the failed magnetic disk device 21 in the minimum unit including the failed part, and then the formatted data is normal. Whether it can be read (without error) in the data for that range (test data)
This is checked by writing and reading (step S12 in FIG. 2). In the test of this step S12, in response to a disk write request from the data processing unit 13,
Since the data (test data) may be overwritten after the partial formatting, the contents of the data are not checked. That is, in step S12, it is checked whether the data can be simply read without error.

Now, the magnetic disk automatic recovery processing mechanism 15
If the result of the check in step S12 indicates that the data can be read normally, it is determined that the partial formatting has been successful (normally ended).

In this case, the magnetic disk automatic recovery processing mechanism 15 excludes the disk write processing for the disk write request from the data processing mechanism 13 through the multiplexed disk access mechanism 11 (step S in FIG. 3).
13) The data corresponding to the portion on the disk medium of the magnetic disk device 21 destroyed by the previous partial formatting is copied from the magnetic disk device 22 as the master disk at that time, and the validity of the data after the copy is copied. Is confirmed (step S14 in FIG. 3). The unit of the exclusive occupation process here is fixed to, for example, a sector, but can be set arbitrarily by a system parameter.

When the validity of the copied data is confirmed in step S14, the automatic recovery processing mechanism 15 determines that the failure of the magnetic disk device 21 has been recovered and logically transfers the disk device 21 to the system. By incorporating, it is restored to the ONLINE state (from the COPY state).

Next, the magnetic disk automatic recovery processing mechanism 15
Releases the exclusion of the disk write processing for the disk write request from the data processing mechanism 13 (step S15 in FIG. 3), and ends the processing.

As a result of the check in step S12,
If the data cannot be read normally, the magnetic disk automatic recovery processing mechanism 15 determines that the partial formatting has failed (abnormal termination), and disconnects the failed magnetic disk device 21 from the system to set it in the OFFLINE state. (Step 16 in FIG. 3). In this case, the subsequent processing is the same as the conventional processing.

Now, the multiplexed disk access mechanism 11
Is a data processing mechanism 13 through the file system 12.
When a disk access request is given from the server, the following disk access processing is performed according to the flowchart of FIG.

First, the multiplexed disk access mechanism 11
Determines whether the disk access request given from the data processing mechanism 13 is a disk write request or a disk read request (step S2 in FIG. 4).
1).

In the case of a disk write request,
The multiplexed disk access mechanism 11 takes exclusion from the copy processing by the magnetic disk automatic recovery processing mechanism 15 (step S22 in FIG. 4), and writes the requested data to both the magnetic disk devices 21 and 22 (step S2 in FIG. 4).
3). Next, the multiplexed disk access module 11 releases the exclusion from the copy processing (step S24 in FIG. 4),
The requested disk writing process ends. When the magnetic disk automatic recovery processing mechanism 15 is not in the operating state, the above steps S22 and S24 become unnecessary.

On the other hand, if it is determined in step S21 that the request is a disk read request, the multiplexed disk access mechanism 11 returns to the normal disk device among the magnetic disk devices 21 and 22 (if both are normal, Then, a read process is performed from a disk device that can be read first (step S25 in FIG. 4).

The case where the present invention is applied to an information processing system having a dual magnetic disk device has been described above.
The present invention is also applicable to a multiplexed disk device including a plurality of magneto-optical disk devices or an information processing system including a multiplexed disk device including a plurality of optical disk devices.

[0039]

As described in detail above, according to the present invention,
When the failure of the disk device is caused by the disk medium, paying attention to the fact that the failure can be recovered by reformatting, the failure information indicating the occurrence of the failure of the disk device indicates that the failure is caused by the disk medium. If the determination is successful, the partial formatting is performed on the disk medium in the minimum unit including the failed part, and when the formatting is completed normally, only the data of the corresponding part is copied from another normal disk device. As a result, the failure of the disk unit was recovered unattended,
The online state can be automatically restored, and costs including maintenance costs can be reduced.

Further, according to the present invention, since the failure recovery by formatting is performed in the minimum unit including the failed part, the time required for formatting and the amount of copying from the master disk are reduced as much as possible, and , The single operation time can be minimized, and the reliability as a multiplexed disk device can be improved.

According to the present invention, in response to a normal disk write request generated during the above-mentioned partial formatting, multiplex writing to a multiplexing disk device including a failed disk device is performed. By applying a configuration that takes exclusion from the normal disk write processing that occurs during this period, the disk write processing can be continued and the operation as a multiplexed disk device can be continued during a series of processing for disk failure recovery. In particular, in the case of writing a disc to an area other than the part to be partially formatted, the reliability is further improved. In addition, even in the case of writing a disc to the target portion for partial formatting, since the exclusion is performed with the copy process, there is no possibility that data will not match between the respective disk devices.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of an information processing system including a dual magnetic disk device to which the present invention is applied.

FIG. 2 is an exemplary flowchart showing a processing procedure when a magnetic disk failure occurs in the multiplexed disk access mechanism 11 in the embodiment.

FIG. 3 is a magnetic disk failure notification mechanism 1 according to the embodiment.
9 is a flowchart showing the procedure of the automatic recovery process in FIG.

FIG. 4 is an exemplary flowchart showing the procedure of a disk access process in the multiplexed disk access mechanism 11 according to the embodiment.

[Explanation of symbols]

10 CPU, 11 multiplexed disk access mechanism, 1
2 ... file system, 13 ... data processing mechanism, 14 ...
Magnetic disk failure notification mechanism, 15: automatic magnetic disk recovery processing mechanism, 20: duplicated magnetic disk drive, 21, 22
2 ... magnetic disk device, 31, 32 ... channel device.

Claims (2)

(57) [Claims] 1. A disk failure automatic recovery processing method for recovering a failure of a multiplexed disk device, comprising determining whether or not a failure is caused by a disk medium from failure information indicating the occurrence of a failure in the disk device. In accordance with the determination result, partial formatting is performed in a minimum unit including a failed portion of the disk medium of the failed disk device, and after the partial formatting is normally completed, only the data of the corresponding portion is copied from another normal disk device, and A disk failure automatic recovery processing method characterized by restoring to a state. 2. A disk failure automatic recovery processing method for recovering a failure of a multiplexed disk device, comprising determining whether a failure is caused by a disk medium from failure information indicating the occurrence of a failure in the disk device. In accordance with the result of the determination, the partial formatting is performed in the minimum unit including the failed portion of the disk medium of the failed disk device, and the multiplexed disk including the failed disk device is responded to a normal disk write request generated during this time. Multiple writing to the device is performed, and after the normal completion of the partial formatting, only the data of the corresponding part is copied from the other normal disk device and restored to the online state while taking exclusion from the normal disk writing process. A disk failure automatic recovery processing method characterized by the above-mentioned.