JP2002215336A - Memory control method, and memory sub-system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent generation of uncorrectable errors which cannot be corrected in medium surface inspection of a disc unit. SOLUTION: In the disc unit 200 for composing a disc array under the control of a disc control device 110, an error-correcting part 202 of which error correcting ability is changeable is provided. Error correction ability is changed between reading for normal operation and reading for medium surface inspection to a disc 203; a correctable error of a specified error correcting length or more is detected as subject to recover of rewriting. Result of medium surface inspection executed, when each disc device 200 has no input or output is periodically taken by the disc control device 110; and a recovery process is conducted to a position, where the correctable error is generated. Uncorrectable errors are thus prevented, and medium surface inspection can be conducted efficiently; without having affecting the input/output process of a central processing unit 100.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a storage device control technique and a storage subsystem, and more particularly to data assurance by detecting a medium surface failure in a magnetic disk device or the like constituting a redundant storage device such as a disk array device. And effective technology.

[0002]

2. Description of the Related Art In a disk controller interposed between a host computer and a group of large-capacity disk units and controlling data transfer between the two, data recorded by the disk unit becomes unreadable due to a medium failure in the disk unit. There's a problem.

[0003] A medium failure may occur when reading is temporarily impossible due to the frictional heat generated by the contact between the medium surface and the head, or when the medium surface is permanently damaged due to dust adhering or scratching. Reading may not be possible in some cases. In the former case, it becomes possible to read again after a lapse of time,
If another disk device constituting the same data group is degraded while reading is disabled or the latter permanent failure occurs, data loss due to exceeding the redundancy occurs.

[0004] A medium surface failure is detected when there is access from a higher-level device. However, if a medium failure occurs in a sector in which data of infrequently accessed data is recorded, the above phenomenon may occur. Therefore, means for detecting and recovering from a medium surface failure at an early stage becomes effective.

As a technique for solving this technical problem, for example, there is a technique described in Japanese Patent Application Laid-Open No. 2000-10736. That is, when the data test of each magnetic disk device is executed and the test is interrupted by the I / O (input / output) request, the address of the next data test is recorded, so that the I / O request from the host computer is recorded. A data test is performed by sewing a gap, and if data that cannot be read normally is detected, the data is restored from the data that could not be read normally using the redundant data of another magnetic disk device. is there.

[0006]

In the above prior art, the disk controller determines whether or not there is I / O from the host computer, and when there is no I / O, performs a medium surface inspection of the magnetic disk device. Although it does not affect the host I / O, in the disk control device having the cache memory, writing to the magnetic disk device is performed asynchronously with the host I / O, and a RAID that configures a data group with a plurality of units is used. In some cases, even if there is a host I / O, there is a disk device that does not perform processing for the I / O. Therefore, it is efficient for the disk controller to determine whether or not the host I / O exists and to perform the medium surface inspection. Not a target.

A magnetic disk device usually has an error correction function, and even if an unreadable bit occurs when data is read, the result of the error correction is reported to the higher rank if it is within the error correctable range. In the above-described conventional technique, the opportunity to execute the recovery process on the disk controller side is when an uncorrectable error occurs. For this reason, it is considered effective to attempt recovery by rewriting even in the case of a collectable error, and to execute the write position change process early when rewriting is impossible.

An object of the present invention is to perform a recovery process on a storage medium at an early stage of an error-correctable correctable error, thereby preventing occurrence of an uncorrectable error that cannot be corrected. Is to provide.

Another object of the present invention is to provide a technique capable of preventing data loss due to occurrence of multiple errors exceeding allowable redundancy in a redundant storage device having a redundant storage device. It is in.

Another object of the present invention is to provide a technique capable of performing a medium surface inspection of a storage medium in a storage device without affecting input / output performance with respect to a host device.

[0011]

According to the present invention, there is provided a first step of executing a process of reading data from a storage medium and detecting a recoverable error that can be recovered by an error correction process; A second step of attempting to overwrite the original storage location of the storage medium.

According to the present invention, in a storage subsystem including a storage device and a storage control device interposed between the storage device and a higher-level device, the storage device stores data from a storage medium in the storage device. The storage control device performs a read process, detects a recoverable error that can be recovered by an error correction process, and records the result as a test result. It has a function of reading the detected data from the storage device and attempting to overwrite the original storage position of the storage device.

More specifically, as an example, in a magnetic disk device that writes and reads data requested by a host computer under the control of a disk controller, a periodic operation is performed on a magnetic disk device without I / O. Means for performing a medium surface inspection, and means for detecting, as a failure, a correctable error having an error correction length equal to or more than a predetermined value in addition to an uncorrectable error, A means for periodically recovering the data and performing recovery processing.

[0014]

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

(Embodiment 1) FIG. 1 is a block diagram showing a system configuration of a disk subsystem which is an example of a storage subsystem for executing a storage device control method according to an embodiment of the present invention.

The disk subsystem according to the present embodiment comprises:
It comprises a disk controller 110 and a plurality of disk devices 200 under its control. A disk device 200 such as a magnetic disk device is connected to a central processing unit 100 as a higher-level device via a disk control device 110.

In the case of the present embodiment, as an example, the plurality of disk devices 200 constitute a redundantly configured disk array such as a RAID, and store and receive data exchanged with a higher-level device such as the central processing unit 100. The generated redundant data is distributed and stored in the plurality of disk devices 200, and the data that has become unreadable when a failure occurs is stored in another disk device 200 according to the redundancy. And so on.

The disk controller 110 has a channel IF
It has a unit 111, a cache memory 112, a disk device control unit 113, and a diagnosis result registration unit 114, and all data transfer between the central processing unit 100 and the disk device 200 is performed via the cache memory 112. Central processing unit 1
00 and the cache memory 112 between the channel IF unit 11
1 is controlled by the disk device control unit 113 between the cache memory 112 and the disk device 200. In addition, a function of reading out a diagnosis result of each disk device 200 and storing it in the diagnosis result registration unit 114 is also provided.

The disk device 200 has a buffer 20
1, an error correction unit 202, a disk 203, an I / O detection unit 204, a current address registration unit 205, and a diagnosis result registration unit 206. The cache memory 112 of the disk control device 110 and the disk 203 of the disk device 200
Data transfer between them is performed via the buffer 201. When reading out data from the disk 203, the error correction unit 202 corrects a missing part in the data within a correctable range and reads out the data to the buffer 201. Although there is an unreadable portion in the data being read, if the error can be corrected by the error correction unit 202, a correctable error occurs and the data on the buffer 201 becomes complete. But,
If the unreadable portion exceeds the correctable range and cannot be corrected, an uncorrectable error occurs and data cannot be read.

In the disk device 200 of the present embodiment,
In the case of normal reading, error correction is performed at the maximum,
By lowering the correction capability at the time of reading the medium surface inspection, a normal correctable error is detected as an error, as shown in FIG. 6, and recorded in the diagnosis result registration unit 206.

The error correction capability can be set arbitrarily. The I / O detection unit 204 can detect the presence or absence of an I / O operation on the disk device 200. In the current address registration unit 205, an address at which the medium surface inspection is performed is registered. The diagnosis result registration unit 206 can register up to n addresses (data storage positions on the disk 203) at which an error has occurred due to the diagnosis and detailed information on the error. Also, the current address registration unit 20
5 and the diagnosis result registration unit 206 are non-volatile, so that the diagnosis results remain even if the power of the disk device 200 is cut off, and can be resumed after recovery.

FIG. 2 is a flowchart showing an example of a flow of processing for detecting a medium failure in the disk device 200 according to the present embodiment. If a certain period of time has elapsed since the disk device 200 last performed the medium surface inspection (Yes in step 301), the I / O to
It is checked whether there is (Step 302). If there is I / O, the medium surface inspection is not performed (N in step 302).
o). If there is no I / O (Yes in step 302),
The data of the address registered in the current address registration unit 205 is read out to the buffer 201 in the disk device 200 (step 303). In the case of this embodiment,
At this time, the error correction processing capability of the error correction unit 202 is set lower than in the normal operation.

If the reading is normally performed (step 304)
No), the address next to the address implemented this time is registered in the current address registration unit 205 (step 30).
6) End. If an error occurs during reading (Yes in step 304), the address executed this time and the error detailed information thereof are registered in the diagnosis result registration unit 206 (step 305). The address next to the address implemented this time is registered in the current address registration unit 205, and the process ends.

In the case of the present embodiment, the error at step 304 is recovered by an error correction process in which the correction capability is intentionally set lower than the error correction process during normal operation, as illustrated in FIG. Error range that is not possible,
A part of the correctable error and the uncorrectable error during the normal operation are included, and are subjected to the recovery processing in the flowchart of FIG. 3 described later.

FIG. 3 is a flowchart showing the flow of processing when each disk device 200 performs a medium surface inspection and the disk control device 110 picks up the result and performs recovery processing on the data at the error occurrence position. . The disk control device 110 periodically picks up the result of the medium surface inspection performed by each disk device 200. After a certain period of time has passed since the last diagnostic result was collected (step 4
00), the diagnosis result is collected from the diagnosis result registration unit 206 of the disk device 200, and the disk control device 110
Is registered in the diagnosis result registration unit 114 (step 40).
1). The disk control device 110 includes a diagnosis result registration unit 11
4 is sequentially read out on the cache memory 112 (step 40).
2). If reading was successful (no error,
(In the case of a correctable error) (Yes at step 403)
s) The data in the cache memory 112 is rewritten to the disk device 200 again (step 405). If reading of the error occurrence address data fails (in the case of an uncorrectable error) (No in step 403),
The data that could not be read is recovered in the cache memory 112 using the data of the redundant disk device (step 404), and rewritten to the disk device 200 (step 405). If the writing has failed (step 406
No), the data write position is reassigned (step 407), and the data is written to the newly assigned write position (alternate area) (step 405).
When writing is completed normally (Ye in step 406)
s), end the recovery process.

The series of processing from step 403 to step 407 is repeated by the number of error addresses included in the diagnosis result obtained from the diagnosis result registration unit 206.

As described above, according to the first embodiment of the present invention, the disk device 200 to which no I / O request has been received from the central processing unit 100 is independently and periodically stored in the disk device 200. Carry out media inspection of
Since the disk control device 110 periodically collects the result and performs the recovery process, the disk 203 in each disk device 200 can be efficiently processed without affecting the I / O.
Media surface inspection can be performed.

Further, the disk device 200 is provided with an error correction section 202 which can arbitrarily set an error correction capability, and can change an error correction rate between normal data reading and reading during medium surface inspection (diagnosis). Therefore, in the case of a correctable error with normal error correction capability, it is possible to perform recovery by rewriting over a certain amount of collectable errors, so that preventive recovery processing at the stage of correctable errors can be performed. The implementation makes it possible to reduce the probability of occurrence of an uncorrectable error.

In the case of a collectable error, since the data itself can be finally read normally, the data recovery process using the redundant data is not required, and the overwriting process using the read data itself is performed. Recovery processing is possible, and even if a collectable error is included in the recovery processing, the efficiency of diagnosis and recovery processing will not decrease,
There is no concern that data loss occurs due to multiple errors in which redundant data and the like also become errors during the recovery processing.

In other words, it is possible to reliably prevent a failure such as data loss due to the occurrence of multiple uncorrectable errors in a plurality of redundantly configured disk devices 200 constituting a disk array or the like.

(Embodiment 2) In Embodiment 1 described above, the error correction capability at the time of diagnosis is made lower than that at the time of normal operation, and a part of the correctable errors is included in the errors to be subjected to the recovery processing. However, in the second embodiment, a case will be described where a correctable error to be subjected to recovery processing is selected based on the magnitude of the error correction length of the correctable error.

That is, in the second embodiment, the interface between the disk control device 110 and the disk device 200 has an error correction length resolution (a function of determining the length) at the time of a correctable error, and the disk control device 110 A recoverable error in which the error correction length generated in the medium surface inspection performed in the disk device 200 is equal to or greater than a certain value (threshold) is detected as a recovery target.

FIG. 5 shows an example of the contents of the diagnosis result registration unit 206 in this embodiment for this purpose. The diagnosis result registration unit 206 records, for each error, at least an address 206a indicating a storage position on the disk 203 of data in which the error is detected by the diagnosis, and an error correction length 206b in the case of a correctable error.

In step 402A of the recovery process by the disk control device 110 illustrated in FIG. 4, the error correction length corresponding to each address 206a of the diagnosis result read from the diagnosis result registration unit 206 of the disk device 200 206b, the error correction length 20
The recovery processing is performed by selecting a value whose value of 6b is larger than a predetermined threshold, that is, a more serious correctable error.

Alternatively, when error information is recorded in the disk device 200, a collectable error having a predetermined error correction length or more is selectively recorded, thereby reducing the amount of recording data and reducing the error correction length in the disk control device 110. Efficiency may be realized by omitting the determination processing.

Step 4 in the flowchart of FIG.
The processing other than 02A is the same as that of the first embodiment of FIG.

As described above, in the case of the second embodiment, the same effect as that of the above-described first embodiment can be obtained. The error correction length of the correctable error is recorded in association with the error correction length, and the correctable error whose error correction length is equal to or greater than a predetermined threshold is targeted for the recovery process, thereby performing the preventive recovery process at the stage of the correctable error, It is possible to reduce the probability of occurrence of an uncorrectable error.

The invention described in the claims of the present application is expressed in a different way as follows.

<1> In a disk controller connected to one or more magnetic disk devices, correctable errors are detected by medium surface inspection by changing the error correction capability between normal reading and medium surface inspection reading. A disk subsystem, comprising: a magnetic disk device capable of performing the above operation; a unit that writes a correctable error detected in the medium surface inspection to the magnetic disk device; and a unit that confirms that data has been correctly written.

<2> A medium surface inspection function in the magnetic disk device and a means for holding an address and detailed information when an error is detected, and a medium surface inspection periodically when there is no I / O to the magnetic disk device By doing
<1> The disk subsystem according to item <1>, wherein the magnetic disk drive having no host I / O has means for performing a medium surface inspection.

<3> The error occurrence address and the error detail information are recorded in the non-volatile memory or stored in the magnetic disk, so that even when the magnetic disk device is stopped due to power interruption, the inspection result is retained and restored from the inspection position after the interruption. The disk subsystem according to item <1>, further comprising a unit capable of performing a medium surface inspection again.

<4> Means for periodically reading the result of the medium surface inspection performed by the magnetic disk drive, reading out the data of the address registered as an error on the cache, writing the read out data to the magnetic disk drive, and correctly writing the data Means for confirming that the writing has failed, means for reassigning the writing position to the magnetic disk when writing has failed, and means for writing when the reading to the cache has failed, using data recorded in the redundant magnetic disk device. The disk subsystem according to item <1>, further comprising a unit for restoring data on the cache by using the disk subsystem.

<5> In a disk controller connecting one or more magnetic disk units, an interface having an error correction length resolution for a collectable error is provided between the magnetic disk unit and the disk controller. A disk subsystem comprising means for detecting a correctable error in a medium surface inspection and performing recovery processing by writing.

Although the invention made by the present inventor has been specifically described based on the embodiments, the present invention is not limited to the above-described embodiments and can be variously modified without departing from the gist thereof. Needless to say, there is.

[0045]

According to the present invention, it is possible to prevent occurrence of an uncorrectable error that cannot be corrected by performing recovery processing on the storage medium at an early stage at the stage of a correctable error that can be corrected. Is obtained.

According to the present invention, in a redundant storage device having a storage device having a redundant configuration, an effect is obtained that data loss due to occurrence of multiple errors exceeding the allowable redundancy can be prevented.

According to the present invention, it is possible to carry out an inspection of a medium surface of a storage medium in a storage device without affecting input / output performance with respect to a host device.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a system configuration of a disk subsystem that is an example of a storage subsystem that executes a storage device control method according to an embodiment of the present invention.

FIG. 2 is a flowchart illustrating an example of an operation of a disk device constituting a storage subsystem that implements a storage device control method according to an embodiment of the present invention;

FIG. 3 is a flowchart illustrating an example of an operation of a disk control device constituting a storage subsystem that implements a storage device control method according to an embodiment of the present invention;

FIG. 4 is a flowchart showing a modified example of the operation of the disk control device constituting the storage subsystem that implements the storage device control method according to one embodiment of the present invention;

FIG. 5 is an explanatory diagram illustrating an example of information used in a disk subsystem that is an example of a storage subsystem that implements a storage device control method according to an embodiment of the present invention;

FIG. 6 is a conceptual diagram showing an example of an operation of a disk subsystem which is an example of a storage subsystem that implements a storage device control method according to an embodiment of the present invention.

[Explanation of symbols]

100: central processing unit; 110: disk control unit (storage control unit); 111: channel IF unit; 112: cache memory; 113: disk unit control unit;
Diagnostic result registration unit, 200: disk device (storage device),
201: buffer, 202: error correction unit, 203: disk, 204: I / O detection unit, 205: current address registration unit, 206: diagnosis result registration unit, 206a: address, 206b: error correction length.

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Takao Sato 2880 Kozu, Odawara-shi, Kanagawa F-term in the Storage Systems Division, Hitachi, Ltd. (Reference) 5B018 HA14 KA21 MA12 5B065 BA01 CA11 EA03 EA15 EA24

Claims (5)

[Claims] A first step of executing a process of reading data from a storage medium and detecting a recoverable error that can be recovered by an error correction process; and a process of reading the data in which the correctable error is detected from the storage medium. A second step of attempting to overwrite the original storage location. 2. The method of controlling a storage device according to claim 1, wherein the storage device is each of a plurality of magnetic disk devices having a redundant configuration configuring a disk array, and in the first step, together with the correctable error. In the error correction process, an unrecoverable error that cannot be recovered is also detected, and when the uncorrectable error is detected, the error recovery process using redundant data stored in another storage device is performed. Obtaining the correct data, and in the second step, confirming whether or not the overwriting of the data has succeeded, and writing the data in an alternative area of the storage medium if the overwriting has failed. Storage device control method. 3. A storage subsystem including a storage device and a storage control device interposed between the storage device and a higher-level device, wherein the storage device transmits data from a storage medium in the storage device. Performing a read process, comprising an inspection function of detecting a recoverable error that can be recovered by an error correction process and recording the result as an inspection result, wherein the storage control device refers to the inspection result of each of the storage devices, A storage subsystem, comprising: a function of reading the data in which the correctable error is detected from the storage device and attempting to overwrite an original storage position of the storage device. 4. The storage subsystem according to claim 3, wherein the storage device is each of a plurality of storage devices in a redundant configuration, and wherein the storage control device is configured to read from a specific storage device based on the inspection result. If the reading of the data fails, a data recovery process is performed using the redundant data stored in the other storage device, the overwriting is attempted using the recovered data, and the overwriting fails. A storage subsystem having a function of storing the data in another alternative area of the storage medium of the storage device. 5. The storage subsystem according to claim 3, wherein the checking function of the storage device detects the correctable error by lowering the error correction processing capability than in a normal operation. The storage subsystem, wherein the controller selects the data having an error correction length equal to or greater than a predetermined value from the data having the correctable error and attempts the overwriting.