JPH01244557A - File recovery system - Google Patents

Abstract

PURPOSE:To recover a file in a short time without stopping the job of a file access processing by identifying an extent where an error occurs with a file allotment unit, and securing the new replacing extent in arbitrary separate volume. CONSTITUTION:When an input/output error occurs during the file access processing executed by a file access processing part 21, an exception processing means 4 is called, an exception input/output processing part 42 retries an input/ output processing, and when the retrial does not succeed, an extent swapping means 5 is called. An extent identification processing part 51 identifies the file base allotment unit including an input/output error generating address as the omitted extent, and an extent swapping processing part 52 secures the space of the size portion of the omitted extent as a replacing extent 36 in the separate volume. Thus, the extent information of the replacing extent 36 is replaced with the extent information of the omitted extent of a file 33, and simultaneously an extent control table is altered, as well.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a file recovery method when a volume failure occurs while accessing a file on a file device. A method for automatically recovering file space extents in the event of a failure.

(Prior Art) Conventionally, the file recovery method in this building is a method in which when a fault occurs in a part of a volume, the track in the faulty part is recovered by a rack to the alternate 1. There is also a method of recovering files to a spare volume by using all saved images that are periodically saved and journal images that are collected when files are updated.

(Problem to be solved by the invention) In the above-mentioned conventional file recovery method using alternating l and racks, recovery is performed in units of 1 to 1 rack. If a failure is detected, there is a problem in that it becomes necessary to recover the entire volume as a failure.

In addition, in the recovery method to a spare volume that uses all backup images and journal images, it is necessary to temporarily stop all jobs that use the failed file, so if the system's M T BF < Mean
There are problems in that the mean time between failures (Till1eBe1ecnFailure) decreases and recovery takes a long time.

(Means for Solving the Problems) In order to solve the above-mentioned problems, the file recovery method provided by the present invention uses a file device that stores files and an extent management unit that manages extent information of a file space. , an update job group having an extent management table in which the extent information managed by the extent management unit is written in advance, and a buffer that temporarily holds data read from the file or data written to the file; , a file access control unit having a file access processing unit that performs file access by reading data from the file into the buffer and writing data held in the buffer to the file in response to a request from the update job group; A data processing system comprising: a method for recovering when a failure occurs in the file access processing unit or a part of the file accessed by the file access processing unit, the file access processing unit of the file access control means; An exception input/output processing module that retries the file access when an input/output error occurs during file access processing, and extent swabbing control that instructs file recovery if the exception input/output processing module retries or fails. and an exception handling means having an extent swap control section of the exception handling means, which identifies the extent in which the error has occurred, and manages the file management section corresponding to this extent 1. An identification processing unit that indicates this in the extent information, and a volume that is different from the volume that contains the extent, secures a replacement extent with the same extent size as the extent, and manages it using the extent management unit. an extent swap processing unit for exchanging the extent information of the relevant extent and the extent information written in the extent management table with the extent information of the extent information written in the extent management table; It is characterized by being able to respect.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a data processing system to which a file recovery method according to an embodiment of the present invention is applied.

This data processing system includes an update job group 1 having an extent management table 11 and a buffer 12 for each job, a file access control means 2 having a file access processing section 21, and extent information of a file 33 and a replacement extent 36. A file device W3 having volumes 31 and 34 including extent management units 32 and 35 to be managed, and an extent swap control unit 4
1 and an exception input/output processing section 42; and an extent swapping section 5 including an extent identification processing section 51 and an extent swapping processing section 52.

In FIG. 1, data flows and information references are represented by solid line arrows, and control flows are represented by broken line arrows. Furthermore, the extent management table 11 and buffer 12 included in the update job group 1 are shown connected by a double line.

When the update job group 1 accesses the file 33 on the volume 31, the extent information of the file 33 in the extent management table 11 of the update job group 1 is fetched from the extent management unit 32 on the volume 31 in advance. It is assumed that there is

Access to the file 33 on the file device 3 is requested from the update job group 1 to the file access control means 2. The file access processing unit 21 of the file access control means 2 accesses the file 33 on the file device 3 while referring to the extent information in the extent management table 11. That is, a process of reading from the file 33 to the buffer 12 and a process of writing a data image on the buffer 12 to the file 33 are performed.

When an input/output error occurs during file access processing performed by the file access processing section 21 of the file access control means 2, the exception handling means 4 is called due to the fault of the input/output error. The exception handling means 4 generally retries input/output processing.

The exception input/output processing unit 42 of the exception handling means 4 retries the input/output processing, and if this does not succeed, the extent targeted for the input/output is a missing extent,
In FIG. 1, the file 33 is indicated by diagonal lines. At this time, the extent swap control section 41 of the exception handling means 4 calls the extent swap means 5.

The extent identification processing unit 51 of the extent swapping means 5 identifies the file base allocation unit (for example, 1 rack unit or smaller unit) that includes the input/output error occurrence address as the missing extent 1, and replaces the volume of the file device 3. The extent management unit 32 in 31 displays information indicating that the missing extent is missing in extent 1 of this missing extent.

The file 33 displayed as missing by the extent identification processing unit 51 is prohibited from use until file recovery is completed. That is, file assignment to the file 33 from subsequent jobs is prohibited.

The extent swap processing unit 52 of the extent swap unit 5 secures a space corresponding to the size of the missing extent in another volume (for example, the volume 34 on the file device 3) as a replacement extent 36.

Next, the extent swap processing unit 52 swaps (swaps) the extent information of the replacement extent 36 and the extent information of the missing extent of the file 33.
, extent management unit 32 in volumes 31 and 34
At the same time, the extent management table 11 is also changed.

If the file access request from the update job group 1 is for reading, the extent information for the replacement extent 36 whose extent has been recovered in this manner is still displayed as missing. This is because the data before the loss has not yet been recovered. Therefore, the file access processing of update job group 1 ends abnormally.

Data recovery of a missing extent can be performed in 1J by, for example, partial file recovery means (not shown) using a full saved image and a journal image.

Furthermore, in the case of a file access request or write from the update job group 1, since the data image before the loss exists on the buffer 12, there is no need to leave a loss indication for the replacement extent. In this case, when the extent swap processing by the extent swapping means 5 is completed, the exception input/output processing unit 42 of the exception handling means 4 refers to the updated extent management table 11 and transfers the data image held in the buffer 12 to the volume 34. write to the replacement extent 36 of. The second process in the case of a write request is performed without stopping the update job group 1, and the file access process from the update job group 1 ends normally.

FIG. 2 shows the extent l- of the extent swapping means 5.
5 is a diagram illustrating an example of a processing overview of an identification processing section 51 and an extent swap processing section 52. FIG. In FIG. 2, parts equivalent to those in FIG. 1 are indicated by the same numbers.

FIG. 2(a) shows an overview of the processing of the extent 1 identification processing unit 51. In the figure, 17 is the identification process of the extent in which an input/output error has occurred for the file 33, 17 is the identification process of the missing extent, and is the defective upper custen 1
It shows the loss display processing to the extent management unit 32 for ~.

In addition, the second [E71 (+1) shows the processing outline of the extent swap processing unit 52, and in the figure, the second [E71 (+1)] shows the process of securing space for the replacement extent 36 in another volume 34, and the point goes to the extent management units 32 and 35. The exten l-replacement process is shown in FIG. The extent l-replacement process is also performed on the extent management table 11 in FIG.

(Effects of the Invention) As explained above, the file recovery method of the present invention identifies, in response to an input/output error that occurs during file access processing, the extension 1 in which the error occurred in file allocation units, and By allocating a new replacement extent 1 in another volume and replacing the extent, you can recover the extent in a short time without stopping the job that was performing file access processing. It has the effect of being able to do something.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a data processing system to which a file recovery method according to an embodiment of the present invention is applied, and FIG. FIG. 2(b) is a diagram showing an example of a processing outline of the extent swap processing section of the extent swap means of FIG. 1. DESCRIPTION OF SYMBOLS 1... Update job group, 2... File access control means, 3... File device, 4... Exception handling means,
5... Extent swap means, 11... Extent management table, 12... Buffer, 21...
File access processing unit, 31.34...Volume, 32.35...Extension management unit, 33...
File, 36... Replacement extent. 41... Extent swap control unit, 42... Exception input/output processing unit, 51... Extend 1 identification processing unit, 52... Extent swap processing unit. -1,2-

Claims (1)

[Scope of Claims] A file device for storing a file having an extent management unit that manages extent information of a file space, and an extent management table in which the extent information managed by the extent management unit is written in advance and the file. an update job group having a buffer for temporarily holding data read from the file or data written to the file, and data read from the file to the buffer according to a request from the update job group and stored in the buffer to the file. In a data processing system comprising: a file access control means having a file access processing unit that accesses the file by writing data, a failure has occurred in a part of the file to which the file access processing unit accesses the file; The method includes: an exception input/output processing unit that retries the file access when an input/output error occurs during file access processing performed by the file access processing unit of the file access control means; an exception handling means having an extent swap control section that instructs file recovery if a retry by the exception input/output processing section fails; an extent identification processing unit that identifies an extent in which an error has occurred and displays that fact in the extent information managed by the file management unit that corresponds to this extent; secures a replacement extent with the same extent size as the extent size, and enters the extent information of the relevant extent and the extent information of the replacement extent with respect to the extent information managed by the extent management unit and the extent information written in the extent management table. 1. A file recovery method, comprising: an extent swapping unit having an extent swapping processing unit for changing an extent;