JP2822869B2 - Library file management device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a library file management apparatus, and more particularly, to a library file management apparatus characterized in detecting and recovering from a failure caused by a system failure or the like.

[0002]

2. Description of the Related Art Data groups independent of each other are generally stored in separate files. However, when the data groups have the same attribute, it is often efficient to centrally manage them as one file in terms of file assignment, space management, and access. The library file management device enables such centralized management, and individual data groups are managed as subfiles.

A library file is generally composed of three areas: a directory area, a block allocation map (BAM) area, and a data area. The directory area is composed of blocks called directory blocks, and manages description information of sub files. The BAM area is composed of blocks called BAM blocks.
Manage the free space in the data area. The data area is composed of blocks called data blocks, and stores data of each subfile.

[0004] In the library file, the allocation of blocks used by the subfile is managed in units of logical tracks (collection of blocks). In a method of extracting an area in units of one logical track as a logical track allocation method, when one logical track is composed of one block,
Every time one block is taken out, input / output of a BAM block occurs and processing efficiency is greatly reduced.

Therefore, in the access to the library file, when an empty logical track is needed, some empty logical tracks are taken out from the BAM area once, pooled in a control table on the memory, and the empty logical tracks in the pool are obtained. When there are no more tracks, some logical tracks are taken out of the BAM area again. When an unnecessary logical track is released, it is returned to the pool. When the pool is full, a list of the remaining unnecessary logical tracks is created, and the list is returned to the BAM area based on the list. ing.

[0006]

In the above-described conventional library file management apparatus, there is a possibility that inconsistency may occur between the data area and the BAM area when the system is down or the job is aborted. For example, if a free space is secured from the BAM area and it is determined that the data area is being used, if a system failure or the like occurs before data is still output to the data area at that time, invalid data is displayed on the data area. Regardless, information indicating that the logical track of the data area is being used is set in the BAM area.
There is a problem that inconsistency occurs and inconsistency occurs in the contents of the directory area, thereby preventing accurate access.

SUMMARY OF THE INVENTION The present invention solves the above-mentioned drawbacks, and can recover the inconsistency of the data area and the BAM area or the inconsistency of the directory area by the failure recovery function even when a failure occurs due to a system failure or the like. It is an object of the present invention to provide a library file management device capable of performing the above.

[0008]

In order to achieve the above object, the present invention provides a data area for storing data of a subfile, a directory area for storing a start address and an end address of the data area, In a library file management apparatus that manages a library file including an allocation map area that stores information indicating whether or not a file is used, the data area and the directory area include time stamp information for each subfile, and There is provided a recovery means for recovering inconsistency of information stored in the directory area or the allocation map area based on the time stamp information of the directory area.

According to another preferred aspect, when accessing the sub-file stored in the data area, the time stamp information of the data area and the time stamp information of the directory area are compared, and if they do not match, it is detected that a failure has occurred. It is configured to include a failure detection unit.

According to another preferred aspect, the restoration means determines whether or not the time stamp information of the data area used in the information of the allocation map area exists in the directory area. Determine, if not present, invalidate the data area, change the information of the allocation map area to unused, recover the inconsistency of the allocation map area, and start from the data area indicated by the start address of the directory area. If the data area where the time stamp information is different from the directory area is detected by performing a sequential search, the data area up to the data area immediately before the data area is validated, and the address of the data area immediately before the directory area is changed to the directory area. The inconsistency of the directory area is restored as the last address of the area.

[0011]

Next, a preferred embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the overall configuration of the library file management device of the present invention. In FIG. 1, the library file management device includes a sub file access processing unit 2, a failure detection processing unit 3, a failure recovery processing unit 4, a memory table 5, and a storage device 11 for storing a library file 12.

The sub-file access processing unit 2 creates, adds,
Perform processing such as deletion. Further, the failure detection processing unit 3 detects that a failure has occurred in the library file 12. Further, the failure recovery processing unit 4 performs processing for recovering the contents of the library file 12 in which the failure has occurred. The memory table 5 stores the time stamp data of the directory block at the time of the failure recovery processing.

As the storage device 11 for storing the library file 12, a device capable of random access is used, and for example, a magnetic disk or an optical disk is used. The storage device 11 stores one or more library files 12. One library file 12 includes one or a plurality of subfiles.

FIG. 2 shows the structure of the library file 12 stored in the disk device 11. The library file 12 includes a directory area 13, a BAM area 14, and a data area 15. Directory area 13
Is composed of a set of directory blocks 21,
The BAM area 14 is constituted by a set of BAM blocks 22, and the data area 15 is constituted by a set of data blocks 23.

FIG. 3 shows a configuration example of each area of the library file 12. A directory block 21 constituting the directory area 13 includes management information 21a and a plurality of directory entries 21b. Management information 21
The number of used bytes is written in a. The directory entry 21b has a subfile name 25a, a start block address 25b, an end block address 25c, a time stamp 25d, and other information 25e.

In the subfile name 25, a name given to each subfile is written. Start block address 25b
The address of the first data block 23 and the address of the last data block 23 in which the data of the subfile is written are set in the end block address 25c. In the time stamp 25d, the date and time when the subfile is created are set. In addition, other information 25e
, An identifier of the sub file is set.

The BAM block 22 constituting the BAM area 14 includes a NEXT pointer 22a, a PRIOR pointer 22b, and a BAM description 22C. The next B is stored in the NEXT pointer 22a and the PRIOR pointer 22b.
Location of AM block 22 and previous BAM block 2
Set the position of 2. The BAM description 22C includes a formatted logical track number 24a, an unused logical track number 24b, and an allocation map 24c indicating whether or not each data block 23 is used. Assignment map 24c
Is set to “1” when the data block 23 is in use and “0” when it is not used.

The data block 23 constituting the data area 15 includes a NEXT pointer 23a, a PRIOR pointer 23b, other information 23c, a time stamp 23d,
It is composed of a plurality of records 1 to n. The position of the next data block 23 and the position of the immediately preceding data block 23 are set in the NEXT pointer 23a and the PRIOR pointer 23b. In the other information 23c, the number of records included in the block, the number of bytes in the block, and the like are set. In the time stamp 23d, similarly to the time stamp 25d of the directory entry, the date and time when the sub file is created are set. Data is written in each record.

Here, the processing contents when creating a subfile in the library file 12 will be described with reference to FIGS. FIG. 4 shows a state when a sub file is opened in the library file 12. First, a directory entry 21b corresponding to a new subfile is created in the directory block 21.

The file name is set in the subfile name 25a of the directory entry 21b, the address of the data block 23 secured first is set in the start block address 25b, and the end block address 25c
Is set to "0". Also, time stamp 25d
Is set by acquiring the date and time at the time of creation. FIG. 4 shows a case where two data blocks 23 are stored in one logical track.

FIG. 5 shows a state in which data is output to the secured data block 23. Data is sequentially written to the secured data block 23. When the data blocks are insufficient, the data blocks 23 are secured and added in logical track units. NEXT pointer 23a, PRIOR pointer 23
b. Also, the directory entry 21
The same value as the time stamp 25d set in b is set in the time stamp 23d of the data block 23.

FIG. 6 shows a state when the subfile is closed. When all data has been written,
Last block address 2 of directory entry 21b
In 5c, the address of the last data block 23 is written.

In a failure due to a system failure at the time of creating a sub-file, if data is not updated in a data block which is in use due to the setting of the BAM block 22, the last block address 25 of the directory entry 21b is not updated.
A case occurs where the value of b does not match the address of the last data block 23. The recovery from such a failure is performed by the failure recovery processing unit 4.

Here, the failure detection by the failure detection processing unit 3 when the sub-file is accessed without executing the failure recovery function and the recovery processing by the failure recovery processing unit 4 when a failure is detected will be described. Here, a case will be described as an example where a system down occurs after writing data to the data block 23 (5) in the data output of FIG.

First, the failure detection processing section 3 determines whether or not the end block address 25c of the directory entry 21b is "0". If it is “0”, it is determined that the failure is before the subfile is closed, and the directory entry 2
The first data block 23 (1) is searched based on the start block address 25b of the 1b, and the NEXT pointer 23
The data blocks 23 are sequentially searched with reference to a. At that time, the time stamp of the data block 23 is compared with the time stamp of the directory entry 21b.
If the two time stamps do not match, it is detected that the system has gone down before writing the data in the data block 23.

In the example of FIG. 5, the data block 23 (5)
Is completed, and the NEXT pointer 2
3a shows the data block 23 (6). However, the data writing to the data block 23 (6) has not been completed, and therefore, a mismatch is detected by comparing the time stamp with the data block 23 (6).

When the inconsistency as described above is detected by the failure detection processing unit 3, the failure recovery processing unit 4 validates the data block 23 (5) up to the immediately preceding data block and checks the last block address 25c of the directory entry 21b. , The address of the data block 23 (5) is set, and a value (NUL) indicating the last block is set in the NEXT pointer 23a of the data block 23 (5).

At the same time, the user is warned to execute the failure recovery function because there is a possibility that there is a data block which is in use on the BAM block 22 but not used on the data block 23. .

When a record is added to the end of the subfile, the corresponding subfile is opened in the addition processing mode, and the time stamp 25d is obtained from the directory entry 21b. Directory entry 2
If there is a free space in the last data block 23 indicated by 1b, a record is added thereto. If there is no sky, a new data block 23 is secured. Then, the time stamp obtained from the directory entry 21b is set in the secured additional data block 23. When all the records have been added, the sub file is closed, and the address of the last data block 23 added to the end block address 25C of the directory entry 21b is set and updated.

When a system failure occurs when a record is added to a subfile, a mismatch state similar to that at the time of creating the subfile described above may occur. Recovery is performed in the same manner as described above. In addition, without executing the failure recovery function,
In the failure detection when the corresponding subfile is accessed, the last data block is searched by the last block address 25c of the directory entry 21b.
If the NEXT pointer 23a of the data block 23 is a value other than the value (NUL) indicating the last block, it can be detected that a system failure has occurred during the addition.

When this state is detected, data blocks are sequentially searched based on the NEXT pointer 23a, and a time stamp mismatch is detected. After detecting the mismatch of the time stamps, a recovery process similar to the above-described process at the time of creating the subfile is performed.

In the process of deleting a subfile, information on the subfile to be deleted is first deleted from the directory entry 21b. Next, all logical tracks used by the sub file are released from the BAM block 22. The data block 23 is not initialized to reduce overhead by reducing I / O.

When the system goes down when the subfile is deleted, a state occurs in which the BAM block 22 is in use but the corresponding subfile does not exist. For recovery, the same recovery processing as the above-described processing at the time of creating the subfile is performed.

Next, the failure recovery function of the failure recovery processing unit 4 will be described in detail with reference to FIGS. First, recovery of a failure of the BAM block 22 by checking inconsistency between the data area 15 and the BAM area 14 will be described. As shown in FIG. 7, all directory blocks 21 on the directory area 13 are read. The read contents are sorted using the time stamp as a key and stored in the memory table 5. Next, the first BAM block 22 is read from the BAM area 14. Then, the contents of the data block 23 set as being used in the allocation map 24c on the BAM block 22 are read. If it is an unused data block on the BAM block 22, no particular reading is performed.

It is checked whether the time stamp 23d of the read data block 23 exists in the memory table 5. If the data block does not exist, it is determined that the data block is invalid due to a system down or the like, and if the allocation map 24c of the BAM block 22 is "1" in use, it is set to the unused state "0". The above processes are sequentially performed on all the data blocks 23. FIG.
5 shows only the value of the assignment map 24c in the BAM description 22c of the BAM block 22.

In the example shown in FIGS. 7 to 9, as shown in FIG. 8, the first data block of the block number 1 and the second data block 23 of the block number 2 are in use on the BAM block 22. Is read. Since the time stamp matches the value of file C in FIG. 7, it is determined as a correct data block. Since the third and fourth data blocks are unused “0” on the BAM block 22, they are not read.

The time stamp of the fifth data block 23 which is in use matches that of the file D. The time stamp of the sixth data block 23 has no corresponding value on the memory table 5. Therefore, it is determined that the data block 23 is unused, and the state of the allocation map 24c on the BAM block 22 is changed to unused.

Next, a description will be given of a process of correcting inconsistency of the directory entry 21b as a second recovery process. First, the first directory entry 21b is read. The head data block 23 of the corresponding subfile is searched from the start block address 25b of the read directory entry 21b.

Next, the NE of the searched data block 23 is
The data blocks 23 are sequentially read by referring to the XT pointer 23a.
And the search is performed until the NEXT pointer 23a indicates the last data block. The time stamp value of the searched last data block 23 is compared with the time stamp of the directory block. As a result of the comparison, if they are different, the data block up to the previous data block 23 is validated, and the last block address 25c of the directory entry 21b is updated by setting the address of the previous data block 23. This process is executed for all directory entries 21b. The recovery of the library file is completed by the above failure recovery processing. Although the present invention has been described with reference to the preferred embodiments,
The present invention is not necessarily limited to the above embodiment.

[0040]

As described above, according to the library file management apparatus of the present invention, even if a failure occurs due to a system failure or the like, the inconsistency in the setting of the data area and the BAM area or the inconsistency of the directory area can be achieved by the failure recovery function. Since the inconsistency can be recovered, the subfile can be correctly accessed even if a failure occurs. In addition, even before recovery from the failure, it is possible to detect that the failure has occurred by accessing.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an overall configuration of a library file management device of the present invention.

FIG. 2 is a diagram showing a configuration of a library file managed by the present invention.

FIG. 3 is a diagram showing a configuration of a block constituting each area included in a library file.

FIG. 4 is a diagram showing an open state when a subfile is created.

FIG. 5 is a diagram showing a data write state when a subfile is created.

FIG. 6 is a diagram illustrating a closed state when a subfile is created.

FIG. 7 is a diagram showing an example of the contents of a directory block.

FIG. 8 is a diagram illustrating a content example of a BAM block.

FIG. 9 is a diagram showing an example of the contents of a data block.

[Explanation of symbols]

 1 Library File Management Device 2 Sub File Access Processing Unit 3 Failure Detection Processing Unit 4 Failure Recovery Processing Unit 5 Memory Table 11 Storage Device 12 Library File 13 Directory Area 14 BAM Area 15 Data Area 21 Directory Block 22 BAM Block 23 Data Block

Continuation of front page (56) References JP-A-4-113426 (JP, A) JP-A-4-314145 (JP, A) JP-A-4-370848 (JP, A) JP-A-5-35568 (JP) , A) JP-A-5-108448 (JP, A) JP-A-5-134895 (JP, A) JP-A-5-257778 (JP, A) Edited by ASCII Publishing Bureau, "Standard MS-DOS Handbook", ASCII Co., July 10, 1984, p. 95-97 (See “CHKDSK”) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06F 12/00 531

Claims (3)

(57) [Claims] 1. A data area for storing data of a subfile, a directory area for storing a start address and an end address of the data area, and an allocation map area for storing information indicating whether or not the data area is used. In the library file management device for managing the library file, the data area and the directory area include time stamp information for each sub-file, and the directory area or the directory area is provided based on the time stamp information of the data area and the directory area. A library file management device, comprising: recovery means for recovering inconsistency of information stored in an allocation map area. And a failure detecting means for comparing time stamp information of the data area with that of a directory area when accessing a sub-file stored in the data area, and detecting that a failure has occurred in the case of a mismatch. The library file management device according to claim 1, wherein: 3. The recovery means determines whether or not the time stamp information of the data area used in the information of the allocation map area exists in the directory area. Invalidate the data area, change the information of the allocation map area to unused, restore the inconsistency of the allocation map area, sequentially search from the data area indicated by the start address of the directory area, and execute the time stamp. If the data area whose information is different from the directory area is detected, the data area up to the data area immediately before the data area is validated, and the address of the data area immediately before is set as the last address of the directory area. 2. The library file management device according to claim 1, wherein the inconsistency of the library file is restored.