JPH1011345A - System and method for file management - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a system and method for file management which can repair hierarchical relation, whose consistency is lost owing to trouble, under simple control. SOLUTION: Unique IDs 48 which are enumerated uniquely in a tree structure are given to (i) nodes 47 held corresponding to the respective nodes constituting the tree structure, and when an (i) node 47 is updated, the unique ID 48 is also newly enumerated. Each time the (i) code 47 is updated, maintenance data 42 including the update state and the latest enumerated unique ID are generated and held in a volume maintenance space 41. If trouble occurs during the update of the (i) node 47, the latest maintenance data 42 are retrieved from the maintenance data 42 showing the completion of update states and the (i) node 47 including the enumerated unit ID 47 newer than the unique ID 43 included in the retrieved maintenance data 42.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a file management system and a file management method for managing data such as files and directories in a tree structure. The present invention relates to a file management system and a file management method that can be performed.

[0002]

2. Description of the Related Art In recent years, computer systems have been introduced in various industries.
It is becoming indispensable for business execution. In addition, with this introduction, the amount of data to be handled has become enormous, and the types of data have been diversified. In such a situation, how efficiently the desired data can be searched greatly affects the performance evaluation of the computer system. In order to perform this data search more efficiently, various data structure concepts have been created to date, and a tree structure exists as one of them.

[0003] The tree structure is a structure in which the shape of the tree is conceptually constituted by the hierarchical relation of data, and has a structure in which data such as files and directories are stored at each node. For example, if your operating system
In a file system constructed by a computer adopting NIX (hereinafter referred to as UNIX computer),
Some files are managed by being divided by directories, and this directory includes an address pointer that points to a subordinate file (which may be a directory). That is, the hierarchical structure of the directory and the files forms the above-described tree structure.

In a file system of a UNIX computer, a disk index node (disk i) is used to indicate files and directories under the file system.
ndex node: Hereinafter, referred to as an inode. ). This inode is composed of mode, owner name, time stamp, size, number of blocks, reference count, direct block and indirect block.

FIG. 7 shows an example of a tree structure of a file system constructed by a UNIX computer.
Here, it is assumed that the file system shown in FIG. 7 is constructed on a write-once storage medium. In this case, for example, when an update such as adding new data to a file whose file name is "file" is performed, the storage area of the data that is the substance of this "file" is moved. It is necessary to update the address (X5) pointing to the storage area of. Therefore, i for bin holding this address (X5)
Since the node also needs to be updated, and updating the bin i-node means moving the storage area of the bin i-node, the root holding the address (X3) indicating the storage area of the bin i-node is updated. The inode for the directory must be updated. That is, when a tree-structured file system is constructed on such a write-once type storage medium, the file under the
All hierarchically related inodes must be updated.

In a conventional file system, data such as a file or a directory to be updated is stored in one i.
It is configured to be managed by the node. Therefore,
If a failure occurs during a series of updating of the i-node as described above, there is a problem that the i-node is destroyed and the file system becomes abnormal.

Further, in the conventional file system, since each directory is pointed to each i-node, even if a failure occurs during the updating of the directory, the directory is destroyed and the i-node is lost. As a result, the file system cannot be referred to.

Further, the conventional file system has a problem that if the file system becomes abnormal due to such causes, the file system cannot be restored to the original state.

[0009]

As described above, in the conventional tree-structured file system, the hierarchical relationship is managed by the i-node, so that a failure occurs during the updating of the i-node. In such a case, there has been a problem that the consistency of the hierarchical relationship is lost and the file system becomes abnormal.

This inode is referred to by a pointer included in the directory.
If the directory is destroyed, it becomes impossible to refer to the inode, and similarly, there has been a problem that the file system is abnormal.

Further, the conventional file system has a problem that, when the file system becomes abnormal due to such causes, the file system cannot be restored to the original state.

The present invention has been made in view of such circumstances, and provides a file management system and a file management method capable of restoring, with simple control, a hierarchical relationship whose consistency has been lost due to a failure. The purpose is to:

[0013]

SUMMARY OF THE INVENTION The present invention is a tree-structured file management system for storing data such as files and directories at each node, and stores node information including a hierarchical relationship between the nodes. In the file management system that stores the node information, the node information includes an identification number uniquely assigned in the tree structure as a component thereof, and the node information is updated with the update of the data. When updating, the numbering means for newly numbering the identification numbers included in the series of node information groups to be updated, and the update status of the series of node information groups and the numbering by the numbering means. Security information storage means for storing security information generated each time the data is updated, including the latest identification number; and when a failure occurs while updating the node information, the update status is updated. Retrieves the latest security information from the security information indicating completion from the security information storage means, and repairs the node information from the identification number included in the searched security information and the identification number included in the node information. And a repairing means.

In the present invention, when a series of related node information is updated, for example, when data is updated, security information corresponding to the node information group is generated and stored. In the present invention, the node information includes an identification number uniquely assigned in the tree structure. The identification number is assigned a unique number by, for example, incrementing an initial value given in advance by the numbering means. When the node information is updated, the identification number included in the node information is newly assigned and rewritten.

On the other hand, the maintenance information generated by updating the series of node information groups includes the status of the update and the latest identification number assigned by the update. When the numbering means increments the number, the identification number with the largest number is the latest identification number.

Here, it is assumed that a failure occurs while updating the node information. If a failure occurs during the update of the node information, in the present invention, the update status included in the latest maintenance information indicates that the update is in progress, thereby making it possible to know the necessity of the repair work. . Then, the restoration work is performed by the restoration means, and the restoration means first searches for the latest maintenance information from the maintenance information indicating that the update status is completed. At this time, the state of the point closest to the failure point at which consistency is guaranteed is held in the retrieved maintenance information, and the node information to be repaired is obtained by obtaining the identification number included in the maintenance information. Identify. In this case, since the numbering means increments the number, the node number whose identification number is larger than the identification number included in the searched maintenance information is replaced with the node information. What is necessary is to make it a restoration target.

For example, in a write-once type storage medium, when updating data, the storage area of the data before update, that is, the data to be updated is set to an undefined state, and the updated data is stored in a newly secured area. I do. Therefore, the repairing means deletes the node information including the identification number having a value larger than the identification number included in the retrieved security information (changes the state to an undefined state), and sets the state to the undefined state when the node information is stored. The node information before the update is returned to be accessible.

That is, in the present invention, the data before updating is used as an archive, so that the integrity of the file system is dramatically improved and the resources are effectively used.

[0019]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a schematic configuration of a computer system according to an embodiment of the present invention. Upon receiving an input request to the disk device 40 from the application program (including the utility program) 20, the file management system 10 of the present embodiment reads out the requested data from the disk device 40 and stores it in the memory 30. On the other hand, when an output request to the disk device 40 is received, the requested data is read from the memory 30 and stored in the disk device 40.

The disk device 40 is a write-once storage medium, and is logically divided into a volume maintenance space 41 and a volume space 44. Therefore, even if logically incorrect input / output is performed on the data stored in the volume space 44, the effect does not reach the volume maintenance space 41.

The volume security space 41 holds security data 42 generated each time an update process occurs. This maintenance data 42 includes, as shown in FIG.
The latest unique ID for making the i-node unique, which will be described later, a security status indicating whether the volume space 44 is open or closed, and a status when the security data 42 is accessed. And a time stamp indicating the time.

In the volume space 44, an actual data group to be managed is held in a tree structure. In this tree structure, a directory 45 and a file 46 are stored at each node, and as information indicating a hierarchical relationship,
An i-node 47 is provided corresponding to each node composed of at least one of the directory 45 and the file 46. As shown in FIG. 3, this inode 47 has
A unique ID is included in addition to conventional information such as a mode and an owner name. The feature of this embodiment is that the unique ID is added as a component of each i-node 47, and its handling will be described below.

FIG. 4 shows an example of a tree structure of a file system constructed on the disk device 40. As shown in FIG. 4, this file system has a root directory as a tree root, and this root directory is
There are two subdirectories, user and bin, under the directory. This bin has two files, cat and file, under its control. Each contact formed by the directory or the file is provided with an inode indicating the hierarchical relationship. Therefore, the file management system 10
By tracing this inode, desired data is searched.

Now, the unique I included in each inode
It is assumed that a value as shown in FIG. 4 is set in D. That is, for example, "1" is set for the unique ID of the root directory i-node, "4" is set for the unique ID of the bin i-node, and "7" is set for the unique ID of the file i-node. It is assumed that The unique ID is assigned in ascending order from a preset initial value (here, “1”), and is incremented when a new inode is generated. A new unique ID that is incremented is also assigned when the ID is updated.

In the state shown in FIG. 4, "7" is set as the unique ID and "closed" is set as the security state in the latest security data 42 held in the volume security space 41. Will be.

Here, consider adding a file file1 under the subdirectory bin. In this case, as shown in FIG. 5, an i-node for pointing to an area where file1 is stored (i-node for file1)
Is generated. Then, the generated file1 i
In order to hold the address (X13) indicating the node, the bin i-node is updated, that is, the storage area of the bin i-node is moved. Further, a new address (X1) indicating the storage area of this bin inode is used.
An update of the inode for the root directory holding 2) occurs.

At this time, the file management system 10
One security data 42 is generated for this series of updates. At the start of the update process, the security status is set to “open”, and when the update process is completed, the latest (ie, maximum) unique ID is set and the security status is set to “closed”. In the case of a write-once type storage medium like the disk device 40 of the present embodiment, when updating data, the storage area of the data before update, that is, the storage area of the data to be updated is set to an undefined state, Store the data in the newly secured area.
Therefore, the inode before update is still held in the volume space 44, and the file management system 10 of this embodiment uses the data before update as an archive.

Here, it is assumed that a failure occurs during the updating of the inode. In this case, the volume maintenance space 4
The security status of the latest security data 42 held in 1 is
In this state, the file management system 10 knows that any of the inodes 47 is in an inconsistent state.

At this time, the file management system 10
The security data 42 held in the volume security space 41 is searched for the security data 42 whose security status is "closed" and whose time stamp is the newest. Then, the unique ID is extracted from the searched maintenance data 42, and from the i-nodes 47 held in the volume space 44, the i including the unique ID having a value larger than the unique ID of the searched maintenance data 42 is included. Search for node 47. Then, the retrieved inode 47 is set as a restoration target, the inode 47 is deleted, and the inode 47 which is an archive of the deleted inode 47 is restored.

Thus, even if a failure occurs during the updating of the inode 47, the integrity of the damaged inode 47 can be restored, and the integrity of the disk device 40 can be improved. Becomes

Here, the operation principle of ensuring the consistency of the file management system 10 will be described with reference to FIG. The file management system 10 determines whether or not the security data 42 exists in the volume security space 41 (step S1).
If the security data 42 does not exist (N in step S1), it recognizes that the disk device 40 is in the initialized state (step S2). On the other hand, if the security data 42 exists (Y in step S1), the file management system 10
Searches for the latest security data 42 from among them (step S3), and refers to whether the security status included in the searched security data 42 is "open" or "closed" (step S4).

If the security status included in the latest security data 42 is "closed" (N in step S4), the file management system 10
Are recognized to be completely maintained (step S
5). On the other hand, when the maintenance state is “open” (Y in step S4), the file management system 10
It recognizes that the consistency of the disk device 40 has been impaired. Then, the file management system 10 searches the security data 42 held in the volume security space 41 for the security data 42 whose security status is “closed” and whose time stamp is the newest. The unique ID included in the secured data 42 obtained is acquired (step S6).

Then, the file management system 10 searches the inode 47 including the unique ID having a value larger than the obtained unique ID from the inode 47 held in the volume space 44, and searches the inode 47. I-node 4 that has been deleted and archived
7 is restored (step S7).

Thus, the file management system 10
Can efficiently ensure the consistency of the disk device 40. In this embodiment, a write-once storage medium has been described as an example of a storage medium. However, the present invention is not limited to this. For example, the present invention can be applied to a rewritable storage medium.

[0035]

As described above in detail, according to the present invention, a unique identification number newly assigned at the time of updating is used as a component of the node information, and the latest identification number and the system security state are used. Since it is separately stored as security information, even if a failure occurs while updating the node information, the damaged hierarchy is determined by the identification number included in the security information and the identification number included in the node information. Since it is possible to restore a dynamic relationship, it is possible to dramatically improve the integrity of the entire system. Further, since the information before update is used as an archive, it is not necessary to secure a new storage area or the like, and it is possible to effectively use resources.

[Brief description of the drawings]

FIG. 1 is an exemplary view showing a schematic configuration of a computer system according to an embodiment of the present invention.

FIG. 2 is an exemplary view showing a configuration of security data according to the embodiment.

FIG. 3 shows a disk index node (i
FIG.

FIG. 4 is an exemplary view showing an example of a tree structure of a file system constructed on the disk device of the embodiment.

FIG. 5 is an exemplary view for explaining an update operation of the file system of the embodiment.

FIG. 6 is an exemplary flowchart for explaining the operation principle of ensuring consistency of the file management system according to the embodiment;

FIG. 7 is a diagram showing an example of a tree structure of a file system constructed by a computer adopting UNIX as an operating system.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... File management system 20 ... Application program 30 ... Memory 40 ... Disk device 41 ... Volume maintenance space 42 ... Security data 43 ... Unique ID 44 ... Volume space 45 ... Directory 46 ... File 47 ... Disk index node (inode) 48 … Unique ID

Claims (7)

[Claims] 1. A tree-structured file management system for storing data such as files and directories at each node, wherein a file holding node information including a hierarchical relationship between the nodes corresponding to each of the nodes. In the management system, the node information includes an identification number uniquely numbered in the tree structure in its constituent element. When the node information is updated along with the update of the data, the target A numbering means for newly numbering an identification number included in a series of node information groups, and an update status of the series of node information groups and a latest identification number assigned by the numbering means. Security information storage means for storing security information generated each time data is updated; and when a failure occurs during the update of the node information, the security status information indicates that the update status is completed. From the security information storage means, and repair means for repairing the node information from the identification number included in the searched security information and the identification number included in the node information. A file management system characterized by becoming. 2. The numbering means assigns a unique identification number by incrementing an initial value given in advance, and the repairing means assigns a unique identification number based on an identification number included in the retrieved security information. 2. The file management system according to claim 1, wherein node information including an identification number having a large value is also targeted for restoration. 3. A file management system in which the data and the node information are stored in a write-once storage medium, wherein the restoration unit deletes the updated node information,
3. The file management system according to claim 1, further comprising means for restoring node information before update, which is in an undefined state due to the update. 4. The file management system according to claim 1, wherein the node information and the security information are separately stored in logically independently secured areas in the same storage device. . 5. A file management system in which each of a plurality of storage devices forms a tree structure, wherein the numbering means numbers a unique identification number in each of the storage devices. The file management system according to 1, 2, 3 or 4. 6. A file management system in which each of a plurality of storage devices forms a tree structure, wherein the numbering means numbers an identification number unique in the system. 2, 3, or 4
The file management system described. 7. A tree-structured file management method for storing data such as a file and a directory at each node, wherein the node information including a hierarchical relation of each node is stored for each of the nodes. In the management method, the node information includes an identification number uniquely assigned in the tree structure as a component thereof. When the node information is updated along with the update of the data, the target A new identification number included in a series of node information groups is newly generated, and is generated each time the data including the update status of the series of node information groups and the latest numbered identification number is updated. The security information is stored, and when a failure occurs during the update of the node information, the security information storage means checks the latest security information from the security information indicating that the update status is completed. File management method is characterized by repairing the node information from the identification number included in the identification number and the node information included in the retrieved security information.