JP5481906B2 - Projection file system management apparatus, projection file system management method, and program - Google Patents

Description

  The present invention relates to a technique for managing a plurality of real file systems as one virtual file system.

  As a technique for managing a plurality of real file systems as one virtual file system, the following technique is known (see, for example, Patent Document 1). In Patent Document 1, a transparent virtual file system (transparent file system) is used in which a plurality of real file systems are used and the directory structure is a superposition of the directory structures of the plurality of real file systems. It is composed. Note that the directory structure is superimposed by combining paths with matching names into one, and leaving paths with unmatched names as they are. Files existing in the combined portion exist on a plurality of real file systems. Each real file system is given a priority. When there is a read request or a write request for a file on the transparent file system, the following processing is performed using the priority.

  When a read request is made, the requested file is searched in order from the highest-priority real file system, and data is read from the corresponding file existing in the real file system in which the corresponding file can be searched first. When a write request is made, the requested file is searched in order from the real file system having the highest priority. If the real file system that can first search for the relevant file is the highest-priority real file system, data is written to the relevant file existing in the real file system. In addition, if the actual file system that was able to search for the file first is other than the real file system with the highest priority, the searched file is copied to the real file system with the highest priority, and the copied file Write data to.

JP 2006-268456 A

  According to the technique described in Patent Document 1, a plurality of real file systems can be managed as one virtual file system. However, since the technology described in Patent Document 1 controls the priority in units of real file systems, when the user accesses a file on the transparent file system, the actual file system is accessed. This file is limited to a file on an actual file system having a high priority, and there is a problem that flexibility is lacking. For example, if the first real file system is a transparent file system having a higher priority than the first real file system, both the first and second real file systems are transparent. Even if the files f1 and f2 corresponding to the files F1 and F2 on the file system exist, the files accessible by the user are always limited to the files f1 and f2 on the first real file system. A file on the first real file system is accessed, and the file f2 cannot be accessed on the second real file system.

[Object of invention]
SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to make it possible to control on a file basis which file on a real file system is accessed by a user in a virtual file system composed of a plurality of real file systems. is there.

A first projection file system management apparatus according to the present invention includes:
Multiple real file systems,
A priority table in which priority information indicating the priority of each file managed by the plurality of real file systems is registered;
The plurality of real file systems are managed as one virtual projection file system in which their directory structures are superimposed, and responds to a processing request specifying a file on the projection file system input from a user process. The file having the highest priority among the files on the real file system corresponding to the specified file is obtained by referring to the priority table, and the requested file is requested by the processing request. Projection file system control means for performing the above operations.

A first projection file system management method according to the present invention includes:
A computer comprising a plurality of real file systems and a priority table in which priority information indicating the priority of each file managed by the plurality of real file systems is registered.
A first step of managing the plurality of real file systems as one virtual projection file system in which their directory structures are superimposed;
In response to the processing request specifying the file on the projection file system input from the user process, the computer has the highest priority among the files on the real file system corresponding to the specified file. A second step of obtaining a file with reference to the priority table and performing an operation requested by the processing request on the obtained file.

The first program according to the present invention is:
A computer having a plurality of real file systems and a priority table in which priority information indicating priority levels of files managed by the plurality of real file systems is registered is caused to function as a projection file system management apparatus. A program for
The computer constitutes one virtual projection file system in which the directory structures are overlapped by the plurality of real file systems, and a processing request specifying a file on the projection file system input from a user process In response, the file having the highest priority among the files on the real file system corresponding to the designated file is obtained by referring to the priority table, and the processing request for the obtained file is obtained. To function as a projection file system control means for performing the operation requested by.

  Advantageous Effects of Invention According to the present invention, in a virtual file system composed of a plurality of real file systems, it is possible to control in units of files which files on which real file systems are to be accessed by the user. Can be obtained.

FIG. 1 is a block diagram showing a configuration example of the first embodiment of the present invention. FIG. 2 is a diagram showing an image of the projection file system. FIG. 3 is a block diagram showing a configuration example of the second embodiment of the present invention. FIG. 4 is a diagram showing an example of the PFS control information file 53. As shown in FIG. FIG. 5 is a diagram showing an example of the PFS link information table 52. FIG. 6 is a diagram illustrating an operation at the time of opening. FIG. 7 is a diagram showing an operation at the time of release. FIG. 8 is a diagram showing an operation at the time of reading. FIG. 9 is a diagram showing an operation at the time of writing. FIG. 10 is a diagram showing an operation when deleting a file. FIG. 11 is a diagram showing the operation at the time of rename. FIG. 12 is a diagram showing the operation when changing the file size. FIG. 13 is a flowchart showing a processing example of the PFS control means 51 when getattr () is issued. FIG. 14 is a flowchart showing a processing example of the PFS control means 51 when open () is issued. FIG. 15 is a flowchart showing a processing example of the PFS control means 51 when truncate () is issued. FIG. 16 is a flowchart showing a processing example of the PFS control means 51 when read () is issued. FIG. 17 is a flowchart showing a processing example of the PFS control means 51 when write () is issued. FIG. 18 is a flowchart showing a processing example of the PFS control means 51 when release () is issued. FIG. 19 is a flowchart showing a processing example of the PFS control means 51 when unlink () is issued. FIG. 20 is a flowchart showing a processing example of the PFS control means 51 when mkdir () is issued. FIG. 21 is a flowchart showing a processing example of the PFS control means 51 when rmdir () is issued. FIG. 22 is a flowchart showing a processing example of the PFS control means 51 when mknod () is issued. FIG. 23 is a flowchart showing a processing example of the PFS control means 51 when rename () is issued. FIG. 24 is a flowchart showing an example of processing when searching for the latest generation file. FIG. 25 is a flowchart illustrating a processing example when determining the output destination (creation destination) of the next generation file.

  Next, embodiments of the present invention will be described in detail.

[First embodiment of the present invention]
Referring to FIG. 1, a projection file system management apparatus 1 according to the first embodiment of the present invention includes a user process 2, a projection file system control means 3, a priority table 4, and an actual file system group 5. Is included. The projection file system management apparatus 1 is realized by a computer.

  The real file system group 5 includes a plurality of real file systems 5-1 to 5-N. In the following description, N = 2. As shown in FIG. 2, the real file system 5-1 includes files FILE1 and FILE2 under the directory DIR1, and files FILE3 and FILE4 under the directory DIR3. The real file system 5-2 includes a file FILE2 under the directory DIR1, a file FILE2 under the directory DIR2, and a file FILE3 under the file system FS2. It is assumed that names FS1 and FS2 are assigned to the real file systems 5-1 and 5-2, respectively.

  In the priority table 4, priority information indicating the priority of each file existing on each real file system 5-1 and 5-2 is registered. In the present embodiment, it is assumed that information indicating a file creation date (registration date) is registered as priority information.

  The user process 2 issues a data read request, a first data write request, and a second data write request to the projection file system control means 3. The data read request designates a file on a projection file system to be read (projection file system will be described later) and requests data read. The first data write request specifies a file on the projection file system as a write destination and requests data write. The second data write request designates a file on the projection file system to be a copy source and requests to write data to the copy destination file (next generation file). The first data write request is issued when file generation management is unnecessary, and the second data write request is issued when file generation management is required.

  The projection file system control means 3 manages the plurality of real file systems 5-1 and 5-2 as one transparent virtual file system in which their directory structures are superimposed. Note that the directory structure is superimposed by combining paths with matching names into one, and leaving paths with unmatched names as they are. When the directory structures of the real file systems 5-1 and 5-2 shown in FIG. 2 are overlapped, for example, the path / DIR1 / FILE2 has the same name in both directory structures (since it exists in both directory structures) Synthesized into one. On the other hand, for example, the path / DIR1 / FILE1 does not match the name (it exists only in one directory structure), so it remains as it is. As a result, the directory structure of the projection file system is as shown in the lower part of FIG. In the projection file system, the file specified by the path / DIR1 / FILE2 and the corresponding file on the actual file system include the file specified by the path / DIR1 / FILE2 on the actual file system 5-1 and the actual file. There are two files: the file specified by path / DIR1 / FILE2 on system 5-2.

  The projection file system control means 3 performs processing in response to a processing request from the user process 2, and includes a request receiving means 3-1, processing target file determining means 3-2, request processing means 3- 3 and next-generation file creation means 3-4.

  The request accepting unit 3-1 has a function of accepting a data read request, a first data write request, and a second data write request from the user process 2.

  The processing target file determining unit 3-2 corresponds to the file on the projection file system specified by the data reading request, the first data writing request, or the second data writing request received by the request receiving unit 3-1. Among the files on the actual file systems 5-1 and 5-2, the file having the highest priority is obtained by referring to the priority table 4. In the present embodiment, it is assumed that a file with a new registration date / time has a higher priority.

  The request processing unit 3-3 has a function of performing processing according to the processing request received by the request receiving unit 3-1 using the file obtained by the processing target file determining unit 3-2. More specifically, when the processing request received by the request receiving unit 3-1 is a data read request, data is read from the file obtained by the processing target file determining unit 3-2 and returned to the user process 2. . If the processing request received by the request receiving unit 3-1 is the first data write request, the data requested by the processing request is added to the file obtained by the processing target file determining unit 3-2. Write. If the processing request received by the request receiving unit 3-1 is a second data write request, a next-generation file creation request including information indicating the file obtained by the processing target file determining unit 3-2 is issued. Next-generation file creation means 3-4 is passed to create the next-generation file, and the requested data is written to the next-generation file created by the next-generation file creation means 3-4.

  The next-generation file creation means 3-4 corresponds to the file specified by the information in the next-generation file creation request passed from the request processing means 3-3 and the file specified by the second data write request. Create a next generation file by copying it to one of the real file systems that don't exist. In addition, information regarding the priority (registration date) of the created next-generation file is additionally registered in the priority table 4.

  The projection file system control means 3 can be realized by a computer, and when realized by a computer, for example, as follows. A disk, a semiconductor memory, and other recording media recording a program for causing the computer to function as the projection file system control means 3 are prepared, and the computer reads the program. The computer realizes the projection file system control means 3 on its own computer by controlling its own operation according to the read program.

  Next, the operation of the present embodiment will be described.

  First, an operation when a data read request is issued from the user process 2 will be described.

  A data read request from the user process 2 is received by the request receiving means 3-1. Now, for example, assume that a file (/ mount-point / DIR1 / FILE2) on the projection file system is specified by a data read request.

  The processing target file determining unit 3-2 refers to the priority table 4, and refers to the files (/ mnt / pfs / FS1 / DIR1 / FILE2, Select the file with the highest priority from mnt / pfs / FS2 / DIR1 / FILE2). Now, for example, if the registration date and time of the corresponding file on the real file systems 5-1 and 5-2 is 2008/10 / xx, 2008/09 / yy, the processing target file determination means 3-2 The file (/ mnt / pfs / FS1 / DIR1 / FILE2) on the real file system 5-1 is selected. The information representing the actual file systems 5-1 and 5-2 corresponding to the files on the projection file system is the information representing the files on the projection file system (such as / mount-point / DIR1 / FILE2). It can be obtained mechanically by replacing mount-point / with / mnt / pfs / FS1 /, / mnt / pfs / FS2 /.

  When the file (/ mnt / pfs / FS1 / DIR1 / FILE2) on the real file system 5-1 is selected by the processing target file determination unit 3-2, the request processing unit 3-3 reads data from the selected file. Return to user process 2.

  Next, an operation when a first data write request is input from the user process 2 will be described. The first data write request from the user process 2 is received by the request receiving unit 3-1. Now, for example, assume that a file (/ mount-point / DIR1 / FILE2) on the projection file system is designated by the first data write request.

  The processing target file determination unit 3-2 performs the same operation as when a data read request is input, and the file on the real file system 5-1 (/ mnt / pfs / FS1 / DIR1 Select / FILE2).

  The request processing means 3-3 requests the user process 2 for the file (/ mnt / pfs / FS1 / DIR1 / FILE2) on the real file system 5-1 selected by the processing target file determining means 3-2. Written data. Further, the information on the priority of the file registered in the priority table 4 is updated to the current date and time.

  Next, an operation when a second data write request is issued from the user process 2 will be described. The second data write request from the user process 2 is received by the request receiving unit 3-1. Now, for example, assume that a file (/ mount-point / DIR1 / FILE2) on the projection file system is designated by the second data write request. The processing target file determination unit 3-2 performs the same operation as when a data read request is input, and the file on the real file system 5-1 (/ mnt / pfs / FS1 / DIR1 Select / FILE2).

  The request processing unit 3-3 generates a next-generation file creation request including information indicating the file (/ mnt / pfs / FS1 / DIR1 / FILE2) selected by the processing target file determination unit 3-2. Pass to 3-4.

  The next generation file creation means 3-4 uses the file (/ mnt / pfs / FS1 / DIR1 / FILE2) specified by the information in the next generation file creation request passed from the request processing means 3-3 as the second Create the next generation file by copying it to one of the actual file systems that do not have a file corresponding to the file (/ mount-point / DIR1 / FILE2) on the projection file system specified by the data write request To do. In addition, information regarding the priority (registration date) of the created next-generation file is additionally registered in the priority table 4. The actual file system to be copied can be, for example, an actual file system with the largest available capacity or an actual file system with the lowest load. Further, when the file (/ mount-point / DIR1 / FILE2) and the file corresponding to all the real file systems 5-1 and 5-2 exist, for example, the following processing is performed. That is, the file with the lowest priority among the files on the real file systems 5-1 and 5-2 corresponding to the file (/ mount-point / DIR1 / FILE2) is deleted, and a request is made instead of the deleted file. Copy the file specified by processing means 3-3. In addition, the priority of the next-generation file is registered in the priority table 4 instead of the priority of the deleted file.

  When the next generation file is created by the next generation file creation means 3-4, the request processing means 3-3 writes data in the file. Further, the information on the priority of the file registered in the priority table 4 is updated to the current date and time.

  According to the present embodiment, in a virtual projection file system composed of a plurality of real file systems, it is possible to control in units of files which files on the real file system are to be accessed by the user. The effect of becoming can be obtained. The reason is that a priority is given to a file on the real file system, and a file to be accessed is determined based on this priority.

  Further, according to the real file system, it is possible to perform generation management of files. The reason is that, in response to the second data write request input from the user process 2, the file having the highest priority among the files on the real file system corresponding to the file specified by the second write request. Is obtained by referring to the priority table 4, and the obtained file is copied to one of the real file systems in which no file corresponding to the file specified by the second data write request exists. This is because the generation file is created and the registration date and time related to the next generation file is registered in the priority table 4.

  Furthermore, in the present embodiment, when there are a plurality of real file systems in which there is no file corresponding to the file specified by the second write request, the real file system having the largest free space or the largest load among them. Since the next generation file is created in a low real file system, it is possible to prevent the load from being concentrated on a specific real file system.

[Second embodiment of the present invention]
Next, a second embodiment of the present invention will be described.

  Referring to FIG. 3, an operating system 100 according to the second embodiment of the present invention includes a user process 10, a virtual file system 20, a user space file system 30, a real file system group 40, and a projection file system. Mechanism 50. In the present embodiment, it is assumed that the operating system 100 is UNIX (registered trademark). Hereinafter, PFS is an abbreviation that indicates a projection file system.

  The real file system group 40 includes a real file system 41 and a real file system 42. Note that a PFS control information file 53 is stored in all directories in each real file system 41 and 42, and the file name and PFS registration date and time are included for each file in the directory. Generations are managed by PFS registration date and time. An example of the PFS control information file 53 is shown in FIG.

  The projection file system mechanism 50 of FIG. 3 includes PFS control means 51 and a PFS link information table 52. The PFS link information table 52 is a table that holds link information for each opened file. The link information includes a file name, a generation protection flag, a write flag, an actual file path, and an open counter. An example of the PFS link information table 52 is shown in FIG. In addition, the projection file system mechanism 50 manages the plurality of real file systems 41 and 42 as one transmissive virtual projection file system in which their directory structures are superimposed.

  The user process 10 requests all file access processing from a virtual file system (VFS) 20.

  The virtual file system 20 is a virtual file system that conceals differences between interfaces provided by individual file systems and provides a common interface in order to handle a plurality of file systems in a unified manner on one operating system. The file access processing is requested from the user to the real file system 41, the real file system 42, and the user space file system (FUSE; Filesystem in Userspace) 30. The virtual file system has already been realized and spread with VFS in Linux and the like, and in this embodiment, a unified interface is provided by the virtual file system when the projection file system is realized.

  The user space file system 30 calls the PFS control means 51 and performs file access processing on the real file system 41 and the real file system 42. In this case, the PFS control means 51 performs actual file access processing through the virtual file system 20. In general, the file system operates in the kernel space at the destination of system calls such as read / write provided by the operating system, and it takes a lot of time and restrictions to construct the file system in the kernel space. The user space file system is a series of mechanisms for building a file system in the user space, and has already been realized and popularized by FUSE in Linux and the like. In this embodiment, the projection file system mechanism 50 is also installed in the user space. It is assumed that the user space file system 30 is provided in order to realize this. Therefore, when the projection file system mechanism 50 is provided in the kernel space, the user space file system 30 is not necessary.

  The PFS control means 51 refers to and updates information for the file entry in the PFS link information table 52 and the PFS control information file 53.

  Next, the operation of the present embodiment will be described.

  First, a description will be given of how a file access request by a user is notified to the PFS control means 51 through the VFS 20 and processing is executed. Next, how the VFS operation is called by extension of file access by the user and how the PFS control means 51 performs processing will be described with reference to FIGS. In particular, the details of the processing of each VFS operation are described in the flowcharts of FIGS. The following embodiments do not cover all operations constituting the file system, and are eight general file access processes (new write / overwrite write / additional writing) that are important for generation management realized in the present invention. write / read / rename / mkdir / rmdir / unlink).

● Notification to the PFS control means 51 of VFS operation by user request and return of execution result

1. A VFS operation is issued from the user process 10 to the VFS 20 by extension of file access by the user.
2. The VFS 20 transfers the issued operation to the FUSE 30.
3. The FUSE 30 transfers the issued operation to the PFS control means 51.
4). The PFS control means 51 processes the issued operation and notifies the FUSE 30 of the result.
5. The processing result is notified to the user process 10 through FUSE30 and VFS20.

● Writing to a new file

  1. The user process 10 executes the open function.

  1-1. The user process 10 requests (getattr) to acquire the attribute of the write target file in the open function. The PFS control means 51 searches for the latest generation file according to the latest generation file search logic (step S1301 in FIG. 13), and returns the information to the user process 10 (step S1302 is no, S1304). The latest generation file search logic will be described later with reference to FIG.

  1-2. The user process 10 confirms that the file does not exist in the open function, and requests creation (mknod) of the file. The PFS control means 51 determines a real file system as a file creation destination according to a next-generation file creation destination decision logic (described later; FIG. 25) (step S2201 in FIG. 22), creates a file (step S2202), A PFS control information file 53 is created, and a file name and a PFS registration date are registered (step S2203).

  1-3. In the open function, the user process 10 requests acquisition of the attribute of the write target file (getattr), and acquires information on the newly created file. The PFS control means 51 searches for the latest generation file and returns the information to the user process 10 (steps S1301 to S1303 in FIG. 13).

  1-4. The user process 10 requests opening of the newly created file in the open function (FIG. 6). The PFS control means 51 creates an entry in the PFS link information table 52. In the entry, a file name on PFS, a generation protection flag, a write flag, an actual file name, and an open count are registered (steps S1401 to S1406 in FIG. 14). If it has already been opened by another process, an entry is acquired, the write flag is turned ON, and the open count is incremented (step S1401 is yes, S1407 to S1409, S1406). When the writing has been opened by another process (step S1408 is yes), an abnormality is notified.

  2. The user process 10 executes a write function.

  2-1. In the write function, the user process 10 requests the writing of the opened file by designating an offset and a buffer (FIG. 9). The PFS control means 51 acquires the real file name from the corresponding entry on the PFS link information table 52, acquires data from the specified buffer, and writes it from the specified offset (steps S1701, S1702, FIG. 17). S1708).

  3. The user process 10 repeats step 2 until all data is written, and after completion, executes the close function.

  3-1. The user process 10 requests release of the file in the close function (FIG. 7). The PFS control means 51 decrements the open count of the entry corresponding to the file on the PFS link information table 52, and deletes the entry if it is 0 (steps S1801 to S1804 in FIG. 18). If there is still an open process, the write flag is turned off and the generation protection flag is turned on (step S1803 is no, S1805 is yes, S1806).

● Truncation and writing of existing files

  1. The user process 10 executes the open function.

  1-1. The user process 10 requests (getattr) to acquire the attribute of the write target file in the open function. The PFS control means 51 searches for the latest generation file and returns the information to the user process 10 (FIG. 13, steps S1301 and S1302 are yes, S1303).

  1-2. The user process 10 checks the existence of a file, an access right, etc., and requests (truncate) a file size change (size is set to 0) (FIG. 12). The PFS control means 51 searches for the latest generation file (step S1501 in FIG. 15), and if the file size is not 0 and the specified size is different from the current size (both steps S1502 and S1503 are no), Determine the actual file system that will be the file creation destination according to the next-generation file creation destination decision logic (described later), copy the file for the specified size, create the PFS control information file 53, file name and PFS registration date Is registered (steps S1505 to 1507). If the file size is 0 and the specified size is also 0 (yes in steps S1502 and S1504), nothing is done.

  1-3. In the open function, the user process 10 requests acquisition of the attribute of the write target file (getattr), and acquires information on the newly created file. The PFS control means 51 searches for the latest generation file and returns the information to the user process 10 (steps S1301 to S1303 in FIG. 13).

  1-4. The user process 10 requests opening of the file (open) in the open function (FIG. 6). The PFS control means 51 creates an entry in the PFS link information table 52. In the entry, a file name on PFS, a generation protection flag, a write flag, an actual file name, and an open count are registered (steps S1401 to S1406 in FIG. 14). If it has already been opened by another process, an entry is acquired, the write flag is turned ON, and the open count is incremented (step S1401 is yes, S1407 to S1409, S1406). When the writing has been opened by another process (step S1408 is yes), an abnormality is notified.

  2. The user process 10 executes a write function.

  2-1. In the write function, the user process 10 requests the writing of the opened file by designating an offset and a buffer (FIG. 9). The PFS control means 51 acquires the real file name from the corresponding entry on the PFS link information table 52, acquires data from the specified buffer, and writes it from the specified offset (steps S1701, S1702, FIG. 17). S1708).

  3. The user process 10 repeats step 2 until all data is written, and after completion, executes the close function.

  3-1. The user process 10 requests release of the file in the close function (FIG. 7). The PFS control means 51 decrements the open count of the entry corresponding to the file on the PFS link information table 52 (steps S1801 and S1802 in FIG. 18), and if 0, deletes the entry (yes in step S1803, S1804). ). If there is still an open process (step S1803 is no, S1805 is yes), the write flag is turned off and the generation protection flag is turned on (step S1806).

● Update and append to existing files

  1. The user process 10 executes the open function.

  1-1. The user process 10 requests (getattr) to acquire the attribute of the write target file in the open function. The PFS control means 51 searches for the latest generation file and returns the information to the user process 10 (steps S1301 to S1303 in FIG. 13).

  1-2. The user process 10 requests opening of the file (open) in the open function (FIG. 6). The PFS control means 51 creates an entry in the PFS link information table 52. In the entry, a file name on PFS, a generation protection flag, a write flag, an actual file name, and an open count are registered (steps S1401 to S1406 in FIG. 14). If it has already been opened by another process, an entry is acquired, the write flag is turned ON, and the open count is incremented (step S1401 is yes, S1407 to S1409, S1406). When the writing has been opened by another process (step S1408 is yes), an abnormality is notified.

  2. The user process 10 executes a write function.

  2-1. In the write function, the user process 10 requests the writing of the opened file by designating an offset and a buffer (FIG. 9). The PFS control means 51 acquires the actual file name from the corresponding entry on the PFS link information table 52, acquires data from the specified buffer, and writes it from the specified offset (Steps S1701 and S1702 in FIG. no, S1708).

  3. The user process 10 repeats step 2 until all data is written, and after completion, executes the close function.

  3-1. The user process 10 requests release of the file in the close function (FIG. 7). The PFS control means 51 decrements the open count of the entry corresponding to the file on the PFS link information table 52 (steps S1801 and S1802 in FIG. 18), and if 0, deletes the entry (yes in step S1803, S1804). ). If there is still an open process (step S1803 is no), the write flag is turned off and the generation protection flag is turned on (step S1805 is yes, S1806).

● Reading files

  1. The user process 10 executes the open function.

  1-1. The user process 10 requests (getattr) the attribute acquisition of the read target file in the open function. The PFS control unit 51 searches for the latest generation file (step S1301 in FIG. 13) and returns the information to the user process 10 (yes in step S1302, S1303).

  1-2. The user process 10 requests opening of the file (open) in the open function (FIG. 6). The PFS control means 51 creates an entry in the PFS link information table 52. In the entry, a file name on PFS, a generation protection flag, a write flag, an actual file name, and an open count are registered (steps S1401 to S1406 in FIG. 14). If it has already been opened by another process (step S1401 is yes, S1407 is no), only entry acquisition and open count increment are performed (step S1406). When the writing has been opened by another process (step S1408 is yes), an abnormality is notified.

  2. The user process 10 executes a read function.

  2-1. In the read function, the user process 10 requests reading of the opened file by specifying an offset and a buffer (FIG. 8). The PFS control means 51 acquires the actual file name from the corresponding entry on the PFS link information table 52, reads and returns the data of the specified size from the specified offset (steps S1601 and S1602 in FIG. 16).

  3. The user process 10 repeats step 2 until all data is read, and after completion, executes the close function.

  3-1. The user process 10 requests release of the file in the close function (FIG. 7). The PFS control means 51 decrements the open count of the entry corresponding to the file on the PFS link information table 52 (steps S1801 and S1802 in FIG. 18). If it is 0, the entry is deleted (step S1803 is yes, S1804).

● Rename

  1. The user process 10 executes the rename function.

  1-1. The user process 10 requests (getattr) the attribute acquisition of the new file name in the rename function. The PFS control means 51 searches for the latest generation file and returns the information to the user process 10 (steps S1301 to S1303 in FIG. 13).

  1-2. The user process 10 requests (getattr) the attribute acquisition of the old file name in the rename function. The PFS control means 51 searches for the latest generation file and returns the information to the user process 10 (steps S1301 to 1303 in FIG. 13).

  1-3. In the rename function, the user process 10 requests the file name change by specifying the new and old file names (FIG. 11). The PFS control means 51 renames all old generation files to temporary file names (steps S2301 to S2304 in FIG. 23), renames the latest generation file to a new file name (step S2305), and sets all old generation files. Is renamed to the new file name, and the entry of the PFS control information file 53 is updated (steps S2306 to S2308).

● Directory creation

  1. The user process 10 executes the mkdir function.

  1-1. The user process 10 requests (getattr) the attribute acquisition of the new directory name in the mkdir function. The PFS control means 51 searches for the latest generation file and returns the information to the user process 10 (steps S1301 and S1302 in FIG. 13 are no, S1304).

  1-2. The user process 10 confirms that the directory name does not exist in the mkdir function, and requests creation (mkdir) of the directory. The PFS control means 51 repeats directory creation with the corresponding path names for all the real file systems 41 and 42 in the real file system group 40 (steps S2001 to S2004 in FIG. 20). At this time, if the specified path name exists as a file (steps S2002 and S2005 are yes and no, respectively), all the directories created so far are deleted (step S2006), and an error is returned.

● Directory deletion

  1. The user process 10 executes the rmdir function.

  1-1. The user process 10 makes a request (getattr) to acquire the attribute of the directory name in the rmdir function. The PFS control means 51 searches for the latest generation file and returns the information to the user process 10 (steps S1301 to S1303 in FIG. 13).

  1-2. The user process 10 confirms that the directory name exists in the rmdir function, and requests deletion (rmdir) of the directory. The PFS control means 51 repeats the directory deletion with the corresponding path name for all the real file systems 41 and 42 in the real file system group 40 (step S2101 in FIG. 21) (steps S2102 to S2105). At this time, if the specified path is other than an empty directory (step S2103 is no), an error is returned.

● File deletion

  1. The user process 10 executes the unlink function.

  1-1. The user process 10 requests (getattr) to acquire the attribute of the file to be deleted in the unlink function.

  1-2. If the user process 10 can delete a file in the unlink function, the user process 10 requests (unlink) the deletion of the file (FIG. 10). The PFS control means 51 deletes all previous generation files (steps S1901 to S1905 in FIG. 19), and finally deletes the latest generation file (step S1906). Deletion of an entry in the PFS control information file 53 (step S1907) is performed each time the file is deleted.

● Latest generation search logic (Fig. 24)

  When the latest generation file is searched from the files on the real file systems 41 and 42 corresponding to the file specified by the path on the projection file system, first, the path indicating the file is decomposed into entries (FIG. 24). Step S2401). For example, when the path is / pfs / DIR1 / FILE2, it is decomposed into pfs, DIR1, and FILE2. Thereafter, the search is started from the first entry (step S2402). For example, if / pfs is the mount point of the projection file system, the search is started from / pfs / DIR1.

  Next, the paths corresponding to all the real file systems 41 and 42 are checked (step S2403). For example, if the names of the real file systems 41 and 42 are FS1 and FS2, respectively, and the mount point is / mnt, the search is started from / mnt / FS1 / DIR1 and / mnt / FS2 / DIR1.

  If the corresponding entry is not found (yes in step S2404), the designated path does not exist in any of the real file systems 41 and 42, so that the process ends abnormally. In addition, for example, when / mnt / * / DIR1 is a file in the real file system 41 and a directory in the real file system 42, a directory and a file with the same name are mixed (step S2405 is yes). And Also, when / mnt / FS1 / DIR1 exists as a file but a file is found in the previous loop (step S2406 is yes), as in / mnt / FS1 / DIR1 / FILE2 is specified. , Abnormal termination.

  If all the determination results in steps S2404 to S2406 are no, it is determined whether all entries have been examined (step S2407). If not checked (step S2407 is no), the next entry is checked (step S2403). Note that / mnt / * / DIR1 / FILE2 is checked after / mnt / * / DIR1. On the other hand, if all entries have been checked (yes in step S2407), the PFS control information file 53 is referred to, and for each entry found last, the registration date and time of the file corresponding to that entry (must be For example, the creation date of the file itself is obtained, and the file with the newest registration date is selected (step S2408). For example, if the registration date and time of the corresponding file FILE2 in the real file system 41 is 2008/10 / xx and the registration date and time of the corresponding file FILE2 in the real file system 42 is 2008/09 / yy, File FILE2 will be selected.

● Next generation file creation logic (Figure 25)

  For all the real file systems 41 and 42 (step S2501 in FIG. 25), the presence date and the presence of the target file are acquired from the PFS control information file 53 (steps S2502 and S2503). If the number of existing generations is less than the user-defined number (step S2504 is no), the real file system of the next-generation file creation destination is determined based on the user-defined policy from the real file system in which the target file does not exist. (Step S2506). If the number of existing generations is greater than or equal to the user-defined number (step S2504 is yes), add the real file system where the target file does not exist and the real file system where the oldest generation exists to the user-defined policy. Based on this, the real file system of the next file generation destination is determined (step S2505). As the user-defined policy, a file system priority with a low load, a file system priority with a large free space, and the like can be considered.

  According to the present embodiment, it is possible to obtain an effect that it is possible to prevent the load from being concentrated on a specific real file system. The reason is that the PFS control information file 53 in which the registration date and time (creation date and time) of each file on the real file system is registered is determined, and the file to be processed is determined based on the registration date and time registered here. He who is Further, according to the present embodiment, it is possible to obtain an effect that file generation management can be performed. The reason is that when writing data to a file, a new generation file is created and the data is written to this new generation file.

DESCRIPTION OF SYMBOLS 1 ... Projection file system management apparatus 2 ... User process 3 ... Projection file system control means
3-1 ... Request acceptance means
3-2… Processing file decision means
3-3 ... Request processing means
3-4 ... Next generation file creation means 4 ... Priority table 5 ... Real file system group
5-1 to 5-N ... Real file system
10 ... User process
20 ... Virtual File System (VFS)
30 ... User space file system (FUSE)
40 ... Real file systems
41, 42 ... Real file system
50 ... Projection file system mechanism
51 ... PFS control means
52 ... PFS link information table
53 ... PFS control information file
100 ... Operating system

Claims (11)

Multiple real file systems,
A priority table in which priority information indicating the priority of each file managed by the plurality of real file systems is registered;
The plurality of real file systems are managed as one virtual projection file system in which their directory structures are superimposed, and responds to a processing request specifying a file on the projection file system input from a user process. The file having the highest priority among the files on the real file system corresponding to the specified file is obtained by referring to the priority table, and the requested file is requested by the processing request. a projection shadow file system management apparatus that includes a projection file system control means for performing an operation was,
And with the processing request, the projection File exists the specified copy source specification data write request on the system to be the copy source,
The projection file system control means responds to the copy source designation data write request input from the user process, and among the files on the real file system corresponding to the file designated by the copy source designation data write request. A file having the highest priority is obtained by referring to the priority table, and the obtained file is one of the real file systems in which no file corresponding to the file designated by the copy source designated data write request exists. The next generation file is created by copying to the next generation file, the operation requested by the copy source designation data write request is performed on the created next generation file, and the next generation file is stored in the priority table. Projection file system management characterized by registering priority information Location. The projection file system management apparatus according to claim 1,
As the processing request, a data read request designating the file on the projection file system to read destination data, and the data write request designating the file on the projection file system to write destination data A projection file system management apparatus characterized by existing. In the projection file system management apparatus according to claim 1 or 2,
The file priority information is information relating to the creation time of a file, and a file having a higher creation priority has a higher priority. The projection file system management apparatus according to claim 3 , wherein
The projection file system control means, when there are a plurality of real file systems in which a file corresponding to the file designated by the copy source designation data write request does not exist in the plurality of real file systems, A projection file system management apparatus, wherein an actual file system having the largest available capacity or an actual file system having the lowest load is selected from the actual file systems. In the projection file system management apparatus according to claim 4 ,
If the file corresponding to the file specified by the copy source specifying data write request exists in all of the plurality of real file systems, the projection file system control means is the oldest generation of the corresponding files. A projection file system management apparatus that selects an actual file system in which a file exists. A computer comprising a plurality of real file systems and a priority table in which priority information indicating the priority of each file managed by the plurality of real file systems is registered.
A first step of managing the plurality of real file systems as one virtual projection file system in which their directory structures are superimposed;
In response to the processing request specifying the file on the projection file system input from the user process, the computer has the highest priority among the files on the real file system corresponding to the specified file. seeking file by referring to the priority table, and a second step of performing an operation requested by the processing request to the determined file a including projecting the shadow file system management method,
The processing by the request, there is the copy source and the projection copy source specification data write request designating the file on the file system, and,
In response to the copy source designation data write request input from the user process, the computer has the highest priority among the files on the real file system corresponding to the file designated by the copy source designation data write request. A file having a high file size is obtained by referring to the priority table, and the obtained file is copied to one of the real file systems in which no file corresponding to the file designated by the copy source designated data write request exists. To create a next-generation file, perform the operation requested by the copy-source-designated data write request for the created next-generation file, and store priority information on the next-generation file in the priority table. A projection file system comprising a third step of registering Management method. The projection file system management method according to claim 6 ,
As the processing request, a data read request designating the file on the projection file system to read destination data, and the data write request designating the file on the projection file system to write destination data A projection file system management method characterized by existing. The projection file system management method according to claim 6 or 7 ,
The file priority information is information relating to the creation time of a file, and a file with a newer creation time has a higher priority. A computer having a plurality of real file systems and a priority table in which priority information indicating priority levels of files managed by the plurality of real file systems is registered is caused to function as a projection file system management apparatus. A program for
The computer constitutes one virtual projection file system in which the directory structures are overlapped by the plurality of real file systems, and a processing request specifying a file on the projection file system input from a user process In response, the file having the highest priority among the files on the real file system corresponding to the designated file is obtained by referring to the priority table, and the processing request for the obtained file is obtained. a program to function as the projection file system control means for performing the requested operation by,
And with the processing request, the projection File exists the specified copy source specification data write request on the system to be the copy source,
The projection file system control means responds to the copy source designation data write request input from the user process, and among the files on the real file system corresponding to the file designated by the copy source designation data write request. A file having the highest priority is obtained by referring to the priority table, and the obtained file is one of the real file systems in which no file corresponding to the file designated by the copy source designated data write request exists. The next generation file is created by copying to the next generation file, the operation requested by the copy source designation data write request is performed on the created next generation file, and the next generation file is stored in the priority table. A program characterized by registering priority information. The program according to claim 9 ,
As the processing request, a data read request designating the file on the projection file system to read destination data, and the data write request designating the file on the projection file system to write destination data A program characterized by the existence. The program according to claim 9 or 10 ,
The file priority information is information relating to the creation time of a file, and a file having a higher creation priority has a higher priority.

Images