JP5497861B2 - Server, file management system, file management method, and file management program - Google Patents

Description

  The present invention relates to a file management system and the like.

  It is important to share system resources and perform exclusive access control, and to ensure system availability and consistency in the event of a failure. There are various conventional techniques for solving this problem.

  For example, there is a conventional technology that supports long-time transactions in a low-reliability network computer cluster environment using inexpensive hardware (Patent Document 2, etc.). This prior art realizes fault tolerance against a server failure and fault tolerance against a client terminal failure due to a transaction by replicating data in preparation for a master change.

  There are conventional technologies that deal with network failures and server failures (Non-Patent Documents 1 and 2, etc.). This prior art can cope with any of a network failure, a server failure, and a client failure by performing an operation to lock a file.

  When a plurality of processor elements execute processing in parallel in one computer, there is a technique for preparing a safe counter and limiting the number of resources that the plurality of processor elements can acquire simultaneously to a certain number or less ( Non-patent document 3).

  In addition, in a conventional client server system or the like, when a client accesses a master resource, a shared lock or an exclusive lock may be performed on the resource. Managed.

JP 2011-210106 A JP2011-210107A

Mike Burrows, "The Chubby lock service for loosely-coupled distributed systems", 7th USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2006. November 8 Tushar Chandra et al., "Paxos Made Live-An Engineering Perspective," PODC'07, 2007. Dijkstra, Edsger W. Cooperating sequential processes (EWD-123). E.W.Dijkstra Archive. Center for American History, University of Texas at Austin. (Original; transcription) (September 1965)

  However, in a system in which a single master is selected from a plurality of servers included in the same network when the network is distributed among a plurality of servers to be synchronized, the lock information of the client with respect to resources due to the failure of the master is displayed. Although another server takes over, there is a problem that inconsistency occurs in the lock information. In the above-described conventional technology, such a problem cannot be solved.

  The embodiment of the disclosure is made in view of the above, and when a network is distributed among a plurality of servers to be synchronized, a single master is selected from a plurality of servers included in the same network. Server, file management system, file management method, and file management program capable of preventing inconsistency in lock information when another server takes over the lock information of a client for a resource due to a master failure The purpose is to provide.

  The server according to the present invention includes a master adjustment unit, a request processing unit, a replication unit, and a restoration processing unit. The master adjustment unit selects a single server having a master role from a plurality of other servers and its own server included in the same network. The request processing unit issues a session ID to the client in response to a connection request from the client, and registers the session ID in the session table in response to a connection request from the client when the server is a server having a master role. When the lock information for the file and the session ID are registered in the file management table in association with the file processing request, if the number of locks for the file is less than the threshold, the lock information for the file and the session ID are If the number of locks on the file has reached the threshold, the lock information for the file and the session ID are registered in association with the lock waiting session, and the number of locks on the file is When it becomes less than the threshold, In association with the lock information and the session ID for the registered file in Deployment is registered in the file management table. Further, the request processing unit issues a session ID to send information to the connected client for monitoring, and when a response from the client does not arrive, lock information corresponding to the session ID issued to the client, Delete from the file management table. The replication unit replicates the session table and file management table of the server having the master role when the local server is a server having the role of replica. When the local server switches from the server having the role of replica to the server having the role of master, the restoration processing unit registers the information of the session table in the session table waiting for deletion and responds to the reconnection request from the client. Then, the same session ID as the session ID included in the reconnection request is deleted from the deletion waiting session table, and the lock information corresponding to the session ID remaining in the deletion waiting session table is deleted from the file management table. .

  According to the server according to the present invention, in a system in which a single master is selected from a plurality of servers included in the same network when a network is distributed among a plurality of servers to be synchronized, When another server takes over the lock information of the client for the resource, it is possible to prevent the occurrence of contradiction in the lock information.

FIG. 1 is a diagram showing a configuration of a file management system according to the present embodiment. FIG. 2 is a functional block diagram showing the configuration of the server according to the present embodiment. FIG. 3 is a diagram illustrating an example of the data structure of the session table according to the present embodiment. FIG. 4 is a diagram showing an example of the data structure of the deletion waiting session table according to the present embodiment. FIG. 5 is a diagram showing an example of the data structure of the file management table according to the present embodiment. FIG. 6 is a functional block diagram showing the configuration of the client according to the present embodiment. FIG. 7 is a flowchart showing a processing procedure of the server according to the present embodiment. FIG. 8 is a flowchart illustrating a processing procedure of processing for restoring a session. FIG. 9 is a flowchart showing a processing procedure of session deletion processing. FIG. 10 is a flowchart showing a processing procedure of request acceptance processing. FIG. 11 is a flowchart illustrating a processing procedure of client monitoring processing. FIG. 12 is a flowchart showing the procedure of the blocking process. FIG. 13 is a flowchart showing a processing procedure when an exclusive lock acquisition request is received. FIG. 14 is a flowchart illustrating a processing procedure when an exclusive lock release request is received. FIG. 15 is a flowchart illustrating a processing procedure when a shared lock acquisition request is received. FIG. 16 is a flowchart illustrating a processing procedure when a shared lock release request is received. FIG. 17 is a flowchart showing the processing procedure of the client according to the present embodiment. FIG. 18 is a diagram illustrating an example of a computer that executes a control program that implements the same function as the server.

  Hereinafter, embodiments of a server, a file management system, a file management method, and a file management program disclosed in the present application will be described in detail with reference to the drawings. In addition, this invention is not limited by this embodiment.

(Embodiment)
A configuration of the file management system according to the present embodiment will be described. FIG. 1 is a diagram showing a configuration of a file management system according to the present embodiment. As shown in FIG. 1, the file management system includes servers 100a, 100b, 100c, 100d, and 100e, clients 200a and 200b, and an application 60. The servers 100a to 100e and the clients 200a and 200b are connected to the network 50.

  In FIG. 1, the servers 100a to 100e and the clients 200a and 200b are shown as an example, but this system may include other servers and clients. In the following description, the servers 100a to 100e are collectively referred to as the server 100 as appropriate. The clients 200a and 200b are collectively referred to as the client 200 as appropriate.

  Each server 100 selects a single master from the servers 100 included in the same network using a majority protocol such as Paxos. As a technique for selecting a single master using the majority protocol, for example, the techniques of Patent Documents 1 and 2 may be used. The server elected as the master functions as a server having the master function. On the other hand, a server that is not elected as a master functions as a server having a role of a replica. In the following description, a server having a master role is referred to as a master, and a server having a replica role is appropriately referred to as a replica.

  For example, in FIG. 1, it is assumed that the servers 100a to 100e are included in the same network and the server 100a is selected as a master. In this case, the server 100a functions as a server having a master role, and the servers 100b to 100e function as servers having a replica role.

  The server 100 having the master role establishes a connection in response to a connection request from the client 200. The server 100 having the master role locks the file in response to the file processing request from the client 200. The lock applied to the file is a shared lock or an exclusive lock. The server 100 having the master role stores the lock information for the file, and notifies the server 100 having the replica role of the lock information for replication.

  When the server 100 having the role of replica switches to the server 100 having the role of master, a reconnection request is accepted from the client 200. The server 100 having the master role accepts a reconnection request for the copied lock information, leaves the lock information corresponding to the client, and deletes other lock information. By executing such processing, it is possible to prevent inconsistencies in the lock information.

  After establishing a connection with the server 100 having the master role, the client 200 executes processing on the file in accordance with an instruction from the application 60. For example, the client 200 applies a shared lock or an exclusive lock to the file, and executes processing on the file.

  In addition, when the server 100 having the master role is switched, the client 200 makes a reconnection request to the server 100 having the master role, and continues the processing on the file. Note that if the reconnection fails, the client 200 notifies the application 60 that the reconnection has failed and stops processing the file.

  Next, an example of the configuration of the server 100 illustrated in FIG. 1 will be described. FIG. 2 is a functional block diagram showing the configuration of the server according to the present embodiment. As illustrated in FIG. 2, the server 100 includes a communication unit 110, a storage unit 140, and a control unit 150.

  The communication unit 110 is a communication device that performs data communication with other devices connected to the network 50. For example, the communication unit 110 performs data communication with the client 200 and another server 100. A control unit 150 (described later) uses the communication unit 110 to exchange data with the client 200 and other servers 200.

  The storage unit 140 is a storage device that stores a session table 140a, a deletion waiting session table 140b, and a file management table 140c.

  The session table 140a is a table having information related to a session between the server 100 having a master role and the client 200. FIG. 3 is a diagram illustrating an example of the data structure of the session table according to the present embodiment. As shown in FIG. 3, the session table 140a associates a session ID, client information, and a lock file entry. Among these, the session ID (Identifier) is information for identifying a session established between the server 100 having a master role and the client 200. The client information is information for uniquely identifying the client 200, and corresponds to, for example, the IP (Internet Protocol) address of the client 200. The entry of the lock file is information indicating a file ID (Fid) that uniquely identifies the locked file and the type of lock. The lock type includes a shared lock and an exclusive lock.

  For example, referring to the second row in FIG. 3, the server 100 having the role of master and the client 200 are connected by the session ID “2”, and the client information of the client 200 is “192.168.”. .10.15 ". Further, the client 200 locks the file with the file ID “1”.

  The deletion-waiting session table 140b is a table having information on a session that could not be restored when the server 100 having the master role is switched. FIG. 4 is a diagram showing an example of the data structure of the deletion waiting session table according to the present embodiment. As shown in FIG. 4, this deletion waiting session table 140b associates session IDs, client information, and lock file entries. The description of the session ID, client information, and lock file entry is the same as that of the session table 140a shown in FIG.

  For example, taking the second row in FIG. 4 as an example, since the session identified by the session ID “2” could not be restored, the connection with the client 200 having the client information “192.168.1.15” is established. However, the client 200 is in a state in which a shared lock is applied to the file with the file ID “1”. This shared lock is released by the restoration processing unit 150d described later.

  The file management table 140c is a table having lock information for files. FIG. 5 is a diagram showing an example of the data structure of the file management table according to the present embodiment. As shown in FIG. 5, this file management table associates a file ID, the upper limit number of locks, a locked session, and a lock waiting session.

  The file ID (Fid) is information for uniquely identifying a file. The upper limit number of locks is the maximum number of shared locks that can be locked simultaneously for a single file. Note that the number of exclusive locks that can be locked simultaneously for a single file is one regardless of the upper limit number of locks.

  The in-lock session indicates a session ID that locks the file and the type of lock. The lock waiting session indicates the session ID and lock type for which the corresponding file is waiting to be locked. When there is a vacant session in the locked session, the lock waiting session shifts to the locked session in the registered order.

  For example, taking the first row in FIG. 5 as an example, the upper limit number of files with file ID “1” is “2”, and two shared locks are applied to the file with file ID “1”. Yes. Such locks are a shared lock with a session ID “2” and a shared lock with a session ID “3”. Further, in the lock waiting session, lock waiting occurs in the order of the exclusive lock by the session ID “4” and the shared lock by the session ID “28”.

  Returning to the description of FIG. The control unit 150 includes a master adjustment unit 150a, a replication unit 150b, a request processing unit 150c, and a restoration processing unit 150d.

  The master adjustment unit 150a is a processing unit that selects a single master from the servers 100 included in the same network using a majority protocol such as Paxos. The master adjustment unit 150a outputs information on whether or not the server 100 is selected as a master to the replication unit 150b, the request processing unit 150c, and the restoration processing unit 150d.

  The master adjustment unit 150a sets a next master selection timer (not shown) when the master is selected, and selects the master again when the next master selection timer times out. That is, the master adjustment unit 150a repeatedly executes a process of selecting a master at a predetermined time interval using a next master selection timer.

  The replication unit 150 b is a processing unit that replicates information stored in the storage unit 140. When the local server 100 is a master, a replication request is made by notifying the server 100 having the role of a replica of the session table 140a and the file management table 140c in the storage unit 140.

  On the other hand, when the local server 100 is a replica, it receives a replication request from the server 100 having the master role. Then, the own server 100 duplicates the session table 140 a and the file management table 140 c of the server 100 having the role of the master and stores it in the storage unit 140.

  The replication unit 150b sets a mastery timer (not shown) when the master is selected by the master adjustment unit 150a. The replication unit 150b repeatedly executes the above process until the mastery timer times out.

  In addition, when the local server 100 is the server 100 having the role of the master and the session table 140a and the file management table 140c are updated, the replication unit 150b transmits the session table 140a and the file management table 140c to the replica. A replication request is made by notifying the server 100 having the above role.

  The request processing unit 150c is a processing unit that operates when the server 100 is elected as a master, establishes a session with the client 200, and performs processing in response to a file processing request. First, in response to a connection request from the client 200, the request processing unit 150c issues a unique session ID to the client 200, associates the issued session ID with the IP address of the client 200, and stores it in the session table 140a. sign up.

  The request processing unit 150c issues a session ID and establishes a session with the client 200, and then updates the session table 140a and the file management table 140c in response to a file processing request from the client 200. For example, the file processing request received from the client 200 includes any of an exclusive lock acquisition request, an exclusive lock release request, a shared lock acquisition request, and a shared lock release request. Hereinafter, processing of the request processing unit 150c when an exclusive lock acquisition request, an exclusive lock release request, a shared lock acquisition request, and a shared lock release request are received will be described in order.

  Processing when the request processing unit 150c acquires an “exclusive lock acquisition request” will be described. This exclusive lock acquisition request includes the file ID to be processed. The request processing unit 150c registers the lock file entry in the session table 140a in association with the session ID. For example, when the session ID is “4” and an exclusive lock for the file ID “1” is requested, the request processing unit 150c determines that the entry “Fid = 1 (exclusive) in the fourth-stage lock file in FIG. Is registered in the session table 140a.

  The request processing unit 150c refers to the file management table 140c and determines whether there is a locked session corresponding to the file ID. When there is no locked session, the request processing unit 150c registers the session ID and the lock type (exclusive lock) in association with the corresponding locked session. On the other hand, when there is a locked session, the request processing unit 150c associates the session ID with the lock type (exclusive lock) and registers it in the lock waiting session.

  Next, processing when the request processing unit 150c acquires an “exclusive lock release request” will be described. This exclusive lock release request includes the file ID to be processed. The request processing unit 150c deletes the lock file entry corresponding to the session ID from the session table 140a. For example, when the session ID is “4” and the release of the exclusive lock for the file ID “1” is requested, the request processing unit 150c sets the entry “Fid = 1” in the lock file in the fourth row in FIG. (Exclusive) "is deleted from the session table 140a.

  The request processing unit 150c refers to the file management table 140c and deletes the value from the locked session corresponding to the file ID. For example, when the exclusive lock is performed on the file ID “1” with the session ID “4”, the value “Sid = 4 (exclusive)” from the locked session of the file ID “1” is set to the file management. Delete from table 140c.

  After deleting the value from the file management table 140c, the request processing unit 150c extracts the top entry of the lock waiting session for the corresponding file ID and adds it to the value of the locked session. When the lock type of the value to be added is shared lock, the request processing unit 150c repeatedly executes a process of adding a value to the locked session within a range not exceeding the lock upper limit number.

  Next, processing when the request processing unit 150c acquires a “shared lock acquisition request” will be described. This shared lock acquisition request includes the file ID to be processed. The request processing unit 150c registers the lock file entry in the session table in association with the session ID. For example, when the session ID is “2” and the sharing lock for the file ID “1” is requested, the request processing unit 150c performs the entry “Fid = 1 (shared) in the second row in FIG. Is registered in the session table 140a.

  The request processing unit 150c refers to the locked session corresponding to the file ID in the file management table 140c. When the number of shared locks in the locked session is less than the upper limit number of locks or when an exclusive lock is not applied, the request processing unit 150c sets the session ID and the lock type (shared) to the corresponding locked session. Lock) and register.

  In contrast, in the locked session, the request processing unit 150c determines the session ID and the lock type (shared lock) when the number of shared locks is equal to or greater than the upper limit number of locks or when an exclusive lock is applied. Are registered in the lock waiting session.

  Next, processing when the request processing unit 150c acquires a “shared lock release request” will be described. This shared lock release request includes the file ID to be processed. The request processing unit 150c deletes the lock file entry corresponding to the session ID from the session table 140a. For example, when the session ID is “2” and the request to release the shared lock for the file ID “1” is requested, the request processing unit 150c has the entry “Fid = 1” in the second-stage lock file in FIG. (Shared) ”is deleted from the session table 140a.

  The request processing unit 150c refers to the file management table 140c and deletes the value from the locked session corresponding to the file ID. For example, if the session ID “2” and the file ID “1” are exclusively locked, the value “Sid = 2 (shared)” from the locked session of the file ID “1” is set to the file management. Delete from table 140c.

  After deleting the value from the file management table 140c, the request processing unit 150c extracts the top entry of the lock waiting session for the corresponding file ID and adds it to the value of the locked session. When the lock type of the value to be added is shared lock, the request processing unit 150c repeatedly executes a process of adding a value to the locked session within a range not exceeding the lock upper limit number.

  As described above, after the request processing unit 150c receives a connection request from the client 200 and establishes a session, the file processing request from the client 200 includes an exclusive lock acquisition request, an exclusive lock release request, a shared lock acquisition request, In response to the shared lock release request, the session table 140a and the file management table 140c are updated.

  Further, the request processing unit 150c monitors the client 200 by periodically sending a ping to the client 200 that has established a session. When the request processing unit 150c does not accept a response from the client 200 within a predetermined time after transmitting the ping, the request processing unit 150c releases the locked file with the session ID corresponding to the client 200. For example, when the request processing unit 150c does not accept a ping response from the client 200 that has established a session with the session ID “2”, the value “Sid = 2 (shared) of the locked session in the file management table of FIG. "Is deleted. The process of deleting the locked session and adding the value of the lock waiting session to the locked session is the same as the process when the exclusive lock release request and the shared lock release request are acquired.

  The restoration processing unit 150d is a processing unit that operates when the server 100 switches from a server having a replica role to a server having a master role, and restores a session with the client 200. First, the restoration processing unit 150d writes all information registered in the session table 140a to the deletion waiting session table 140b, and sets a session deletion timer (not shown). In addition, the restoration processing unit 150d deletes all information in the session table 140a.

  The restoration processing unit 150d receives a reconnection request from the client 200 before the session deletion timer times out. This reconnection request includes the session ID issued from this server before the server having the master role is switched. The restoration processing unit 150d compares the session ID included in the reconnection request with the deletion waiting session table 140b, and records a record having the same session ID as the session ID included in the reconnection request from the deletion waiting session table 140b. delete. In addition, the restoration processing unit 150d registers the session ID included in the reconnection request in the session table 140a.

  When the session deletion timer times out, the restoration processing unit 150d deletes the corresponding value from the locked session of the file management table 140c based on the session ID and the lock file entry included in the deletion waiting session table 140b. To do.

  When the lock type indicated in the entry of the lock file is an exclusive lock, the restoration processing unit 150d executes the same process as the process when the “exclusive lock release request” described above is acquired. In addition, when the lock type indicated in the lock file entry is a shared lock, the restoration processing unit 150d executes the same process as the process performed when the above-described “shared lock release request” is acquired.

  Next, the configuration of the client 200 shown in FIG. 1 will be described. FIG. 6 is a functional block diagram showing the configuration of the client according to the present embodiment. As illustrated in FIG. 6, the client 200 includes a communication unit 210, an input unit 220, a display unit 230, a storage unit 240, and a control unit 250.

  The communication unit 210 is a communication device that performs data communication with other devices connected to the network 50. For example, the communication unit 210 performs data communication with the server 100 and the application 60. The control unit 250 described later exchanges data with the server 200 and the application 60 using the communication unit 210.

  The input unit 220 is an input device that inputs various data to the client 200. For example, the input unit 220 corresponds to a keyboard, a mouse, a touch panel, or the like. The display unit 230 is a display device that displays various data. For example, the display unit 230 corresponds to a display or the like.

  The storage unit 240 is a storage device that stores session information 240a. The session information 240a is information having a session ID when the client 200 establishes a session with the server 100 having the master role, for example.

  The control unit 250 includes a connection request unit 250a, a request unit 250b, a reconnection request unit 250c, and a notification unit 250d.

  The connection request unit 250a is a processing unit that establishes a session with the server 100 having a master role. Specifically, the connection request unit 250 a transmits connection requests to all the servers 100 connected to the network 50. After transmitting the connection request, the connection request unit 250a receives the session ID from the server 100 having the role of the master, and registers the received session ID in the session information 240a. Then, the connection request unit 250a transmits a reception confirmation to the server 100 having the master role. A session between the client 200 and the server 100 is established by the connection request unit 250a executing the series of processes described above.

  The connection request unit 250a sets a life / death monitoring timer (not shown), and when receiving communication from the server 100 having the role of the master before the life / death monitoring timer times out, the connection requesting unit 250a confirms reception to the master. Send and repeat the process of resetting the alive monitoring timer. Note that the connection request unit 250a requests the reconnection request unit 250c to reconnect before receiving a communication from the server 100 having the master role before the alive monitoring timer times out.

  The request unit 250b is a processing unit that makes a file processing request to the master while the client 200 establishes a session with the server 100 having the role of the master. For example, the request unit 250b makes an exclusive lock acquisition request, an exclusive lock release request, a shared lock acquisition request, a shared lock release request, and the like as a file processing request. The request unit 250b may receive an instruction from the application 60 and make a file processing request according to the content of the instruction.

  The reconnection request unit 250 c is a processing unit that makes a reconnection request to the server 100. Specifically, the reconnection request unit 250c sets a reconnection timer (not shown) when a reconnection request is received from the connection request unit 250a. Then, the reconnection request unit 250c registers the session ID registered in the session information 240a in the reconnection request, and transmits the reconnection request to all the servers 100 connected to the network.

  When a reply is obtained from the server 100 having the master role before the reconnection timer times out, the reconnection request unit 250c means that the session has been established again, and the subsequent processing is performed by the connection request unit 250a. To ask.

  Upon receipt of the request, the connection request unit 250a sets a life / death monitoring timer, and when communication from the server 100 having the role of the master is received before the life / death monitoring timer times out, the reception request to the master is confirmed. Send and repeat the process of resetting the alive monitoring timer. Note that the connection request unit 250a requests the reconnection request unit 250c to reconnect before receiving a communication from the server 100 having the master role before the alive monitoring timer times out.

  On the other hand, when the reconnection request unit 250c cannot obtain a reply to the reconnection request from the server 100 having the master role before the reconnection timer times out, the reconnection request unit 250c obtains a reply to the reconnection request. No notification is output to the notification unit 250d.

  The notification unit 250d is a processing unit that notifies the application 60 of information indicating that reconnection has failed when the reconnection request unit 250c fails to reconnect. When the reconnection fails, for example, the application 60 limits the file processing request to the master from the request unit 250b.

  Next, a processing procedure of the server 100 according to the present embodiment will be described. FIG. 7 is a flowchart showing a processing procedure of the server according to the present embodiment. As shown in FIG. 7, the server 100 selects a master by Paxos (step S101), and determines whether or not the server 100 is selected as a master (step S102).

  If the server 100 is not elected as a master (No at Step S102), the server 100 sets a master lease timer (Step S103). The server 100 starts accepting a replication request from the master (step S104).

  The server 100 determines whether or not the master lease timer has timed out (step S105). If the master lease timer has not timed out (No at Step S105), the server 100 proceeds to Step S105 again.

  On the other hand, when the master lease timer times out (step S105, Yes), the server 100 closes the replication request reception from the master (step S106), and proceeds to step S101.

  By the way, when the server 100 is elected as a master (step S102, Yes), the server 100 sets a master lease timer (step S107). The server 100 sets the next master selection timer (step S108).

  If the server 100 is the previous master and the previous master lease timer has not timed out (step S109, Yes), the server 100 proceeds to step S111. On the other hand, if the server 100 is not the previous master, or if the previous master lease timer has timed out (No in step S109), the server 100 proceeds to step S110.

  The server 100 restores the session with the client 200 (step S110), and proceeds to step S111.

  The server 100 starts accepting a request from the client 200 (step S111), and determines whether or not the next master selection timer has timed out (step S112). If the next master calculation timer has not timed out (No at Step S112), the server 100 proceeds to Step S112 again.

  If the next master selection timer times out (step S112, Yes), the server 100 proceeds to step S101 and continues the process. On the other hand, the server 100 determines whether or not the master lease has timed out (step S113). If the master lease timer has not timed out (step S113), the server 100 proceeds to step S113 again. On the other hand, when the master lease timer times out (step S113, Yes), the server 100 blocks the request reception from the client (step S114).

  Next, the process for restoring the session with the client shown in step S110 of FIG. 7 will be described. FIG. 8 is a flowchart illustrating a processing procedure of processing for restoring a session. As shown in FIG. 8, the server 100 reads the session information before the master change in the session table 140a, and writes all the information in the deletion waiting session table 140b (step S121). The server 100 deletes all information in the session table 140a (step S122).

  The server 100 sets a session deletion timer (step S123), and determines whether or not the session deletion timer has timed out (step S124). If the session deletion timer has not timed out (No at Step S124), the server 100 proceeds to Step S124 again.

  If the session deletion timer times out (step S124, Yes), the server 100 determines whether or not the deletion waiting session table 140b is empty (step S125). If the deletion waiting session table 140b is empty (step S125, Yes), the server 100 ends the process of restoring the session.

  On the other hand, when the deletion waiting session table 140b is not empty (step S125, No), the server 100 selects an arbitrary session in the deletion waiting session table 140b (step S126). The server 100 executes session deletion processing (step S127), and proceeds to step S125.

  Next, the session deletion process shown in step S127 of FIG. 8 will be described. FIG. 9 is a flowchart showing a processing procedure of session deletion processing. As shown in FIG. 9, the server 100 acquires all lock file entries related to the session ID from the session table 140a (step S131).

  The server 100 performs an automatic unlocking operation based on the lock file entry (step S132). In step S132, when the file to be unlocked is shared and locked, the server 100 performs the same process as the process for the shared lock release request process. On the other hand, when the file to be unlocked is exclusively locked, the same processing as that for the exclusive lock release request is performed.

  Next, a description will be given of the processing procedure of request reception processing from the start of the request from the client shown in step S111 in FIG. 7 to the blocking of the request reception from the client in step S114. FIG. 10 is a flowchart showing a processing procedure of request acceptance processing.

  As shown in FIG. 10, the server 100 determines whether or not there is a request from the client 200 (step S141). When there is no request from the client 200 (No at Step S141), the server 100 proceeds to Step S141 again.

  If there is a request from the client (step S141, Yes), the server 100 determines whether or not a connection request has been received from the client 200 (step S142). When the server 100 accepts the connection request (step S142, Yes), the server 100 issues a session ID and registers it in the session table 140a (step S143).

  The server 100 notifies the client 200 of the session ID (step S144), starts a client monitoring process (step S145), and proceeds to step S141.

  In step S142, if the server 100 has not received a connection request from the client 200 (step S142, No), the server 100 determines whether or not it is a reconnection request (step S146). When the server 100 receives the reconnection request (step S146, Yes), the server 100 deletes the corresponding session ID from the deletion waiting session table 140b (step S147). The server 100 adds the session ID to the session table 140a (step S148), and proceeds to step S144.

  In step S146, if the server 100 has not received a reconnection request from the client 200 (step S146, No), the server 100 executes processing for the file according to the processing content (step S149). The server 100 transmits the processing result to the replica (step S150).

  The server 100 determines whether or not a reply is obtained from the replica (step S151). When a reply is obtained from the replica (step S151, Yes), the server 100 proceeds to step S141. On the other hand, when a reply cannot be obtained from the replica (step S151, No), the request acceptance from the client 200 is blocked (step S152).

  Next, the client monitoring process when the client monitoring process shown in step S145 of FIG. 10 is started will be described. FIG. 11 is a flowchart illustrating a processing procedure of client monitoring processing.

  As illustrated in FIG. 11, the server 100 transmits a ping to the client 200 after waiting for a predetermined time (step S171). The server 100 sets a timer (step S172), and determines whether or not there is a reply from the client before the timeout (step S173).

  If there is a reply from the client before the timeout (Yes at Step S173), the server 100 proceeds to Step S171. On the other hand, when there is no reply from the client before the timeout (No at Step S173), the server 100 executes a session deletion process (Step S174).

  Next, the blocking process for blocking the request acceptance from the client shown in step S152 in FIG. 10 will be described. FIG. 12 is a flowchart showing the procedure of the blocking process. As shown in FIG. 12, the server 100 stops the process for accepting requests from the client 200 (step S181).

  The server 100 stops all client monitoring processes (step S182) and deletes all session information (step S183).

  Next, processing when the server 100 receives an exclusive lock acquisition request, an exclusive lock release request, a shared lock acquisition request, and a shared lock release request from the client 200 will be described.

  FIG. 13 is a flowchart showing a processing procedure when an exclusive lock acquisition request is received. As shown in FIG. 13, the server 100 acquires the file ID to be locked (step S201), and determines whether or not the value of the locked session corresponding to the file ID is empty (step S202).

  If the value of the locked session corresponding to the file ID is empty (step S202, Yes), the server 100 adds a value to the locked session (step S203).

  On the other hand, if the value of the locked session corresponding to the file ID is not empty (No at Step S202), the server 100 adds a value to the lock waiting session (Step S204).

  FIG. 14 is a flowchart illustrating a processing procedure when an exclusive lock release request is received. As shown in FIG. 14, the server 100 acquires the file ID to be locked from the file management table 140c (step S211). The server 100 deletes the value corresponding to the file ID from the locked session (step S212).

  The server 100 determines whether or not all the lock waiting session values corresponding to the file ID are empty (step S213). If the values of the session waiting for lock corresponding to the file ID are all empty (step S213, Yes), the server 100 ends the process.

  On the other hand, if the values of the session waiting for lock corresponding to the file ID are not all empty (step S213, No), the server 100 extracts the first entry of the session waiting for lock of the file ID, and the value of the session being locked (Step S214).

  The server 100 determines whether or not the value added to the locked session is an exclusive lock (step S215). When the value added to the locked session is an exclusive lock (step S215, Yes), the server 100 ends the process.

  On the other hand, if the value added to the locked session is not an exclusive lock (step S215, No), the server 100 determines whether the lock waiting session is not empty (step S216). If the session waiting for lock is empty (No in step S216), the server 100 ends the process.

  On the other hand, if the lock waiting session is not empty (step S216, Yes), the server 100 determines whether or not the head of the lock waiting session is a shared lock (step S217). When the head of the lock waiting session is not the shared lock (step S217, No), the server 100 ends the process.

  On the other hand, when the head of the session waiting for lock is a shared lock (step S217, Yes), the server 100 determines whether or not the number of locked sessions has reached the upper limit (step S218). If the number of locked sessions has reached the upper limit number (step S218, Yes), the server 100 ends the process.

  On the other hand, if the number of locked sessions has not reached the upper limit (step S218, No), the server 100 extracts the top entry of the lock waiting session, adds it to the locked session (step S219), and step S216. Migrate to

  FIG. 15 is a flowchart illustrating a processing procedure when a shared lock acquisition request is received. As shown in FIG. 15, the server 100 acquires the file ID to be locked (step S221), and determines whether or not the value of the locked session for the file ID is empty (step S222).

  If the value of the locked session for the file ID is empty (step S222, Yes), the server 100 adds a value to the locked session (step S223) and ends the process.

  On the other hand, if the value of the locked session for the file ID is not empty (step S222, No), the server 100 determines whether the value of the locked session is a shared lock (step S224).

  When the value of the locked session is not the shared lock (step S244, No), the server 100 adds the session waiting for the lock (step S225).

  On the other hand, the server 100 determines whether or not the value of the locked session has reached the upper limit number (step S226). When the number of locked sessions has not reached the upper limit number (step S226, Yes), the server 100 proceeds to step S223. If the value of the session being locked has reached the upper limit number (step S226, No), the server 100 proceeds to step S225. On the other hand, if the value of the locked session has not reached the upper limit number (step S226, Yes), the server 100 proceeds to step S223.

  FIG. 16 is a flowchart illustrating a processing procedure when a shared lock release request is received. As illustrated in FIG. 16, the server 100 acquires a file ID to be locked from the file management table 140c (step S231). The server 100 deletes the value corresponding to the file ID from the locked session (step S232).

  The server 100 determines whether or not all the lock waiting session values corresponding to the file ID are empty (step S233). If all the values of the session waiting for lock are empty (step S233, Yes), the server 100 ends the process.

  On the other hand, when all the values of the lock waiting session are not empty (step S233, No), the server 100 determines whether or not the head of the lock waiting session is a shared lock (step S234). When the head of the lock waiting session is not the shared lock (No in step S234), the server 100 ends the process.

  On the other hand, when the head of the session waiting for lock is the shared lock (step S234, Yes), the server 100 determines whether or not the number of locked sessions has reached the upper limit (step S235). If the number of locked sessions has reached the upper limit number (step S235, Yes), the server 100 ends the process.

  On the other hand, if the number of locked sessions has not reached the upper limit (No in step S235), the server 100 takes out the top entry of the lock waiting session and adds it to the locked session (step S236).

  The server 100 determines whether or not the lock waiting session is not empty (step S237). If the lock waiting session is empty (step S237, No), the server 100 ends the process. On the other hand, when the lock waiting session is not empty (step S237, Yes), the server 100 proceeds to step S234.

  Next, a processing procedure of the client 200 according to the present embodiment will be described. FIG. 17 is a flowchart showing the processing procedure of the client according to the present embodiment. As shown in FIG. 17, the client 200 transmits a connection request to all the servers 100 connected to the same network (step S301).

  The client 200 receives the session ID from the master (step S302), and transmits a timer reception confirmation to the master (step S303). The client 200 sets an alive monitoring timer (step S304).

  The client 200 determines whether communication has been received from the master before the alive monitoring timer times out (step S305). If the client 200 receives communication from the master before the alive monitoring timer times out (step S305, Yes), the client 200 proceeds to step S303.

  On the other hand, if the client 200 has not received communication from the master before the alive monitoring timer times out (No in step S305), the client 200 sets a reconnection timer (step S306).

  The client 200 makes a reconnection request to all the servers 100 connected to the same network (step S307). This reconnection request includes a session ID. The client 200 determines whether a reply is obtained from the master before the reconnection timer times out (step S308). If a reply is obtained from the master before the reconnection timer times out (Yes in step S308), the client 200 proceeds to step S303.

  If the reply is not obtained from the master before the reconnection timer times out (step S308, No), the client 200 transmits information indicating that reconnection could not be performed to the application 60 (step S309).

  The client 200 receives the response from the application 60 (step S310), and stops the file processing using the shared lock and the exclusive lock (step S311).

  Note that the client 200 can execute file operations as needed in the section of steps S303 to S305 in FIG. Here, the file operation corresponds to an exclusive lock acquisition request, an exclusive lock release request, a shared lock acquisition request, and a shared lock release request. The server 100 having the role of the master that has received the file operation executes the processes of FIGS. 13 to 16 described above.

  Next, the effect of the system according to the present embodiment will be described. In the system according to the present embodiment, when the server 100 is a server having a master role, the server 100 issues a session ID to the client 200 in response to a connection request from the client 200, and the session ID is issued. In addition to registering in the session table 140a, in response to a file processing request from the client 200, the lock information for the file and the session ID are associated and registered in the file management table 140c. Further, when the server 100 is a server having a role of replica, the server 100 duplicates the session table 140a and the file management table 140c of the server having the role of master. In addition, when the server 100 switches from the server having the role of replica to the server having the role of master and receives a reconnection request from the client 200, the server 100 stores the information in the session table 140a as the session table waiting for deletion. 140b, deletes the same session ID as the session ID included in the reconnection request from the deletion waiting session table 140b, and stores the lock information corresponding to the session ID remaining in the deletion waiting session table 140b in the file management table. Delete from 140c.

  By executing such processing, even when the network is distributed among a plurality of synchronized servers, even in a system that selects a single master from a plurality of servers included in the same network, When another server takes over the lock information of the client for the resource due to the failure of the master, it is possible to prevent inconsistency in the lock information.

  Further, when the number of locks on the file is less than the upper limit, the server 100 associates the lock information on the file with the session ID and registers them in the file management table. By executing such processing, the number of resources used can be limited.

  Further, when the file processing request from the client 200 requests a shared lock for the file, the server 100 displays the lock information and the session ID for the file when the number of locks for the file is less than the upper limit number. When the file processing request from the client 200 requests an exclusive lock on the file, the file lock information and the session ID are registered when the file is not locked. Are registered in the file management table 140c. By executing such processing, the exclusive lock and the shared lock can be appropriately applied to the file while limiting the number of resources used.

  Further, when the server 100 having the master role is switched, the client 200 makes a reconnection request to the server 100 having the master role, and does not receive a response to the reconnection request from the server 100 having the master role. In such a case, the application 60 is notified of information on the reconnection failure. By performing such processing, it is possible to prevent the client 200 from performing file processing using a shared lock or exclusive lock on the server 100 in a situation where the connection with the server 100 is not established.

  Next, an example of a computer that executes a control program that realizes the same function as the server 100 described in the embodiment will be described. FIG. 18 is a diagram illustrating an example of a computer that executes a control program that implements the same function as the server. As illustrated in FIG. 18, the computer 300 includes, for example, a memory 301, a CPU (Central Processing Unit) 302, a hard disk drive interface 303, a disk drive interface 304, a serial port interface 305, a video adapter 306, A network interface 307, and these units are connected by a bus 308.

  As illustrated in FIG. 18, the memory 301 includes a ROM (Read Only Memory) 301a and a RAM (Random Access Memory) 301b. The ROM 301a stores a boot program such as BIOS (Basic Input Output System). The hard disk drive interface 303 is connected to the hard disk drive 303a as illustrated in FIG. The disk drive interface 304 is connected to the disk drive 304a as illustrated in FIG. For example, a removable storage medium such as a magnetic disk or an optical disk is inserted into the disk drive 304a. As illustrated in FIG. 18, the serial port interface 305 is connected to, for example, a mouse 305a and a keyboard 305b. The video adapter 306 is connected to a display 306a, for example, as illustrated in FIG.

  Here, as illustrated in FIG. 18, the hard disk drive 303 a stores, for example, an OS (Operating System) 1, an application program 2, a program module 3, and program data 4. That is, the control program for executing the processing of the server 100 is stored in, for example, the hard disk drive 303a as the program module 3 in which a command executed by the computer 300 is described. Specifically, the program module 3 in which the procedures corresponding to the master adjustment unit 150a, replication unit 150b, request processing unit 150c, and restoration processing unit 150d described in the above embodiment are described is stored in the hard disk drive 303a. . Further, data used for information processing by the control program, such as data stored in the storage unit 140 described in the above embodiment, is stored as program data 4 in, for example, the hard disk drive 303a or the memory 301. Then, the CPU 302 reads the program module 3 and program data 4 stored in the hard disk drive 303a to the RAM 301b as necessary, and executes each procedure.

  The program module 3 and the program data 4 relating to the control program are not limited to being stored in the hard disk drive 303a, but may be stored in, for example, a removable storage medium and read out by the CPU 302 via the disk drive or the like. Good. Alternatively, the program module 3 and the program data 4 relating to the control program are stored in another computer connected via a network (LAN (Local Area Network), WAN (Wide Area Network), etc.), and via a network interface. It may be read by the CPU 302.

  Next, an example of a computer that executes a control program that implements the same function as the client 200 will be described with reference to FIG. A control program for executing the processing of the client 200 is stored in, for example, the hard disk drive 303a as the program module 3 in which a command executed by the computer 300 is described. Specifically, the program module 3 in which the procedures corresponding to the connection request unit 250a, the request unit 250b, the reconnection request unit 250c, and the notification unit 250d described in the above embodiment are described is stored in the hard disk drive 303a. . Further, like data stored in the storage unit 240 described in the above embodiment, data used for information processing by the control program is stored as program data 4 in, for example, the hard disk drive 303a. Then, the CPU 302 reads the program module 3 and program data 4 stored in the hard disk drive 303a to the RAM 301b as necessary, and executes each procedure. The other description is the same as the description of the computer that executes the control program that implements the same function as the server 100 described above.

1 OS
2 Application program 3 Program module 4 Program data 50 Network 60 Application 100a, 100b, 100c, 100d, 100e Server 110, 210 Communication unit 140, 240 Storage unit 140a Session table 140b Deletion waiting session table 140c File management table 150, 250 Control unit 150a Master adjustment unit 150b Replication unit 150c Request processing unit 150d Restoration processing unit 200a, 200b Client 220 Input unit 230 Display unit 240a Session information 250a Connection request unit 250b Request unit 250c Reconnection request unit 250d Notification unit 300 Computer 301 Memory 301a ROM
301b RAM
302 CPU
303 Hard Disk Drive Interface 303a Hard Disk Drive 304 Disk Drive Interface 304a Disk Drive 305 Serial Port Interface 307 Network Interface 308 Bus

Claims (6)

A master coordinator that selects a single server having a master role from a plurality of other servers and its own server included in the same network;
When the own server is selected as the server having the master role by the master adjustment unit , in response to a connection request from the client, in response to the file processing request from the client, lock information and session ID for the file, When the number of locks for a file is less than the threshold indicated in the file management table, the lock information for the file and the session ID are associated and registered in the file management table. ,
When the number of locks on the file reaches the threshold indicated in the file management table, the lock information on the file and the session ID are associated with each other and registered in the lock waiting session, and the number of locks on the file is the file management When it becomes less than the threshold shown in the table, the lock information for the file registered in the lock waiting session and the session ID are associated with each other and registered in the file management table,
Issue a session ID and send information to the connected client for monitoring. If no response is received from the client, the lock information corresponding to the session ID issued to the client is deleted from the file management table. And
A request processing unit for locking the file based on a session ID and lock information registered in the file management table ;
A replication unit that replicates a session table and a file management table of a server having a master role when the master adjustment unit is selected as a server having a replica role by the master adjustment unit ;
When the local server switches from the server having the role of replica to the server having the role of master, the information of the session table replicated from the server having the role of master is registered in the session table waiting for deletion, and from the client In response to the reconnection request, the same session ID as the session ID included in the reconnection request is deleted from the deletion waiting session table, and lock information corresponding to the session ID remaining in the deletion waiting session table is stored. And a restoration processing unit for deleting from the file management table replicated from the server having the role of the master . When the file processing request from the client requests a shared lock on the file, the request processing unit locks the file when the number of locks on the file is less than the threshold indicated in the file management table. Information and session ID are associated with each other and registered in the file management table, and when the number of locks on the file reaches the threshold indicated in the file management table, the lock information on the file is associated with the session ID. When the number of locks on the file becomes less than the threshold indicated in the file management table, the lock information and the session ID associated with the file registered in the lock waiting session are associated with each other. Register in the file management table ,
If the file processing request from the client requests an exclusive lock on the file, if the file is not locked, the file lock information and the session ID are associated with each other and registered in the file management table. When the lock is applied to the client, the lock information for the file and the session ID are associated with each other and registered in the lock waiting session. When the lock on the file is released, the lock information is registered in the lock waiting session. The server according to claim 1, wherein lock information and a session ID associated with the registered file are registered in the file management table in association with each other. A file management system comprising a plurality of servers having a master or replica role and a plurality of clients connected to the server having a master role,
The server
A master coordinator that selects a single server having a master role from the same network;
When the master adjustment unit selects the server as a server having the master role, in response to a connection request from the client, a session ID is issued to the client, and the session ID has the master role When registering in the session table copied from the server and registering in the file management table in association with the lock information and session ID for the file in response to the file processing request from the client, the number of locks for the file is If it is less than the threshold shown in the file management table, the lock information for the file and the session ID are associated with each other and registered in the file management table,
When the number of locks on the file reaches the threshold indicated in the file management table, the lock information on the file and the session ID are associated with each other and registered in the lock waiting session, and the number of locks on the file is the file management When it becomes less than the threshold shown in the table, the lock information for the file registered in the lock waiting session and the session ID are associated with each other and registered in the file management table,
Sending and monitoring information to the connected client by issuing a session ID, and if the response from the client does not arrive, lock information corresponding to the session ID issued to the client has the role of the master Delete from the file management table replicated from the server ,
A request processing unit for locking the file based on a session ID and lock information registered in the file management table ;
A replication unit that replicates a session table and a file management table of a server having a master role when the master adjustment unit is selected as a server having a replica role by the master adjustment unit ;
When the local server switches from the server having the role of replica to the server having the role of master, the information of the session table replicated from the server having the role of master is registered in the session table waiting for deletion, and from the client In response to the reconnection request, the same session ID as the session ID included in the reconnection request is deleted from the deletion waiting session table, and lock information corresponding to the session ID remaining in the deletion waiting session table is stored. And a restoration processing unit for deleting from the file management table replicated from the server having the role of the master . In response to an instruction from the application, the client performs the file processing request to a server having a master role,
When the server having the master role is switched, the client makes the reconnection request to the server having the master role, and does not receive a response to the reconnection request from the server having the master role. The file management system according to claim 3, further comprising a notification unit that notifies the application of information that failed to reconnect. A file management method executed on a server,
A master adjustment step of selecting a single server having a master role from a plurality of other servers and its own server included in the same network;
When the server is elected as a server having the master role by the master adjustment step , in response to a connection request from the client, a session ID is issued to the client, and the session ID is registered in the session table. When the lock information for a file and the session ID are registered in the file management table in association with the file processing request from the client, the number of locks for the file is less than the threshold indicated in the file management table In addition, the file lock table and the session ID are associated with each other and registered in the file management table,
When the number of locks on the file reaches the threshold indicated in the file management table, the lock information on the file and the session ID are associated with each other and registered in the lock waiting session, and the number of locks on the file is the file management When it becomes less than the threshold shown in the table, the lock information for the file registered in the lock waiting session and the session ID are associated with each other and registered in the file management table,
Issue a session ID and send information to the connected client for monitoring. If no response is received from the client, the lock information corresponding to the session ID issued to the client is deleted from the file management table. Request processing step to
A replication step of replicating the session table and file management table of the server having the role of the master when the server is elected as a server having the role of the replica by the master adjustment step ;
When the local server switches from the server having the role of replica to the server having the role of master, the information of the session table replicated from the server having the role of master is registered in the session table waiting for deletion, and from the client In response to the reconnection request, the same session ID as the session ID included in the reconnection request is deleted from the deletion waiting session table, and lock information corresponding to the session ID remaining in the deletion waiting session table is stored. A restoration step of deleting the file from the file management table replicated from the server having the role of the master and locking the file based on the session ID and lock information registered in the file management table. File management method. A master adjustment step of selecting a single server having a master role from a plurality of other servers and its own server included in the same network;
When the server is selected as a server having the master role by the master adjustment step , in response to a connection request from the client, a session ID is issued to the client, and the session ID is registered in the session table. When the lock information for a file and the session ID are registered in the file management table in association with the file processing request from the client, the number of locks for the file is less than the threshold indicated in the file management table In addition, the file lock table and the session ID are associated with each other and registered in the file management table,
When the number of locks on the file reaches the threshold indicated in the file management table, the lock information on the file and the session ID are associated with each other and registered in the lock waiting session, and the number of locks on the file is the file management When it becomes less than the threshold shown in the table, the lock information for the file registered in the lock waiting session and the session ID are associated with each other and registered in the file management table,
Issue a session ID and send information to the connected client for monitoring. If no response is received from the client, the lock information corresponding to the session ID issued to the client is deleted from the file management table. Request processing steps to
A replication step for replicating a session table and a file management table of a server having a master role when the server is elected as a server having a replica role by the master adjustment step ;
When the local server switches from the server having the role of replica to the server having the role of master, the information of the session table replicated from the server having the role of master is registered in the session table waiting for deletion, and from the client In response to the reconnection request, the same session ID as the session ID included in the reconnection request is deleted from the deletion waiting session table, and lock information corresponding to the session ID remaining in the deletion waiting session table is stored. A restoration processing step for deleting the file from the file management table replicated from the server having the role of the master and locking the file based on the session ID and lock information registered in the file management table ;
File management program that causes a computer to execute.

Images