JPH05334161A - Multi-server system - Google Patents

Abstract

PURPOSE:To provide a multi-server system for automatically holding the contents of files in plural file server devices in the same state. CONSTITUTION:In a master file server, a writing processing part 12 or an updating information processing part 14 receives a writing request or updating information to a self-disk and rewrites the disk through an I/O processing part 11. An updating information forming part 13 informs the updating information to another file server device. In a slave file server, a writing processing part 12 receives a writing request to its own disk and an updating information forming part 13 forms updating information for the disk and informs the updating information to the other file server device. An updating information processing part 14 receives the updating information and rewrites the disk through an I/O processing part 11.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is a multi-server system that enables a host computer to access its own disk in a network file system, and maintains the same file contents among a plurality of file server devices. Related to the propagation of file update information for.

[0002]

2. Description of the Related Art In recent years, a plurality of personal computers jointly use a database that is centrally managed in a computer network. For this reason, various file servers having a function of accessing a file device of another computer connected to the network as well as accessing a file device in the computer's own system have been developed.

Conventionally, an NFS (Network File System: USA) has been used as a network system in which a host computer of a client (hereinafter referred to as a client) can mount a file system on a file server (hereinafter referred to as a server) and read / write the contents of a disk of the server. Sun Microsystems, Networking on the Sun Workstatio
n, Sun User's Manual). In NFS, application software and common files for users can be placed on the server and used by multiple clients.

In such a file server, application software and the like are frequently used by many clients. Therefore, in consideration of the load on the server, it should be distributed to a plurality of servers (hereinafter referred to as a multi-server system). There are many. An example of a conventional system will be described below with reference to the drawings. FIG. 12 shows an example of the configuration of a system using a conventional file server. In the figure, the client a mounts the server A and the client b mounts the server B. It shows a case where two clients mount two servers. Two
One server is provided because application software and the like are frequently used by many clients, and are distributed to a plurality of servers in consideration of the load on the server.

[0005]

However, the conventional technique has the following two problems. One is that, when files having the same content exist in a plurality of servers, the files are updated independently in each server, so that the content of the application software differs between the servers. It is a point. For example, if a file X having the same content exists on the servers A and B, and the client a updates the content of the file X of the server A to X ′, the update is not propagated to the server B, and therefore the file is transmitted between the servers. Contents are different like X and X '. To avoid this, if the servers A and B are providing the same file, the update of the file must be performed manually for each of the same update operations.
It will be troublesome twice.

Another is that even if the server mounted by the client goes down or the load on the server becomes high, even if the server mounted by the client is switched, file access in the same environment as before can be performed. The problem is that it will disappear. For example, even if the client b switches the mount to the server A when the server B goes down or the load increases, the server A
Since it is not guaranteed that the file contents of the server B and the server B are the same, the client b cannot access the file in the same environment as before.

In view of the above problems, it is an object of the present invention to provide a multi-server system that enables the contents of files to be automatically kept the same between servers.

[0008]

In order to solve the above-mentioned problems, a multi-server system of the present invention is a multi-server system provided with a plurality of file server devices capable of accessing their own disks from other computers in a computer network. Therefore, when another computer on the network receives a write request for its own disc, or when it receives update information from another file server device, it writes it to its own disc and updates the same update information as the write request. When a master file server device to be given to another file server device and a write request to its own disk from another computer on the network are received, update information having the same contents as the write request is given to the master file server device to give a master file. Update from server device Upon receiving the broadcast, it is made from the slave file server device to write to their own disk.

The master file server device receives input / output processing means for reading / writing data from / to the disk and a write request for the own disk from another computer on the network, and rewrites the disk through the input / output processing means. A write processing unit, a master update information processing unit that receives update information from another file server device, and rewrites the disk through the input / output processing unit, and an update having the same content as the write request notified from the master write processing unit. The master update information generating means may generate information and notify other file server devices of the update information.

The slave file server device has an input / output processing means for reading / writing data from / to a disk, a slave receiving means for receiving a write request to its own disk from another computer on the network, and a notification from the slave receiving means. Generated update information having the same contents as the write request and notifying the master file server device of the update information, and the update information from the master file server device and receiving the update information from the master file server device. And a slave update information processing means for rewriting

Further, in the master file server device, the master update information generating means adds a serial number to the update information to be generated, and when there is no normal response to the update information notified to the slave file server device, Retaining update information, in the slave file server device, the slave update information processing means determines whether or not the serial numbers of the received update information are in numerical order,
The slave update information generation means may notify the master file server device of request information requesting update information of a flying number when it is determined that the serial numbers of the update information are not in numerical order.

[0012]

According to the multi-server system of the present invention, the master file server device receives the update request from the other computer on the network for its own disk or the update information from the other file server device. When it receives it, it writes it to its own disk and gives update information having the same contents as the write request to another file server device. When the slave file server device receives a write request for its own disk from another computer on the network, the slave file server device gives update information having the same contents as the write request to the master file server device and receives the update information from the master file server device. And write it on your own disc.

That is, in the master file server device, the master write processing means / master update information processing means receives the write request / update information for its own disk and rewrites the disk through the master input / output processing means. Furthermore, the master update information generation means is
The update information is notified to another file server device. In the slave file server device, the slave write processing means receives a write request for its own disk, and the slave update information generation means generates the update information and notifies the other file server device. The slave update information processing means receives the update information and rewrites the disk through the slave input / output processing means.

[0014]

Embodiments of the multi-server system of the present invention will be described below with reference to the drawings. The multi-server system includes one master file server device and other slave file server devices. FIG. 1 is a block diagram showing the configuration of a file server device (having both a master and a slave) used in a multi-server system according to an embodiment of the present invention.

An input / output processing unit 11 performs input / output processing to / from the disk. A write processing unit 12 processes a write request from the client. An update information generation unit 13 generates / stores file update information to another server in response to a write request from the client. An update information processing unit 14 processes update information from another server.

FIG. 2 is a block diagram showing the configuration of the master file server device. Reference numeral 21 denotes a master input / output processing unit that performs input / output processing to / from the disk. A master write processing unit 22 receives a write request for its own disk from another computer on the network, and writes it to the disk through the master input / output processing unit 21.

Reference numeral 23 denotes a master update information generation unit, which generates a write request notified from the master write processing unit 22 or update information having the same contents as the update information notified from the master update information processing unit 24, and updates the update information. The information is notified to another file server device. A serial number is added to this update information. When there is no normal response to the update information notified to the slave file server device, the update information is held.

The master update information processing section 24 receives update information from another file server device and writes it to the disk through the master input / output processing section 21. FIG. 3 is a block diagram showing the configuration of the slave file server device. Reference numeral 31 is a slave input / output processing unit that reads / writes data from / to the disk. A slave write processing unit (slave receiving unit) 32 receives a write request from another computer on the network to its own disk.

A slave update information generator 33 generates update information having the same content as the write request notified from the slave receiver 32, and notifies the master file server device of the update information. When the slave update information processing unit 34 determines that the serial numbers of the update information received from the master file server device are not in numerical order, it generates request information requesting update information of the flying number, and the master file server device To notify.

Reference numeral 34 denotes a slave update information processing unit, which receives update information from the master file server device and writes it to the disk through the slave input / output processing unit 31. At this time, it is determined whether or not the serial numbers of the received update information are in the numerical order, and if they are not in the numerical order, the update information is temporarily stored without being written to the disk, and the slave update information generation unit 33 is notified to that effect. To do.

The multi-server system in this embodiment is composed of a plurality of file server devices (one is a master file server device and the other is a slave file server device), is connected to a computer network, and can be freely accessed from a network-like computer. To be done. Here, as a specific example, it is composed of three file server devices,
As shown in FIGS. 4, 6 and 8, the file server device A is the master file server device and the file server device B is
And C are slave file server devices (hereinafter referred to as server A, server B, and server C).

Regarding the operation of the multi-server system configured as described above, (1) when there is a write request to the master server, (2) when there is a write request to the slave server, (3) Description will be given separately when one of the servers is down. (1) When a write request is made to the disk of the master server A.

FIG. 4 shows how the update information in this case propagates.
FIG. 5 shows a sequence of information shown in FIG. When the write request from the computer of the client a is notified to the server A (41 in FIG. 4, S51 in FIG. 5), the master write processing unit 22 of the server A is requested via the master input / output processing unit 21. The master information updating unit 23 is instructed to generate an update information record when the master I / O processing unit 21 receives the fact that the writing is normally completed (42). The master information updating unit 23
A corresponding update information record is generated and the server B,
It is transmitted to C (43, S52, S53). The format of this update information record is shown in FIG. Reference numeral 510 is an update information record, which is information 5 indicating that it is update information.
11, an identification number 512 indicating the update order, and a plurality of information 513, 514, 515 (file path name, update type, data) indicating the update content.

In the servers B and C that have received the update information record 510, the update information processing unit 4 first analyzes the content of the record and writes it to the disk through the input / output processing unit 1 (44) and returns the response record. 520 is generated and sent to the server A (S54, S55). This reply record 520, as shown in FIG. 10, is information 521 indicating that it is a reply to the update information record 510.
And 522 indicating the identification number of the slave server and the identification number 523 of the update information. When the reply record arrives, the server A knows that the update has been normally performed. In this way, the writing performed on the file of the server A is reflected in the same file of the servers B and C, and the same content is maintained.

Further, the update information record 510 sent by the server A is provided with an identification number 512 in order. By using this number, the servers B and C temporarily store the received update information record when there is an update information record with a number smaller than the received update information record 510 that has not been received yet. Then, by sending the request record 530 to the server A, the update information record that has not been received is received from the server A, and the records are processed in numerical order to guarantee the processing order. As shown in FIG. 10, the request record 530 includes information 531 indicating a request, 532 indicating the identification number of the slave server, and 533 indicating the identification number of the update information record 51 to be requested.

(2) When there is a write request to the disk of slave server B. FIG. 6 shows how the update information propagates in this case, and FIG. 7 shows a sequence of information propagated between the servers. When receiving a write request from the computer of the client b to the disk of its own host (61, S71), the slave receiving unit 32 of the server B is received.
Requests the slave update information generator 33 to generate update information. The slave update information generation unit 33 generates an update information record 510 based on the request and sends it to the server A (62, S72).

The server A that has received the update information record generates the reply record 520 shown in FIG.
(S73). In the server A, the master update information processing unit 24 then analyzes the content of the update information record,
Writing to the disk is performed through the master input / output processing unit 21 (63). If the writing is completed normally, the master update information generation unit 23 generates the update information record 510 and sends it to the servers B and C (64, S74, S7).
5).

In response to this, the servers B and C perform the above (1).
By the same process as described above, writing is performed on the disc of its own (65) and the reply record 520 is sent to the server A (S76, S77). (3) When one of the servers is down. FIG. 8 shows how update information is propagated when one of the slave servers is down, and FIG. 9 shows a sequence of information propagated between the servers.

When the master server A sends the update information record 510 to the slave servers B and C (S92) and the server B is down, a normal response is returned from the server C (S93). There is no response from the server B. In such a case, if the master server A receives no response even after a certain period of time, the master update information generation unit 23 saves the service update information record as a save record 610 in the disk or the memory. The format of this saved record is shown in FIG. The saved record 61 is a list 611 of identification numbers of servers to which the update information record should be sent.
And the update information record 612 to be transmitted, and the update information record 510 has a format in which an identification number list of servers that have not responded is added.

After that, when the master server A sends other update information to the servers B and C (S95), a normal response is returned from the server C (S96), but from the server B, the above (1) The request information 530 is returned as described in () (S97). In response to this, the master server A sends the previous update information (S98). The server B returns a write normal response to the disk by the previous update information (S99), and writes it on the disk by the subsequent update information. In this way, it is possible to maintain the file identity even if the server that was down is restored.

In the above (3), when the server B does not respond, the master server A also sends the next update information to the server B, waits for the request information from the server B, and sends the saved record 610. However, the recovery of the server B that is down from the master server A is not actively detected.
Therefore, when the server B does not respond, the master server A periodically sends recovery confirmation information (update information may be sent) to the server B to determine whether or not the response has been received, and restores. In this case, the stored records are sent to the server B in the order of identification numbers. In this way, the file update while the server B is down can be reflected on the disk of the server B to maintain the file identity.

[0032]

As described above, according to the present invention, in a multi-server system in which another host computer can access its own disk in a network file system, a plurality of file server devices are propagated by propagating update information. There is an effect that the identity of the file contents can be automatically maintained between them.

In addition, by adding serial numbers to the update information sent from the master server, the slave server has a function of rearranging the update information based on the serial numbers and reflecting it in its own disk in that order.
There is an effect that the update order of the same file can be made the same between servers. Furthermore, when another file server is down, by saving the update information to that file server, the contents of the same file can be kept the same with the temporarily down server. effective.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a file server device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a master file server device in the embodiment.

FIG. 3 is a block diagram showing a configuration of a slave file server device in the embodiment.

FIG. 4 is a diagram showing how update information is propagated when a write request is made to the disk of the master server in the embodiment.

FIG. 5 is a diagram showing a sequence of information propagated between servers when a write request is made to a disk of a master server in the embodiment.

FIG. 6 is a diagram showing how update information is propagated when a write request is made to a disk of a slave server in the embodiment.

FIG. 7 is a diagram showing a sequence of information propagating between servers when a write request is made to a disk of a slave server in the embodiment.

FIG. 8 is a diagram showing how update information is propagated when one of the slave servers is down in the embodiment.

FIG. 9 is a diagram showing a sequence of information propagating between servers when one of the slave servers is down in the embodiment.

FIG. 10 is a diagram showing an example of a format of data exchanged between servers in the embodiment.

FIG. 11 is a diagram showing a format example of data stored in the master server when the slave server is down in the embodiment.

FIG. 12 is a conceptual diagram of communication between a conventional server and a client computer.

[Explanation of Codes] 11 Input / Output Processing Unit 12 Write Processing Unit 13 Update Information Generation Unit 14 Update Information Processing Unit 21 Master Input / Output Processing Unit 22 Master Write Processing Unit 23 Master Update Information Generation Unit 24 Master Update Information Processing Unit 31 Slave Input Output processing unit 32 Slave writing processing unit (slave reception unit) 33 Slave update information generation unit 34 Slave update information processing unit 510 Update information record 520 Response record 530 Update information request record 610 Update information storage record sent from master server to slave server

Claims (4)

[Claims] 1. A multi-server system in a computer network, comprising a plurality of file server devices capable of accessing its own disk from another computer, wherein a write request is made to the own disk from another computer on the network. When receiving the update information from another file server device, the master file server device that writes the update information having the same content as the write request to the other file server device and writes it to its own disc When a computer receives a write request for its own disc, it gives update information having the same contents as the write request to the master file server device, and when it receives update information from the master file server device, it writes it to its own disc. When Multiserver system, characterized in that Ranaru. 2. The master file server device receives input / output processing means for reading / writing data from / to the disk and write request to its own disk from another computer on the network, and writes the disk through the input / output processing means. Master write processing means for rewriting, master update information processing means for receiving update information from another file server device, and rewriting the disk through the input / output processing means, and write request notified from the master write processing means. 2. The multi-server system according to claim 1, further comprising: master update information generation means for generating the update information of and updating the update information to other file server devices. 3. The slave file server device comprises an input / output processing means for reading / writing data from / to a disk, a slave receiving means for receiving a write request from another computer on the network to the own disk, and a slave receiving means. Generate update information having the same content as the write request notified from the slave update information generating means for notifying the master file server device of the update information, and receiving the update information from the master file server device,
2. The multi-server system according to claim 1, further comprising slave update information processing means for rewriting the disk through the input / output processing means. 4. The master file server device according to claim 2, wherein the master update information generating means adds a serial number to the update information to be generated, and when there is no normal response to the update information notified to the slave file server device. In the slave file server device according to claim 3, the slave update information processing means determines whether or not the serial numbers of the received update information are in numerical order, and the slave update information generating means A multi-server system characterized in that, when it is determined that the serial numbers of the update information are not in numerical order, the master file server device is notified of request information requesting update information of the flying numbers.