JPH06139213A - Computer system - Google Patents

Abstract

PURPOSE:To automatically carry on an associative processing between a client and a server by enabling the client to automatically recognize a processor which is substituted as a new server. CONSTITUTION:When trouble occurs to a processor (A) 11-1 which operates as the server of a single system 12, another processor substitutes for the server function of the processor. At this time, the single system 12 changes the server name in order according to the priority. When the server name matches the name of the server processor actually substituted for the new server, the single system 12 is connected to the processor. Consequently, the single system 12 can automatically recognize the processor which is substituted in the closely coupled multiprocessor system 11 as the server and automatically carries on the associative processing by the different processor.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a composite type of a loosely coupled multiprocessor system in which a plurality of processors are coupled and a client computer for issuing various data processing requests to the loosely coupled multiprocessor system. Regarding computer systems.

[0002]

2. Description of the Related Art Generally, a cooperative distributed system using a loosely coupled multiprocessor system is realized by a client / server model. In this case, each processor constituting the loosely coupled multiprocessor system serves as a server, and a workstation or the like connected to the loosely coupled multiprocessor system serves as a client. Functions are distributed among a plurality of processors that form a loosely coupled multiprocessor system, and a server function that can be provided for each processor is defined.

In a client / server type system using such a loosely coupled multiprocessor system,
The client requests various functions by designating the name of the processor that serves as the server.

However, in such a conventional system, if a failure occurs in the server and the processing is interrupted during the linked processing of the client and the server, the processing is interrupted.
It was not possible to restore the environment at the time of the interruption and continue the linked processing.

This is because, in the conventional method, even if the server can be replaced by another processor in the loosely coupled multiprocessor system, the client cannot recognize which processor has become the new server. For this reason, conventionally, there is a drawback that if the server fails during the client / server model interlocking process, it is necessary to restart the entire system.

[0006]

Conventionally, when a failure occurs in the server and the processing is interrupted during the interlocking processing of the client and the server, the processing such as restarting the entire system is required. Therefore, it was not possible to restore the environment at the time of interruption and continue the linked processing.

The present invention has been made in view of the above circumstances, and enables a client to automatically recognize a processor that has been replaced as a new server so that a failure occurs in the server during the interlocking process between the client and the server. It is an object of the present invention to provide a computer system capable of automatically continuing the linked processing by restoring the environment at the time of interruption even if the processing is interrupted due to the occurrence of a problem.

[0008]

Means and Actions for Solving the Problems
A loosely coupled multiprocessor system configured by coupling a plurality of processors and a client terminal connected to the loosely coupled multiprocessor system via a network, and the processor provides a service in response to a request from the client terminal. In a client / server model computer system provided, a loosely coupled multiprocessor system monitors a processor operating as a server of the client terminal and detects whether or not a failure has occurred, and when the failure has been detected. When the server function of the failed processor is replaced by another predetermined processor and the new processor name of which the server function is replaced and the server name transmitted from the client terminal are different, the client computer To And means for issuing error notification, the client terminal includes a management table in which priority of the processor to be assigned as a server is defined, in the event of failure of the processor running as a server,
Means for changing the server name in order of the priority of the processors defined in the management table and transmitting to the loosely coupled multiprocessor system; and within the loosely coupled multiprocessor system according to the presence / absence of error notification from the loosely coupled multiprocessor system. And a means for recognizing the new processor name replaced by and a means for changing the server name to the recognized new processor name and continuing the processing request.

In this computer system, when a processor operating as a server of a client computer fails, another server replaces the server function of that processor. At this time, in the client terminal, the connection destination processor name is changed in order of priority, and when the connection destination processor name and the server processor name actually assigned as a new server match, the client terminal is connected to that processor. It
Therefore, the client terminal can automatically recognize the alternative processor in the loosely-coupled multiprocessor system as a server, and some failure occurs in the server during the interlocking process between the client and the server and the processing is interrupted. Also, it becomes possible to restore the environment at the time of interruption and automatically continue the linked processing.

[0010]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows the system configuration of a computer system according to an embodiment of the present invention. This computer system includes a loosely coupled multiprocessor system 11 and
The loosely coupled multiprocessor system 11 is composed of a single system 12 such as a workstation connected via a network 10.

The loosely coupled multiprocessor system 11 includes
It is provided with four processors 11-1 to 11-4 which operate under independent operating systems. Further, these processors 11-1 to 11-4 have a configuration in which they are coupled by an inter-channel communication mechanism and share the magnetic disk device 111. This magnetic disk device 111
The checkpoint file 111a is stored in. The checkpoint file 111a is used for restarting the processing of the processor interrupted by the occurrence of a failure, and the checkpoint information indicating the operating environment of each processor is collected therein for each predetermined quiescent point. It Checkpoint file 111
a can be referred to by any processor.

The processors 11-1 to 11-4 are communication processors (ACP; A) which form a high-speed inter-processor coupling mechanism.
Messages are sent and received between each other via a dd-on node concatenation processor. Processor 11-1 to 1
Each of 1-4 provides various server functions to the single system 12. Various server functions such as a file server, a database server, and a mail server are prepared as server functions.

The stand-alone system 12 includes these processors 11
It functions as a client that uses the server functions of -1 to 11-4, and the processing is executed in association with the standalone system 12 and the server. In this case, the client / client between the loosely coupled multiprocessor system 11 and the stand-alone system 12
The server-type interprocess communication is realized by calling a procedure such as a remote procedure on the single system 12 side. The procedure called by the standalone system 12 includes the server name. When a procedure is called on the side of the stand-alone system 12, it is sent to the processor as a service request to the server. The server executes the service corresponding to the called procedure and sends it back to the single system 12 which is the client. Upon receiving the result, the standalone system 12 resumes execution from the next step after the called procedure.

One of the processors 11-1 to 11-4 functions as a master processor for controlling the loosely coupled multiprocessor system 11 as a whole. The master processor monitors the operating conditions of the other processors by message communication with other processors, detects the occurrence of a failure, and performs processing such as substituting a server of a processor interrupted by the failure with another processor. . The environment at the time of interruption is restored using the checkpoint file 111a.

Which processor is to be assigned as a new server is determined by using a master management table provided in the master processor. This master management table manages the load status, owned resources, etc. for each processor.

That is, when the master processor detects a failure of a processor operating as a server, the master processor refers to the management table and has the least load among other processors having sufficient resources to realize the server function. Select a processor and assign the selected processor to a new server.

In the figure, the processor 11-1 operating as a server with respect to the client process 121 of the stand-alone system 12 fails and the processor 11-1
Is shown as being assigned as a new server.

The standalone system 12 determines which processor a new server is assigned to by using the processor management table 122. The processor management table 122 defines the priority order of the processors that can be assigned to the single system 12 as a server. When a failure occurs in the server processor, the standalone system 12 changes the connection destination processor name in the priority order defined in the management table 122 and notifies the master processor of the change. The connection destination processor name is sequentially changed until the connection destination processor name and the server processor name actually assigned as the new server match. Whether there is a match is recognized by an error notification issued by the master processor. FIG. 2 shows the processor management table 1
An example of a specific configuration of 22 is shown.

Here, the processor (A) 11-1, the processor (B) 11-2, the processor (C) 11-3,
The case where the priority order of the servers for the single system 12 is assigned in the order of the processor (D) 11-4 is shown.

When the processor (A) 11-1 functions as a server of the stand-alone system 12 and the processing of the processor (A) 11-1 is interrupted due to the occurrence of a failure, the stand-alone system 12 is A) The processor having the highest priority among the processors other than 11-1, here the processor (B) 11-2 is the processor (A) 11-
Recognize that it works as an alternative to 1. Then, the processor name (B) is transmitted to the master processor as the server name. At this time, when the master processor issues an error notification, the server name is further changed to the processor name (C) having the next highest priority. FIG. 3 shows an example of a specific configuration of the loosely coupled multiprocessor system 11.

Here, it is assumed that the processor (C) 11-3 operates as a master processor. In this case, the processor (C) 11-3 is provided with the master management table 112.
Manages the load status and resources of each processor. Next, the processor substitution process controlled by the master processor will be described with reference to the flowchart of FIG.

The master processor (C) 11-3 sends a message indicating normal operation to another processor 1 within a certain period.
Whether or not the message is transmitted from 1-1, 11-2, 11-4 is detected, and the processor in which the message is not continuously transmitted within a certain period is regarded as having a failure. When the master processor (C) 11-3 detects a failure of the processor (A) 11-1 operating as a server of the stand-alone system 12 (step S11), it first notifies the stand-alone system 12 of the failure occurrence. The master management table 112 is referred to (step S12), and the alternative processor of the processor (A) 11-1 is recognized (step S13). As an alternative processor, a single system 12
The least loaded of the processors that have sufficient resources to implement the server function for
For example, the processor (D) 11-4 is selected.

Next, the master processor (C) 11-3
Issues a restart request to the alternative processor (D) 11-4 (step S14). Processor (D) 1
1-4 reads the checkpoint information of the processor (A) 11-1 from the checkpoint file 111a, and restores the environment when the processor (A) 11-1 is operating normally (step S15). Then, the processor (D) 11-4, instead of the processor (A) 11-1, restarts the processing from the time when the processing of the server is interrupted, and continuously executes the server function (steps S16 and S1).
7). Next, the update processing of the connection processor in the client will be described with reference to FIG.

When the stand-alone system 12 receives the fault occurrence notification from the master processor (C) 11-13 (step S21), it refers to the processor management table 121 to recognize the processor newly assigned as the server. , The server name is changed to the processor name of the next priority of the current server, and it is checked whether the server name matches the new server processor name (step S2).
2, S23).

Whether or not the server name matches the new server processor name can be recognized by the error notification from the master processor (C) 11-3. If the server processor name and the new server processor name are different, an error notification is issued, and the server processor name is further changed according to the error notification. In this example, the processor (D)
Since 11-4 is assigned as a new server processor, when the server name is changed to the processor (D), the server name matches the new server processor name,
After that, the interlocking process is restarted (step S24).

As described above, in this embodiment, when a failure occurs in the processor (A) 11-1 operating as a server of the stand-alone system 12, another processor substitutes the server function of that processor. . At this time, in the standalone system 12, the server processor name is sequentially changed in the order of priority, and when the server processor name and the server processor name actually assigned as a new server match, the standalone system 12 is connected to the processor. It Therefore, the processor replaced in the loosely coupled multiprocessor system 11 is replaced by the single system 1
2 can be automatically recognized as a server processor, and even if the server is interrupted and some processing is interrupted during the interlocking processing of the client and server, the interlocking processing can be separated by restoring the environment at the time of interruption. It becomes possible to continue automatically with the processor.

[0028]

As described above in detail, according to the present invention,
The client can automatically recognize the processor replaced as a new server, and the environment at the time of the interruption is restored even if the server fails and the processing is interrupted while the client and server work together. As a result, the linked processing can be automatically continued by another processor.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a computer system according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of specific contents of a processor management table provided in the client in the embodiment.

FIG. 3 is a diagram showing an example of a specific configuration of a loosely coupled multiprocessor system according to the first embodiment.

FIG. 4 is an exemplary flowchart for explaining a server processor alternative process in the loosely coupled multiprocessor system according to the first embodiment.

FIG. 5 is a flowchart illustrating server name update processing in the client of the embodiment.

[Explanation of symbols]

11 ... Loosely coupled multiprocessor system, 12 ... Single system, 11-1 to 11-4 ... Processor, 112 ... Master management table, 121 ... Processor management table.

Claims (1)

[Claims] 1. A loosely coupled multiprocessor system configured by coupling a plurality of processors, and a client terminal connected to the loosely coupled multiprocessor system via a network, and responding to a request from the client terminal. In a client / server model computer system that provides services by the processor, a loosely coupled multiprocessor system monitors a processor operating as a server of the client terminal and detects whether or not the failure has occurred; Is detected, the processor substitute means for substituting the server function of the faulty processor for another predetermined processor, the new processor name of which the server function is replaced, and the server name transmitted from the client terminal are different. When the cry And a means for issuing an error notification to the host computer, wherein the client terminal stores a management table in which the priority order of the processors assigned as the server is defined, and the management table in the management table when a failure occurs in the processor operating as the server. Means for changing the server name in order of the priority of the defined processor and transmitting to the loosely coupled multiprocessor system; and a means for replacing in the loosely coupled multiprocessor system according to the presence or absence of an error notification from the loosely coupled multiprocessor system A computer system comprising: means for recognizing a new processor name; and means for changing the server name to the recognized new processor name and continuing a processing request.