JP2785992B2 - Server program management processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Overview] Regarding a management processing method of a server program for managing an active server and a standby server deployed in a multiprocessor system, completely non-stop operation of services provided by the active and standby servers A dual space management server that executes management processing for non-stop operation of the active and standby servers is deployed on at least three or more processor modules, and the identification number of the processor module is further increased. The memory unit having a writing area and a writing area for the live notification information of the active dual-space management server is prepared, and the identification number of the processor module is written in the memory unit of the dual-space management server. When the server is not present, the server is set as the active dual space management server, and the The active setting unit that writes the identification number of the file, the live notification unit that periodically writes the live notification information to the memory unit when the own server is set as the active dual space management server, When the server is not set as a server, it monitors the survival notification information in the memory unit to monitor the occurrence of a failure in the active dual-space management server. The current monitoring unit writes the identification number of its own processor module in the memory unit, and when the current monitoring unit detects a failure, it is specified by the identification number of the processor module before update written in the memory unit. And a restart instruction unit that instructs the dual space management server to perform a restart.

[Industrial applications]

The present invention relates to a management processing method of a server program for managing an active server and a standby server deployed in a multiprocessor system, and in particular, realizes a completely non-stop operation of the active and standby servers deployed in a multiprocessor system. The present invention relates to a management processing method of a server program which enables the server program.

[Conventional technology]

The computer system adopts a configuration in which a server program for providing a general-purpose service is prepared and data processing is executed. Then, in order to prevent the service from being unable to be provided due to the server program going down, a plurality of identical server programs are prepared, and one of them is made to function as the active server, and the remaining one is used as the active server. One or more server programs function as standby servers, and when the active server goes down, pass control to one of the standby servers to prevent suspension of service provision. I am taking it.

On the other hand, in recent years, a data processing system is often configured with a multiprocessor configuration in order to improve the reliability and speed of the data processing by decentralizing the data processing.

In such a multiprocessor system, the active server and the standby server are deployed in separate processor modules, the operation information of the active server is saved in a non-volatile memory, and an abnormality (server crash) of the active server or the active server is saved. If the active server cannot be service-provided due to an error (PM crash) in the processor module in which the server is deployed, the standby server deployed in another processor module takes over the information for taking over from its non-volatile memory. And take over the service processing of the active server.

[Problems to be solved by the invention]

However, in a multiprocessor system,
As described above, the non-stop processing performed by the active server and the standby server itself complicates the control, and the reliability cannot be reduced because both the active server and the standby server cannot cope with a simultaneous crash. Will bring.

Therefore, in order to construct a more reliable nonstop operation system, a dual space management server that performs management processing for nonstop operation of services provided by the active and standby servers is prepared. The space management server starts the active server and the standby server and monitors their existence,
It is conceivable to adopt a configuration in which a takeover instruction process for the standby server, a new standby start process for the server to be newly set as the standby server, a restart process for the server that has been operating as a current server, and the like are executed. . However, in order to construct such a non-stop operation of the server program by the dual space management server, it is necessary to construct a non-stop mechanism of the dual space management server itself.

The present invention has been made in view of such circumstances,
Complete non-stop of the active and standby servers deployed in a multiprocessor system by realizing the non-stop of the dual space management server that performs management processing for the operation of the service provided by the active and standby servers without stop It is an object of the present invention to provide a new server program management processing method for realizing a simplified operation.

[Means for solving the problem]

 FIG. 1 is a diagram illustrating the principle of the present invention.

In the figure, reference numeral 1 denotes a data processing device equipped with the present invention, which is constituted by a multiprocessor system;
2 is a processor module, which is a data processing device 1
3 is an active server deployed in the processor module 2;
A server that provides a predetermined service, 4 is one or a plurality of standby servers that are deployed in the processor module 2 and is provided for backup of the active server 3, and 5 is at least three or more processor modules 2
Is a dual-space management server deployed on

The dual space management server 5 includes the active and standby servers 3, 4
Server management unit 50 that executes management processing for non-stop operation of the server, a survival notification unit 51 that writes a survival notification to the memory unit 7 described later, and a survival of the dual space management server 5 of the other processor module 2. A restart that instructs the active monitoring unit 52 that monitors the survival notification information of the memory unit 7 written by the notification unit 51 and the server management unit 50 of the dual space management server 5 of the other processor module 2 that has gone down. An instructing unit 53 and an active setting unit 54 that sets the own server as the active dual space management server by writing the identification number of the own processor module into the memory unit 7 are provided.

Reference numeral 7 denotes a memory unit having an operation information management unit 70 for managing operation information of the dual space management server 5. In this operation information management unit 70, the survival notification information and the identification number of the processor module are written.

[Action]

In the present invention, when a boot request is received from a kernel (not shown), the dual space management server 5 of each processor module 2
The active setting unit 54 performs a process of checking whether or not the identification number of the processor module 2 has been written in the operation information management unit 70 of the memory unit 7. As will be understood from the description below, when the identification number of the processor module 2 is written, it means that the dual-space management server 5 deployed in the processor module 2 of the identification number is operating as an active use. .

In this activation start state, which processor module 2
Since the dual space management server 5 is not set as a working server, the operation information management unit 70 is in a state where the identification number of the processor module is not written. from now on,
First, the active setting unit 54 of the dual space management server 5 that has checked the identification number of the processor module of the operation information management unit 70
Will detect that there is no active dual space management server 5.

The active setting unit 54 that has detected that there is no active dual-space management server 5 instructs the server management unit 50 of its own dual-space management server 5 to operate as the active dual-space management server 5, and The identification number of the own processor module is written in the operation information management unit 70 of the memory unit 7.

Upon receiving the instruction to enter the active mode, the server management section 50 starts to execute the management processing of the active and standby servers 3 and 4 and alives to the liveness notification section 51 of its own dual space management server 5. An instruction is given to write a notification, and in response to the instruction, the survival notification unit 51 performs a process to periodically write to the operation information management unit 70 that it is alive. Then, in response to the survival notification written to the operation information management unit 70, the active monitoring unit 52 of the dual space management server 5 of the other processor module 2 monitors the live state of the dual space management server 5 operating as the active. to go into. Thus, when the multiprocessor system is started,
Any one of the dual space management servers 5 will operate as the active one, and the remaining at least two dual space management servers 5 will operate as the standby.

When an error occurs in the server management unit 50 of the dual space management server 5 that operates as the active server, the active and standby servers 3 and 4 are activated.
When the management processing of the two-space management server 5 cannot be executed, the survival notification unit 51 of the dual-space management server 5 stops writing the survival notification to the operation information management unit 70. In response to the suspension of the write processing, the active monitoring unit 52 of the other processor module 2 which first detects this state sends the active dual space management server to the server management unit 50 of its own dual space management server 5. 5
And writes the identification number of its own processor module in the operation information management unit 70 of the memory unit 7. Then, in accordance with the instruction to enter the active server, the server management unit 50 starts executing the management processing of the active and standby servers 3 and 4. According to this processing, nonstop operation of the service provided by the active and standby servers 3 and 4 is realized.

The server management unit 50 that has turned to the current operation instructs the survival notification unit 51 of its own dual space management server 5 to write a survival notification, and upon receiving this instruction, the survival notification unit 51
Processes the operation information management unit 70 to periodically write that it is alive, and upon receiving the written alive notification, the active monitoring unit 52 of the dual space management server 5 of the other processor module 2 , In preparation for the downtime of the dual space management server 5 which has been turned to the current use.

Then, the restart instructing unit 53 of the dual space management server 5 that has been turned to the current use is the dual space management server specified by the identification number of the processor module before update written in the operation information management unit 70 of the memory unit 7. 5 server management unit 50
The server management unit 50 is regenerated by executing the restart process for (the one that has been operating as before) so that at least two or more dual space management servers 5 are always on standby. Process to be in state. By this processing, even if an abnormality occurs in the dual-space management server 5 that is operating as a working server and an abnormality occurs simultaneously in the dual-space management server 5 that is selected as a standby, at least one dual-space management server 5 can be used. Since the management server 5 is left on standby, the operation of the active and standby servers 3 and 4 does not stop.

〔Example〕

 Hereinafter, the present invention will be described in detail with reference to examples.

FIG. 2 shows an embodiment of the present invention. In the figure, the same components as those described in FIG. 1 are indicated by the same symbols. Reference numeral 6 denotes a kernel which is a nuclear program developed in each processor module 2;
A non-volatile memory 71 corresponding to the memory unit 7 shown in the figure is a processor number storage unit constituting the operation information management unit 70 shown in FIG. 1, and stores the processor number of the processor module 2 which deploys the active dual space management server 5. A counter 72 stores the operation information management unit 70 shown in FIG. 1, and is used to write a count value by which the active dual space management server 5 advances. Here, when the system is started, a processor number that does not exist is set as an initial value in the processor number storage 71, and “0” is set as an initial value in the counter 72.

In this embodiment, a server program (the active server 3 and the standby server 4) is deployed in all the processor modules 2 constituting the multiprocessor system, and the non-stop operation of the active and standby servers 3 and 4 is performed. Is shown in which the dual space management server 5 that executes the management process of FIG. If the number of the processor modules 2 has a margin, the processor module 2 that deploys the dual space management server 5 can be configured differently from the processor module 2 that deploys the active and standby servers 3 and 4. Even in the case of adopting this configuration, a configuration is adopted in which the dual space management server 5 is deployed on at least three or more processor modules 2 in order to cope with a compound crash of the dual space management server 5. Become.

The dual space management server 5 prepared as described above functions as a server program for executing management processing for nonstop operation of the active and standby servers 3 and 4, as described in FIG. More specifically, the activation processing of the active server 3 and the standby server 4 and their survival monitoring processing, the takeover instruction processing to the standby server 4, the new standby activation processing to the server to be the new standby server 4 Then, a process such as a restart process of the server which has been operating as a current server is executed.

Next, the operation processing of the present invention will be described in detail with reference to the flowchart shown in FIG. 3 executed by the dual space management server 5 of each processor module 2.

First, in step 1, a system IPL is performed. When the IPL is executed in this manner, the kernel 6 of each processor module 2 performs a process of activating the dual space management server 5 of its own processor module 2 in step 2. When started in this way, the dual space management server 5 of each processor module 2 refers to the processor number storage 71 in the subsequent step 3 to
It is checked whether or not there is a dual space management server 5 that is set as the working space. When the system is started, a nonexistent processor number is written in the processor number storage 71 as described above as an initial value. from now on,
First, the dual space management server 5 referring to the processor number storage unit 71 detects that there is no active dual space management server 5.

The bi-spatial management server 5, which first detects that the active bi-spatial management server 5 does not exist, proceeds to step 4 and executes an initialization process to operate as the active bi-spatial management server 5. The dual space management server 5, which has been made active according to the conversion processing, performs processing to start the execution of management processing for nonstop operation of the active and standby servers 3, 4. Then, in the following step 5, the processor number of the processor module 2 in which it is deployed is stored in the processor number storage unit 71, and the count value of the counter unit 72 is incremented according to a predetermined cycle. Then, the process enters a loop process of writing survival information indicating that the nonvolatile memory 7a is currently operating.

On the other hand, in the processing of step 3, the twin space management server 5, which refers to the processor number storage unit 71 later than the dual space management server 5 set as current, uses the current space according to the processor number stored in the processing of step 5. Since the presence of the dual space management server 5 is detected, the process proceeds to step 6 and the standby dual space management server 5
The dual space management server 5 that has been on standby according to this initialization process operates to enter the backup process of the dual space management server 5 that is currently operating. In this way, when the multiprocessor system is started, one of the dual space management servers 5 operates as an active server, and all the remaining dual space management servers 5 (at least two or more exist). ) Will wait and work. And
In the subsequent step 7, the standby dual space management server 5 refers to the count value of the counter unit 72 according to a predetermined cycle to determine whether the active dual space management server 5 is alive or not. Enter the monitoring loop process.

If an abnormality occurs in the dual-space management server 5 that is operating as a working device, the process of step 5 is not performed, so that the count value of the counter unit 72 stops at a constant value and does not advance. From now on, according to the processing of step 7, the standby dual space management server 5, which first detects the stop of the increment of the count value of the counter unit 72, proceeds to step 8, and operates the current dual space By taking over the processing of the management server 5 and operating as a new active dual-space management server 5, processing is performed to execute the management processing for the non-stop operation of the active and standby servers 3, 4, and Proceeding to step 5, the processor number of the processor module 2 in which the processor module 2 is loaded is stored in the processor number storage section 71, and a new step of incrementing the count value of the counter section 72 is executed. In this way, a process is started to write the survival information indicating that the nonvolatile memory 7a is currently operating. In accordance with the registration processing of the new active dual-space management server 5, the other standby dual-space management servers 5 will detect the survival of the active dual-space management server 5 in the processing of step 7, so that The loop processing for monitoring whether or not the dual space management server 5 operating as it is still alive is continued.

The kernel 6 of the processor module 2 to be deployed in the active dual-space management server 5 then proceeds to step 9
Then, is the cause of the down of the double space management server 5 (which is the dual space management server 5 which has been operating as the current one) being caused by the server crash of the double space management server 5 or the PM of the processor module 2? Determine if you are in a crash.

If it is determined in step 9 that it is due to a server crash, the process proceeds to step 10, in which the active twin space management server 5 restarts the down dual space management server 5, and then restarts. The bi-spatial management server 5 proceeds to step 6 via step 3 and is processed to operate as a standby bi-spatial management server 5. On the other hand, when it is determined in step 9 that the crash is caused by the PM crash, the process returns to step 1 to execute the PM initialization process, which is the same process as in the system IPL. By restarting the management server 5 and proceeding to step 6 via step 3, the standby dual space management server 5
Will be processed.

Thus, according to the present invention, the active and standby servers 3,
The bi-spatial management server 5 that executes the management process 4 is configured such that at least two or more bi-spatial management servers 5 other than the active one are always registered in a standby state.

FIG. 4 schematically shows the processing contents of the present invention described above.

That is, as shown in FIG.
Starts the corresponding dual space management server 5 (in the figure) when the initialization of the processor module 2 is completed, and each activated dual space management server 5 refers to the processor number storage unit 71 to make the active It is checked whether or not the dual space management server 5 exists, and first, the current dual space management server 5 is used.
Space management server 5 (in the figure) that detects that there is no (the space management server 5 of PM01 in the example of this figure)
After writing its own PM number in the processor number storage unit 71 (in the figure), it initializes itself (in the figure) as the current working. Thereafter, the dual space management server 5 of another processor module 2 refers to the counter unit 72 (in the figure), and thereby the current dual space management server 5 (PM in the example of this figure).
When the presence of the dual space management server 5) 01 is detected, it is initialized (in the figure) as a standby dual space management server 5.

Then, as shown in FIG. 4 (b), the dual space management server 5 (the dual space management server 5 of PM01 in the example shown in FIG. 4) operating as an active device increments the count value of the counter 72 during normal operation. On the other hand, the bi-spatial management server 5 that operates as a stand-by, refers to the stepped state of the count value of the counter unit 72 (in the drawing), and the current bi-spatial management server 5 operates normally. Monitors whether it is running. In accordance with this monitoring process, when it is detected that the current dual-space management server 5 crashes (in the figure), the dual-space management server 5 (PM in the example of this figure) that first detects the crash.
03 dual space management server 5), the processor number storage 71
Is changed to its own processor number (in the figure), a takeover process (in the figure) for operating as the current working is performed, and the operation starts as the active dual space management server 5.

Then, as shown in FIG. 4 (c), if a server crash occurs after the crash recovery, the newly used dual space management server 5 (in this example, the dual space management server 5 of PM03). Restarts the dual space management server 5 of the currently active processor module 2 (in the figure), and if a PM crash occurs, the kernel 6 of the currently active processor module 2 becomes The dual space management server 5 is restarted (in the figure). Thereafter, the restarted twin space management server 5 is initialized (in the figure) as a standby twin space management server 5.

Although the illustrated embodiment has been described, the present invention is not limited to this. For example, in the embodiment, the configuration in which the active and standby servers 3 and 4 and the dual space management server 5 are deployed on the same processor module 2 is disclosed. However, the present invention is not limited thereto. May be developed in separate processor modules 2.

〔The invention's effect〕

As described above, according to the present invention, even if an abnormality occurs in the dual space management server that executes the management processing of the active and standby servers deployed on the multiprocessor system,
Since at least one dual space management server is always configured to be left as a standby, the management processing of the active and standby servers does not stop. from now on,
The non-stop operation of the service provided by the active and standby servers deployed in the multiprocessor system will be realized.

[Brief description of the drawings]

FIG. 1 is a block diagram of the principle of the present invention, FIG. 2 is an embodiment of the present invention, FIG. 3 is a flowchart executed by a dual space management server, and FIG. 4 is an explanatory diagram of the processing of the present invention. In the figure, 1 is a data processing device, 2 is a processor module, 3 is an active server, 4 is a standby server, 5 is a dual space management server, 6 is a kernel, 7 is a memory unit, 7a is a non-volatile memory, and 50 is a server management unit Reference numeral 51 denotes a survival notification unit, 52 denotes an active monitoring unit, 53 denotes a restart instruction unit, 70 denotes an operation information management unit, 71 denotes a processor number storage unit, and 72 denotes a counter unit.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G06F 15/177 G06F 15/167

Claims (1)

(57) [Claims] 1. A server program management system for managing an active server (3) and a standby server (4) deployed in a multiprocessor system, wherein at least three or more processor modules (2) are provided. Then, a dual space management server (5) for executing management processing for non-stop operation of the active and standby servers (3, 4) is developed,
Further, a writing area for the identification number of the processor module,
A memory unit (7) having an area for writing the survival notification information of the active dual space management server is adopted, and the dual space management server (5) includes a processor module in the memory unit (7). When the identification number is not written, the own server is set as the active dual space management server and the memory unit (7)
An active setting unit (54) for writing the identification number of the own processor module to the server; and an alive notification that periodically writes alive notification information to the memory unit (7) when the own server is set as the active dual space management server. And monitoring the occurrence of a failure in the active dual space management server by monitoring the survival notification information in the memory unit (7) when the own server is not set as the active dual space management server. When detecting the occurrence of a failure, the current monitoring unit (52) that sets its own server as the current dual space management server and writes the identification number of its own processor module into the memory unit (7); When 52) detects the occurrence of a failure, the dual space management server (5) specified by the identification number of the processor module before update written in the memory unit (7) Restart instruction unit that instructs the restart for (5
And 3) a server program management processing method.