KR100437184B1 - Method for Automatic Release of Database on Stanby Loading in Switching System - Google Patents

Abstract

The present invention relates to the automatic release of the write-protected state of the database during standby loading, in standby loading of the exchange system. In particular, the loading block responsible for the standby loading automatically releases the write-protected state of the database by monitoring the states of the processors. The present invention relates to a method for automatically freeing a database during standby loading of an exchange system.

In the case of the conventional standby loading, the operator must release the write-protection state of the DB directly. Therefore, if the write-protection state of the DB is not released by the operator's mistake, the application can continuously try to write or cannot perform the operation related to the DB. Due to this there was a problem that the operation of the exchange is not possible at all.

The present invention by automatically monitoring the state of the processor at the same time after the standby loading and release the write-protected state of the DB, it is possible to ensure the stable running of the package and the stability of the exchange service after the transfer, the exchanger due to operator error It is effective to prevent the disorder.

Description

Method for Automatic Release of Database on Stanby Loading in Switching System}

The present invention relates to the automatic release of the write-protected state of the database during standby loading, in standby loading of the exchange system. In particular, the loading block responsible for the standby loading automatically releases the write-protected state of the database by monitoring the states of the processors. The present invention relates to a method for automatically freeing a database during standby loading of an exchange system.

In general, standby loading of an exchange means loading a new version of a program into a standby processor among redundant processors in order to minimize service downtime when a program is changed in an operating exchange system. To run a program.

This standby loading system is a redundant system management processor (hereinafter referred to as 'SMP') responsible for the operation and management of the entire system including the standby loading as shown in Figure 1 of the accompanying drawings (10-A, 10-B) and a plurality of redundant main processors (hereinafter referred to as' MP's) which are loaded by the SMPs 10-A and 10-B and serve the main functions of the exchange (20-A, 20- B) and a hard disk drive (hereinafter referred to as 'HDD') 30 in which a loaded program is stored.

Hereinafter, since the duplicated processor has the same structure, the processor will be described mainly in an active state.

The SMP 10 -A includes a system loading handler that performs a loading server function (hereinafter referred to as SLH) 11 -A and a processor loading processor that performs a loading client function ( Processor Loading Handler (hereinafter referred to as 'PLH') 12-A is mounted.

The plurality of MPs 20 -A are mounted with a PLH 21 -A which performs a loading client function, and is connected to the SMP 10 -A by an Inter Processor Communication (IPC).

The HDD 30 is composed of a root (ROOT) 31 and a standby (SBY) 32 directory, which are logical paths that are logically divided when performing a standby loading, and the SMP 10-A and a SCSI (Small Computer System Interface). Leads to.

The operation of the standby loading system configured as described above will be described in detail with reference to FIG. 2.

Since the standby loading method of SMP and MP is the same, it demonstrates based on SMP.

First, an active SLH 11-A is referred to as a processor loading data (hereinafter referred to as 'PLD') that means an application and a database (hereinafter, referred to as 'DB') used by the application. ), A new version of the package including the operating system and the like is recorded in the SBY 32 (step S1).

Then, the synchronization with the SMP 10-B in the standby state is stopped so that the redundancy is terminated, and the SMP 10-A to which it belongs is kept performing the active operation independently (step S2).

Thereafter, the new version of the package recorded in the SBY 32 directory is loaded into the SMP 10-B in the standby state (step S3). Then, the old version of the package is loaded into the SMP 10-A that is currently active, and the new version of the package is loaded into the standby SMP 10-B and is ready to be driven.

Thereafter, the processors are switched so that the SMPs 10 operating in the active or standby mode do the opposite operations in accordance with the existing synchronization (step S4). Then, the SMP 10-A that has been active is in a standby state, and the SMP 10-B that is in the standby state is activated and is loaded with a new version of the package (step S5).

At this time, PLH (not shown in the drawing) of the SMP 10-B that is active by switching over writes a new DB to be used by a new version of the application to ensure the integrity of the DB at the moment of switching. The reason for making the DB into the write inhibit state as described above is described below with reference to FIG. 3.

First, suppose there is a redundant SMP and a redundant MP, and a new version of the package is loaded on each standby processor.

In this state, if the processor is switched, and as a result, both the SMP and the MP are switched and the loaded new version of the package is driven, no problem occurs. However, if only one of the SMP and the MP is switched, for example, as shown in FIG. Similarly, when only SMP is transferred, a new version of the package is driven by the transferred SMP (10-A), and an old version of the package is driven by the unchanged MP (20-A). The PLD mismatch of the package being driven causes a catastrophic failure of the exchange operation.

Therefore, the DB must be made write-protected when switching processors.

After that, the operator checks the transfer status of all processors, and if the processors are normally transferred, directly releases the write-protection state of the DB (steps S6 and S7), and recovers the abnormal state of the processor if there is an unchanged processor. After that, the write-protection state of the DB is released (steps S6, S9, and S7).

If the system is operated for a long time and no problem occurs, the active SMP package is copied to the standby SMP to duplicate the SMP again (step S8).

As described above, the standby loading system requires the operator to manually release the write-protection state of the DB. Therefore, if the DB-protection state of the DB is not released by the operator's mistake, the application continuously attempts to write or executes a task related to the DB. The impossibility of performing the exchange makes it impossible to operate the exchange at all.

In addition, even if the operator releases the DB's write protection status normally, the application releases the database to the DB until the write protection status of the DB is released after switching because the release time is after all versions of the package have been run. There is a problem in that a write request is required, which causes an overload of the exchange or a specific application cannot properly handle the DB write prohibition error.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and an object thereof is to improve the stability of the exchange by allowing the loading block responsible for standby loading to automatically monitor the state of the processors to release the write prohibition of the DB.

1 is a block diagram of a conventional standby loading system.

2 is a flowchart for explaining the operation of the conventional standby loading system.

3 illustrates a PLD structure mismatch of a processor in a conventional standby loading system.

4 is a block diagram of a standby loading system according to the present invention;

Figure 5a is a flowchart for explaining the operation of the standby loading system according to the present invention.

5B is a flowchart for explaining the operation of the state monitoring processing unit according to the present invention;

5C is a flowchart for explaining the operation of the SMP located state monitoring processing unit according to the present invention;

5D is a flowchart for explaining the operation of the state monitoring processing unit located in the MP according to the present invention;

Explanation of symbols on the main parts of the drawings

10-A: SMP in active state 10-B: SMP in standby state

11-A: active SLH 12-A: active PLH

13-A: Active state monitoring processing unit 20-A: Active state MP

20-B: MP 21-A in standby state: PLH in active state

22-A: Active state monitoring processor 30: HDD

31: ROOT 32: SBY

A feature of the present invention for achieving the object as described above, in the standby loading of the exchange system, the process of the state monitoring processing unit of the management processor that is started when the processor is switched to check the switching state of all the processors in the system; A process of releasing a write prohibition state of the DB when the transfer state of all the processors is normal, transmitting a package execution signal to the management processor to execute the package, and starting the operator interface block when the transfer state is abnormal; The present invention provides a method for automatically releasing a database during standby loading of an exchange system.

Herein, the switching state check may include: generating, by the state monitoring processor of the management processor, a list of all processors; Requesting a loading directory of a package loaded by a processor from a state monitoring processor located at each processor at a predetermined time after initializing a response reception state from the listed processors; If there is a response from the status monitoring processor located in each processor within the predetermined time, recording the received package loading directory; if there is no response, abnormally recording the transfer state of the processor and requesting the loading directory again in the next cycle Wow; Recognizing a transfer state as normal when the received package loading directory is identical for all processors, and requesting a package loading directory at regular intervals by initializing the response receiving state when the received package loading directory is identical. It is done.

The package loading directory response may include: checking, by a state monitoring processor located at each processor, whether there is a request of a package loading directory from a state monitoring processor of the management processor at a predetermined time period; Responding to a package loading directory received from the management processor when a package is loaded, when the package loading directory is requested; When there is no request of the package loading directory, the state of the database is examined to check whether there is a request of the package loading directory in the next period, and if the database is in the write-protected release state, the database is automatically terminated.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Standby loading system of the exchange according to the present invention is a redundant SMP (10-A, 10-B) and a plurality of redundant MP (20-A, 20-B) and HDD ( 30).

Hereinafter, since the duplicated processor has the same structure, the processor 10-A and 20-A which are active will be described. The configuration of the processors 10-B and 20-B in the standby state, which is not shown in the drawing, will be described. The block will be described by substituting the sign (-A) of the corresponding component block which is active and replacing the sign (-B) indicating that it is in a standby state.

SMP (10-A) includes the SLH (11-A) that performs the loading server function, the PLH (12-A) that performs the loading client function, and monitors the status of all processors in the system to release the write-protected state of the DB. The state monitoring processing unit 13-A is mounted.

The plurality of MPs 20-A are equipped with a PLH 21-A that performs a loading client function, and a state monitoring processor 22-A that monitors the state of the MP. The SMPs 10-A are connected to the IPC. Connected.

The HDD 30 is composed of a ROOT 31 and an SBY 32, which are package loading directories that are logically divided into functions during standby loading, and are connected to the SMP 10-A by SCSI.

The operation of the standby loading system having the above configuration will be described in detail with reference to FIG. 5A.

Since the standby loading method of the SMP and the MP is the same, it will be described based on the SMP.

First, the active SLH 11-A writes a new version of the package into the SBY 32 of the HDD 30 (step S10).

Then, the synchronization with the standby SMP 10-B is stopped so that duplication is terminated, and the SMP 10-A to which it belongs is allowed to continue active operation independently (step S11).

Thereafter, the new version of the package recorded in the SBY 32 is loaded into the SMP 10-B operating in standby (step S12). Then, the old version of the package is loaded into the SMP 10-A that is currently active, and the new version of the package is loaded into the standby SMP 10-B and is ready to be driven.

At this time, the SLH 11-A stores directory information of the HDD 30 in which the package to which the SMP 10-B or the MP 20-B is loaded is stored. -B) to the state monitoring processor (13-B, 22-B). This package loading directory is used later to determine if a package is old or new.

Then, the SMP 10-A which has been switched to the active state by switching the processor is shifted to the standby state, and the SMP 10-B which is in the standby state is shifted to the active state to drive the loaded new version package (step S13). ).

Then, the PLH 12-B of the SMP 10-B that has transitioned to the active state makes the DB to be used by the new version of the package write-protected to ensure the integrity of the DB at the moment of switching, and the status monitoring processor 13-B. B) is started (steps S14, S15).

Therefore, if the state monitoring processor 13-B checks the states of the processors and checks that the states of all the processors are normally operated and the transfer is normal, the state monitoring processor 13-B releases the write protection state of the DB and transmits a package execution signal to the package. Allow all programs to run, then exit automatically (step A).

On the other hand, the PLH 12-B which has set the DB to write-protected and starts the state monitoring processing unit 13-B transmits the state monitoring processing unit 13-B within the corresponding time after setting a predetermined time (TMR 1). It is confirmed whether or not a package execution signal is received (step S16).

As a result of the confirmation, when the package execution signal is received, the package is executed (step S17). When the package is normally executed, the package of the SMP 10-B in operation is transferred to the SMP 10-A in standby state by switching ( Dual Copy) to duplicate the SMP (step S18).

However, if the package execution signal is not received within a predetermined time, the operator interface block (not shown) is started to warn the operator of the processor state mismatch, and then waits for the reception of the package execution signal again (steps S16 and S19). ).

5B is a flowchart for explaining the operation (step A) of the state monitoring processing unit according to the present invention.

The state monitoring processing unit 13-B started up by the PLH 12-B confirms whether or not the processor is switched (step S20).

This is to distinguish the state monitoring processing unit 13-B because it is started even when the processor is restarted because a problem occurs during the switching in addition to the moment when the standby state of the processor is active by the processor switching.

As a result of the check, if the processor is not switched, it is checked whether it is in a standby loading state, whether the processor is restarted, and whether the DB is in the write-protect state (steps S22, S23, S24).

This is because, when the processor is restarted, the operator may set the DB to be write-protected even when the processor is not being loaded. In this case, the status monitoring processor 13-B should not automatically release the write-protection state of the DB.

Accordingly, the state monitoring processor 13-B determines whether the processor to which the processor belongs is the SMP 10-B if the processor is transferred or not transferred, but the processor is restarted, standby loading, and the DB is write-protected. It is checked whether or not it is B) (step S21).

The state monitoring processing unit 13-B of the SMP 10-B creates a list of the MPs 20-B in the exchange as shown in FIG. 5C (step S25), and sets the response reception state to 'NO'. After initializing to ', set a certain time (TMR 2) to the status monitoring processing unit 22-B of all the MP (20-B) in the period of the time (TMR 2) loading directory of the package currently loaded in each processor Information is requested (steps S26, S27).

Then, it is checked whether or not there is a response to the package loading directory from the state monitoring processing unit 22-B within a predetermined time TMR 2 (step S28).

As a result of the check, when a response to the package loading directory is received from the state monitoring processing unit 22-B of the MP 20-B that has requested the package loading directory information within a predetermined time, the received package loading directory is recorded (step). S29), if there is no response, the transfer state of the MP is abnormally recorded (step S31).

Thereafter, the above steps (steps S28, S29, S31) are repeated until the package loading directory of all the MPs 20-B is recorded, and when the package loading directory of all the MPs is finished, all recorded package loading directories are written. It is checked whether it matches (step S32).

As a result of the above check, if the package loading directories of all the processors match, it is recognized that the switchover has been normally performed, and after the release of the DB write protection state (step S33), the package execution signal is transmitted to the PLH 12-B and itself ( 13-B) ends (step S34).

However, if the package loading directory is inconsistent, the response reception state is initialized and the information on the package loading directory is requested to the state monitoring processing unit 22-B of all the MPs 20-B again (steps S32, S26, S27). .

At this time, if the status monitoring processing unit 13-B fails to transmit the package execution signal due to a mismatch of the package loading directory, the operator is operated by the PLH 12-B waiting for reception of the package execution signal at a predetermined time (TMR 1). If the interface block is started up (steps S16 and S19), and the switching of the corresponding MPs is performed by the operator, and the package loading directories of all the MPs match, the state monitoring processor 13-B releases the write-protection state of the DB (steps S16). S33) The package execution signal is transmitted to the PLH 12-B (step S34), and the PLH 12-B executes the package (step S17).

On the other hand, the state monitoring processing unit 22-B of the MP 20-B is the state monitoring processing unit 13-B of the SMP 10-B within a predetermined time TMR 2 as shown in FIG. 5D. From step S35, it is checked whether or not there is an information request for the loading directory of the package loaded in the MP.

If there is a corresponding request as a result of the checking, the package loading directory received from the SLH 11-A is read during standby loading, and a corresponding response signal is transmitted to the state monitoring processing unit 13-B of the SMP 10-B ( Steps S36 and S37, if there is no request, the DB is checked to see if it is in the write-protect state (steps S38 and S39).

Therefore, when the write prohibition state of the DB is released, that is, when the state monitoring processing unit 13-B of the SMP 10-B determines the switching state of the processors and automatically releases the write prohibition of the DB, the state monitoring is no longer performed. Since the function of the processing unit 22-B is unnecessary, it terminates by itself, and if the DB is in the write-protected state, it is returned from the state monitoring processing unit 13-B of the SMP 10-B at the next cycle of the predetermined time (TMR 2) again. It is checked whether there is an information request for the loading directory of the package loaded in the MP (steps S39 and S35).

As described above, the state monitoring processor 13-B of the SMP 10-B automatically releases the DB write prohibition after confirming the transfer state of the processors. This will be described in detail.

During standby loading, each processor goes through the states of Normal, Standby Loading, Normal_Standby_Old, Normal_Standby_New, Abnormal, and Dualcopy.

Normal is the state in which the processor is operating normally, Standby Loading is loading the new version of the package into memory, and Normal_Standby_Old is the normal state but it is not running but the old version is loaded.

In addition, Normal_Standby_New is normal, but not running, and a new version of the package is loaded, Abnormal is a state in which the processor is not operating abnormally, and Dualcopy is copying the memory value of the active processor to the standby processor. .

Of these, processor switching is possible only when the processor state is Normal, Normal_Standby_Old, and Normal_Standby_New.

Table 1 below shows the processor switching state according to the standby state of the processor A and the processor B after the transfer, and Table 2 shows the processor A after transfer collected by the state monitoring processor 13-B according to Table 1 above. And a processor switching state according to the loading directory of the active side of the processor B.

Index Processor A Processor B Transfer One Normal Normal normal 2 Normal Normal_Standby_Old Conditional normal 3 Normal Normal_Standby_New Conditional normal 4 Normal_Standby_Old Normal Conditional normal 5 Normal_Standby_Old Normal_Standby_Old normal 6 Normal_Standby_Old Normal_Standby_New abnormal 7 Normal_Standby_New Normal Conditional normal 8 Normal_Standby_New Normal_Standby_Old abnormal 9 Normal_Standby_New Normal_Standby_New normal

Index Processor A Processor B Transfer One New / old New / old normal 2 New New normal Old New abnormal 3 New Old abnormal Old Old normal 4 New New normal New Old abnormal 5 New New normal 6 New Old abnormal 7 Old New abnormal Old Old normal 8 Old New abnormal 9 Old Old normal

Here, the active sides of the processor A and the processor B are always Normal, and if not, the switching state is always abnormal regardless of the condition. And, in the case of conditional normal among the transfer state of Table 1, it means that the determination result varies depending on the situation.

For example, if the standby sides of Processor A and Processor B are in Normal, Normal_Standby_Old state after switching (Index 2), respectively, the active side of Processor B currently has a new version of the package, According to whether the old version package or the new version package is loaded, the state monitoring processor may determine whether the transfer state is normal or abnormal.

As described above, the state monitoring processor determines the switching state of all processors by matching the package loading directory transmitted by the in-processor state monitoring processor in the active state in response to the state of the processor in the standby state after the changeover, and writing the DB. Release the ban.

In addition, the embodiments according to the present invention are not limited to the above-described embodiments, and various alternatives, modifications, and changes can be made within the scope apparent to those skilled in the art.

As described above, the present invention by automatically monitoring the status of the processor at the same time after the standby loading and release the write-protected state of the DB, it is possible to ensure the stable running of the package and the stability of the exchange service after the transfer, It is effective to prevent the failure of the exchange by mistake.

Claims (3)

A standby loading of an exchange system, the method comprising: checking, by a state monitoring processor of a management processor, that is started when a processor is switched, a switching state of all processors in the system; A process of releasing a write prohibition state of the DB when the transfer state of all the processors is normal, transmitting a package execution signal to the management processor to execute the package, and starting the operator interface block when the transfer state is abnormal; Automatic release of the database at the time of standby loading of the exchange system comprising a. The method of claim 1, The switching state check may include: generating, by the state monitoring processor of the management processor, a list of all processors; Requesting a loading directory of a package loaded by a processor from a state monitoring processor located at each processor at a predetermined time after initializing a response reception state from the listed processors; If there is a response from the status monitoring processor located in each processor within the predetermined time, recording the received package loading directory; if there is no response, abnormally recording the transfer state of the processor and requesting the loading directory again in the next cycle Wow; Recognizing a transfer state as normal when the received package loading directory is identical for all processors, and requesting a package loading directory at regular intervals by initializing the response receiving state when the received package loading directory is identical. Automatic release of database during standby loading of the exchange system. The method of claim 2, The package loading directory response may include: checking, by a state monitoring processor located at each processor, whether there is a request of a package loading directory from a state monitoring processor of the management processor at a predetermined time period; Responding to a package loading directory received from the management processor when a package is loaded, when the package loading directory is requested; When there is no request of the package loading directory, the state of the database is examined to check whether there is a request of the package loading directory in the next cycle, and if the database is in the write-protected release state, the database is automatically terminated. How to automatically release the database on standby loading.