JP5567444B2 - Program, application server device control method, application server device - Google Patents

Description

  The present invention relates to a program, a control method for an application server device, and an application server device.

  When the client makes an inquiry to the database server via the application server, for example, issues an SQL command, the SQL command may not be completed within a predetermined time. If the SQL command is not completed within a predetermined time, the execution of the application program becomes an obstacle in the client. Therefore, when the SQL command is not completed within a predetermined time, the application server performs various failure processes such as performing an error notification process for the client.

  For example, in the server processing apparatus, when the service thread receives a request from the client, the timeout time of the requested process and its own thread ID are written in the timeout table, and when the processing of the request ends, the timeout time in the timeout table And the thread ID is deleted, and the monitoring thread periodically reads the timeout time from the timeout table and compares it with the current time to detect the service thread that has not finished processing by the set timeout time. It has been proposed that when a service thread that has timed out is found, the service thread is forcibly terminated to notify the client of an error.

  In the product sales data processing system, it is possible to acquire the number of times that each application process has requested to connect to the database server periodically, and the acquired connection request number is compared with the set number of connection requests. If the number of times exceeds the set number, the database server is formed so that it is possible to determine that there is a strong probability that a failure will occur in the database server. It has been proposed to make it bootable.

JP-A-8-263325 JP 2000-122971 A

  If an inquiry to the database server, for example, an SQL command is not completed within a predetermined time, the application server issues a cancellation of the SQL command to the database server, for example. If the cancellation of the SQL command is successful, the SQL command can be forcibly terminated.

  However, if there is any abnormality in the application server, the cancellation of the SQL command may fail. When the cancellation of the SQL command fails, the SQL command cannot be forcibly terminated, and the SQL command issued thereafter may not be executed.

  An object of this invention is to provide the program of the application server which restarts a database server when the inquiry to a database server is not performed.

The disclosed program is a program that causes a computer to operate as an application server. The program includes a query step for issuing a query to a database server to a computer, an execution step for executing the query issued in the query step for the database server, and a query issued in the query step as monitoring targets. A predetermined number of queries are monitored within the overall monitoring time, which is the time for monitoring the time-out in which the query is not executed in the execution step in a predetermined period and the time for monitoring the query issued in the query step The monitoring step of restarting the database server when the server times out.

  According to the disclosed program, the database server can be restarted when an inquiry to the database server such as an SQL command is not executed due to some abnormality in the application server or the like.

It is a figure which shows an example of a structure of an application server system. It is a figure which shows an example of the monitoring table. It is a figure which shows an example of monitoring information. It is a figure which shows an example of a structure of an application server apparatus. It is explanatory drawing of a SQL monitoring process. It is explanatory drawing of a SQL monitoring process. It is explanatory drawing of a SQL monitoring process. It is explanatory drawing of a SQL monitoring process. It is a SQL process flowchart. It is a SQL registration process flowchart. It is a SQL registration cancellation process flowchart. It is a SQL monitoring process flowchart. It is a database monitoring process flowchart. It is a timeout monitoring process flowchart. It is explanatory drawing of the SQL monitoring process which this inventor examined.

  FIG. 15 is an explanatory diagram of the SQL monitoring process examined by the present inventors, and is an explanatory diagram for a case where the application server fails to cancel the SQL command.

  The application processing thread 201 requests the application timeout thread 203 to register for the monitoring process (P # 101), and then requests the SQL command execution unit 204 to execute the SQL command (P # 102). The SQL command execution unit 204 requests the database server 205 to execute the requested SQL command.

  Thereafter, the application processing thread 202 requests the SQL command execution unit 204 to execute the SQL command (P # 102 '). The SQL command execution unit 204 executes the SQL command requested to be executed on the database server 205. Prior to the request in P # 102 ', the application processing thread 202 requests the application timeout thread 203 to register for monitoring processing.

  At this time, since the execution of the SQL command does not end, the application processing thread 201 requests the application timeout thread 203 to cancel the SQL command that does not end. In response to this, the application timeout thread 203 calls a cancel processing execution unit that is a subroutine for executing the cancel processing, and the called cancellation processing execution unit attempts to cancel the SQL command to the SQL command execution unit 204. (P # 103).

  However, since the SQL command execution unit 204 is executing the SQL command from the other application processing thread 202 in response to the request in P # 102 ', the SQL command in P # 103 cannot be canceled.

  If the database server 205 is not normal, the SQL command execution unit 204 cannot cancel the SQL command requested to be executed by the database server 205.

  Thus, when the cancellation of the SQL command fails, the SQL command cannot be forcibly terminated. Further, if an attempt is made again to execute the SQL command after that, there is a possibility that the SQL command is not executed. Therefore, there is a possibility that the retry of executing the SQL command is wasted.

  The disclosed program, the control method of the application server apparatus, and the application server apparatus monitor the inquiry to the database server, for example, the timeout of the SQL command, and restart the database server when the timeout becomes a predetermined state .

  FIG. 1 is a diagram illustrating an example of a configuration of an application server system.

  The application server system includes an application server 1, a database server 2, a spare database server 2 ′, and a plurality of clients 3. The database server 2 includes a database (DB) 21, and the spare database server 2 ′ includes a spare database (DB) 21 ′. The spare database 21 ′ is a database that stores the same data as the database 21. The application server 1 is connected to the database server 2 via the network 4. The application server 1 is connected to each of the plurality of clients 3 via the network 5. The network 4 may be the same as the network 5. Further, the application server 1 and the database server 2 may be provided as one server device.

  Each client 3, in other words, the client device is a computer that makes an inquiry to the database server 2. Specifically, each client 3 transmits an inquiry request to the database server 2 to the application server 1.

  The application server 1 includes a plurality of application processing units 11, an execution processing unit 12, a monitoring processing unit 13, a monitoring information storage unit 14, and a monitoring table storage unit 15.

  The plurality of application processing units 11 are provided according to the plurality of clients 3 connected to the application server 1. Actually, the application processing unit 11 is a thread generated for each client 3 by the control program of the application server 1. The application processing unit 11 issues various queries to the database server 2 in response to various requests to the database server 2 received from the client 3. For example, the application processing unit 11 issues an SQL command to the database server 2 to the execution processing unit 12.

  In the following, a case where the query to the database server 2 is an SQL command will be described. The query to the database server 2 may be a command other than the SQL command.

  The execution processing unit 12 executes an SQL command for the database server 2 issued by the plurality of application processing units 11 to the database server 2. Actually, the execution processing unit 12 transmits an SQL command to the database server 2 via the network 4. The database server 2 executes the received SQL command and returns a response to the received SQL command to the execution processing unit 12. Specifically, the database server 2 searches the database 21 based on the received SQL command, and transmits the search result to the execution processing unit 12 as an answer to the received SQL command.

  The monitoring processing unit 13 monitors the SQL commands issued by the plurality of application processing units 11. The SQL command to be monitored is an SQL command registered in the monitoring table 151 stored in the monitoring table storage unit 15. The monitoring table 151 will be described later with reference to FIG.

  The monitoring processor 13 monitors the timeout of the SQL command. Actually, the monitoring processing unit 13 is a thread generated by the control program of the application server 1. The timeout of the SQL command is a state in which the SQL command is not executed by the execution processing unit 12 during a predetermined period. The period is a time during which the monitoring processing unit 13 monitors the execution of the SQL command, and can be determined empirically and is determined in advance.

  The monitoring processing unit 13 restarts the database server 2 when a predetermined number of SQL commands time out. Specifically, the monitoring processing unit 13 transmits a restart command to the database server 2. The database server 2 executes the received restart command. Thereby, the various control programs of the database server 2 are temporarily stopped, the OS (operating system) of the database server 2 is restarted, and the various control programs of the database server 2 are restarted when the OS is restarted. .

  During the restart of the database server 2, for example for several minutes, the client 3 cannot access the database server 2. In other words, the execution processing unit 12 cannot execute the SQL command. In practice, a spare database server 2 ′ of the database server 2 to be restarted is provided, and the client 3 can access the spare database server 2 ′ until the restart of the database server 2 is completed. It is possible. Since the spare database 21 ′ stores the same contents as the database 21, the client 3 does not actually need to be aware of the restart of the database server 2.

  The database server 2 is restarted when a predetermined number, in other words, the number of SQL commands necessary for restarting time out. The number required for restart can be determined empirically and is predetermined.

  As will be described later, the monitoring processing unit 13 includes an overall monitoring time as a parameter in the monitoring processing in the monitoring information storage unit 14. The monitoring information storage unit 14 will be described later with reference to FIG. The overall monitoring time determines the time for monitoring the SQL commands issued by the plurality of application processing units 11. When the SQL command issued by any of the plurality of application processing units 11 times out first, the monitoring processing unit 13 starts counting the total monitoring time. The monitoring processing unit 13 restarts the database server 2 when a predetermined number, in other words, the number of SQL commands necessary for restarting time out within the entire monitoring time.

  The “number necessary for restart” is, for example, the total number of SQL commands that can be issued within the overall monitoring time. The total number of SQL commands that can be issued within the overall monitoring time is determined based on the overall monitoring time and the lower limit SQL issue interval. In this case, the database server 2 is restarted when all of the SQL commands that can be issued within the entire monitoring time have timed out. Thereby, as described later, it is possible to prevent the database server 2 from being frequently restarted.

  The “number required for restart” is not limited to the total number of SQL commands that can be issued within the overall monitoring time, but may be other numbers. For example, the number calculated based on the total number of SQL commands that can be issued within the overall monitoring time, in other words, the value calculated by multiplying the total number by a predetermined ratio is set as the number necessary for restarting. Also good. Further, the number determined regardless of the total number of SQL commands that can be issued within the overall monitoring time may be set as the number necessary for restarting.

  FIG. 2 is a diagram illustrating an example of the monitoring table.

  The monitoring table storage unit 15 stores a monitoring table 151. The monitoring table 151 is basically created by the monitoring processing unit 13 by registering the SQL command requested to be registered based on the SQL registration processing request from the application processing unit 11. The monitoring table 151 stores a thread ID, a DB command, a start time, and a cancel time for each SQL command.

  The SQL command is represented by a command ID such as “SQL # 1”. The command ID is identification information that uniquely identifies the SQL command. The thread ID is identification information that uniquely identifies the application processing unit 11 that has issued the corresponding SQL command, in other words, the application processing thread. The DB command is the content of the corresponding SQL command, in other words, the description content of a SELECT statement or the like. The SQL command, thread ID, and DB command are stored in the monitoring table 151 when registration is requested based on the SQL registration processing request from the application processing unit 11.

  The start time is the time when execution of the corresponding SQL command is started, and is actually the time when the corresponding SQL command is registered in the monitoring table 151. The start time is acquired from a timer provided in the application server 1. The start time is stored in the monitoring table 151 when the application processing unit 11 requests registration of the SQL command.

  The cancel time is the time when execution of the corresponding SQL command is cancelled. The cancellation time is acquired from a timer provided in the application server 1. The cancel processing time is requested by the monitoring processing unit 13 when the monitoring processing unit 13 requests the execution processing unit 12 to cancel the process, and when the response to the request for cancellation processing is returned from the execution processing unit 12 to the monitoring processing unit 13. Stored in the monitoring table 151.

  FIG. 3 is a diagram illustrating an example of monitoring information.

  The monitoring information storage unit 14 stores various parameters in the SQL monitoring process. The monitoring information storage unit 14 includes an overall monitoring time area 141, a monitoring interval area 142, a lower limit SQL monitoring interval area 143, a lower limit SQL issue count area 144, a lower limit SQL issue interval area 145, an overall monitoring start time area 146, and an SQL timeout count area. 147, a previous SQL execution time area 148, and an SQL issue number area 149. Information stored in these areas 141 to 149 is a parameter in the SQL monitoring process.

  The overall monitoring time area 141 stores the overall monitoring time. The overall monitoring time can be determined empirically and is predetermined. The total monitoring time is a time for monitoring all the SQL commands issued by the plurality of application processing units 11, in other words, the SQL commands registered in the monitoring table 151 in the SQL command monitoring process.

  The monitoring interval area 142 stores the monitoring interval. The monitoring interval can be determined empirically and is determined in advance. The monitoring interval is an interval at which the monitoring processing unit 13 executes monitoring of the SQL command in the monitoring processing of the SQL command.

  The lower limit SQL monitoring interval area 143 stores the lower limit SQL monitoring interval. The lower limit SQL monitoring interval can be obtained empirically and is determined in advance. The lower limit SQL monitoring interval is a condition that does not contribute to restart of the database server 2 even if a predetermined number of SQL commands time out in the monitoring process of the SQL command, in other words, an exception condition for restart. Determine. Specifically, when the monitoring interval of the SQL command is smaller than the lower limit SQL monitoring interval, the number of SQL commands timed out at that time or SQL commands timed out until that time are necessary for restarting the database server 2 Not counted as. Alternatively, even when a predetermined number of SQL commands time out, the database server 2 is not restarted if the monitoring interval of the SQL commands is smaller than the lower limit SQL monitoring interval.

  The lower limit SQL issue number area 144 stores the lower limit SQL issue number. The lower limit SQL issuance number can be obtained empirically and is determined in advance. The lower limit SQL issuance number defines an exception condition for restart in the monitoring process of the SQL command. Specifically, when the number of issued SQL commands is smaller than the lower limit number of issued SQL commands, the number of SQL commands timed out at that time or SQL commands timed out until that time are required for restarting the database server 2 Not counted as. Alternatively, even when a predetermined number of SQL commands have timed out, the database server 2 is not restarted if the number of issued SQL commands is smaller than the lower limit number of issued SQL commands.

  The lower limit SQL issue interval area 145 stores the lower limit SQL issue interval. The lower limit SQL issue interval can be determined empirically and is determined in advance. The lower limit SQL issue interval is used to calculate the restartable time, as will be described later.

  The overall monitoring start time area 146 stores the overall monitoring start time. The total monitoring start time is the time at which the monitoring processing unit 13 starts monitoring the SQL command in the SQL command monitoring processing, in other words, the SQL command issued by one of the plurality of application processing units 11 first times out. It is time. Furthermore, the overall monitoring start time is the time when the cancel processing execution unit, which is a subroutine that first executes the cancel processing, is called for the SQL command issued by any of the plurality of application processing units 11. The overall monitoring start time is acquired from the cancel time of the monitoring table 151 as will be described later. The overall monitoring start time is stored in the overall monitoring start time area 146 by the monitoring processing unit 13 when the monitoring processing unit 13 starts monitoring the SQL command.

  The SQL timeout number area 147 stores the SQL timeout number. The number of SQL timeouts is the number of SQL commands that timed out in the SQL command monitoring process. The number of SQL timeouts is counted every time the SQL command times out by the monitoring processing unit 13 and stored in the SQL timeout number area 147.

  The previous SQL execution time area 148 stores the previous SQL execution time. The previous SQL execution time is the issue time of the SQL command issued last time in the SQL command monitoring process, in other words, immediately before the latest SQL command. The previous SQL execution time is the issuance of the SQL command issued immediately before the latest SQL command in the monitoring table 151 by the monitoring processing unit 13 when the monitoring processing unit 13 registers the latest SQL command in the monitoring table 151. With reference to the time, it is stored in the previous SQL execution time area 148.

  The SQL issue number area 149 stores the SQL issue number. The number of SQL issuances is the number of SQL commands issued from a plurality of application processing units 11 in the SQL command monitoring process. The number of SQL issuances is counted each time an SQL command is issued by the monitoring processing unit 13 and stored in the SQL issuance number area 149.

  FIG. 4 is a diagram illustrating an example of a hardware configuration of the application server.

  The CPU 101 controls the application server 1 according to a control program stored in the ROM 103. For example, the CPU 101 executes an application processing program, an execution processing program, and a monitoring processing program on the RAM 102 which is a main memory. Thereby, the application processing unit 11, the execution processing unit 12, and the monitoring processing unit 13 are realized. The application processing program, the execution processing program, and the monitoring processing program are stored in the recording medium 107 such as a CD-ROM or DVD, input from the recording medium 107 to the hard disk 106, and loaded from the hard disk 106 to the RAM 102. The monitoring information storage unit 14 and the monitoring table storage unit 15 are provided in the RAM 102, for example.

  The input device 104 is a keyboard, for example, and may include a mouse or the like. The output device 105 is a display, for example, and may include an output device such as a printer. The CPU 101, the RAM 102, the ROM 103, the input device 104, the output device 105, and the hard disk 106 are connected to each other via the bus 100.

  FIG. 5 is an explanatory diagram of the SQL monitoring process in the application server.

  In FIG. 5, a plurality of application processing threads 31 are generated as the application processing unit 11. An application timeout thread 131 is generated as the monitoring processing unit 13. The execution processing unit 12 is referred to as an SQL command execution unit 121.

  The application processing thread 31 receives an inquiry request from the corresponding client 3 to the database server 2. In response to this, the application processing thread 31 requests the application timeout thread 131 to register to the SQL monitoring process, in other words, registration to the monitoring table 151 (P # 1), and then execute the SQL command. The command execution unit 121 is requested (P # 2). The SQL command execution unit 121 requests the database server 2 to execute the SQL command requested to be executed.

  Thereafter, the application processing thread 32 receives an inquiry request from the corresponding client 3 to the database server 2. In response to this, the application processing thread 32 requests the application timeout thread 131 to register for the SQL monitoring process (P # 1 ′), and then requests the SQL command execution unit 121 to execute the SQL command (P #). 2 '). The SQL command execution unit 121 executes the SQL command requested to be executed on the database server 2.

  The application timeout thread 131 periodically monitors the execution of the entire SQL command requested to be registered in the SQL monitoring process by the application processing thread 31 and the application processing thread 32 (P # 3). The SQL command requested to be registered in the SQL monitoring process is referred to as a monitoring target SQL command. The monitoring target SQL command is an SQL command registered in the monitoring table 151.

  When the execution of the monitoring target SQL command does not end within a predetermined period, in other words, when the monitoring target SQL command times out, the application timeout thread 131 includes a cancel process execution unit that is a subroutine for executing a cancel process. Call (P # 4). The called cancellation process execution unit executes the SQL command cancellation to the SQL command execution unit 121.

  In the process of P # 4, the application timeout thread 131 stores the cancel time for the timed-out SQL command in the monitoring table 151. The cancellation time is acquired from a timer provided in the application server 1. The application timeout thread 131 sets the overall monitoring start date and time in the overall monitoring start time area 146 based on the cancel time stored in the monitoring table 151 for the timed-out SQL command.

  For example, when the execution processing unit 12 is not executing an SQL command from another application processing thread 32, the SQL command execution unit 121 can cancel the SQL command requested to be executed by the database server 2. When the database server 2 is normal, the execution processing unit 12 can cancel the SQL command requested to be executed by the database server 2. Even if the SQL command issued by the plurality of application processing units 11 is canceled, the monitoring processing unit 13 sets the canceled SQL command as the monitoring target SQL command.

  On the other hand, when the execution processing unit 12 is executing an SQL command from another application processing thread 32, the SQL command execution unit 121 cannot cancel the SQL command requested to be executed by the database server 2. When the database server 2 is not normal, the SQL command execution unit 121 cannot cancel the SQL command requested to be executed by the database server 2.

  As described above, regardless of whether the cancellation is successful or unsuccessful, the SQL command for which the cancel processing execution unit that executes the cancel processing is called is counted as a time-out SQL command. As a result of the above, when there is a cause for the SQL command to time out in the database server 2, as shown in FIG. 6 below, the monitoring target SQL is timed out continuously.

  FIG. 6 is an explanatory diagram of the SQL monitoring process in the application server.

  For example, SQL # 1, which is the first SQL command, is issued at time t11. However, SQL # 1 is not executed, and a timeout of SQL # 1 occurs at time t12. As a result, counting of the entire monitoring time starts, and all SQLs are monitored. For this purpose, the monitoring processing unit 13 includes an overall monitoring time counter.

  Thereafter, SQL # 2, which is the second SQL command, is issued at time t13. However, SQL # 2 is not executed, and a timeout of SQL # 2 occurs at time t14. Thereafter, a plurality of SQL commands are issued. However, all issued SQL commands are not executed, and a timeout occurs for all issued SQL commands.

  Here, it is assumed that 100 SQL commands are issued in total including SQL # 1 and # 2. In addition, in 100 SQLs, SQL # 100, which is the 100th SQL command, is finally issued at time t15 and is not executed, and a timeout of SQL # 100 occurs at time t16.

  For example, it is assumed that the overall monitoring time when counting is started from time t12 reaches a predetermined time, for example, 600 seconds at time t17. Therefore, the time-out of all issued SQL commands has occurred until time t17 when the overall monitoring time has elapsed. In other words, the entire monitoring time has elapsed, and timeouts of all issued SQL commands have occurred. Therefore, the database server 2 is restarted at time t17. Thereby, it can be avoided that the retry of the SQL command is repeated in vain.

  Thereafter, SQL # 101 is issued at time t18, and SQL # 101, the 101st SQL command, is normally executed and terminated at time t19. Furthermore, at time t110, the 102nd SQL command SQL # 102 is issued, and the number of SQL issuances is counted. However, SQL # 102 is not executed, and a timeout of SQL # 102 occurs at time t111. As a result, counting of the entire monitoring time starts, and all SQLs are monitored.

  Thereafter, SQL # 103, which is the 103rd SQL command, is issued at time t112, is normally executed, and ends at time t113. Thereby, the count of the entire monitoring time is stopped, and monitoring for all SQL is stopped.

  As shown in FIG. 6, an exception for restarting the database server 2 is provided. The monitoring processing unit 13 uses the restart exception condition stored in the monitoring information storage unit 14 so as not to count the timed-out SQL command as the number necessary for restarting the database server 2, or in advance Even when a predetermined number of SQL commands time out, the database server 2 is not restarted.

  FIG. 7 is an explanatory diagram of the SQL monitoring process in the application server.

  For example, as described above, the monitoring processing unit 13 includes a lower limit SQL issue number as an exception condition to be excluded from the case of restarting the database server 2. The lower limit number of issued SQL statements determines the number of issued SQL commands issued by a plurality of application processing units 11.

  As shown in FIG. 7, when the predetermined number of SQL commands time out within the overall monitoring time, the monitoring processing unit 13 determines in advance that the number of SQL commands that timed out is less than the lower limit SQL issuance number. The case where a predetermined number of SQL commands time out is excluded from the case where the database server 2 is restarted.

  In FIG. 7, for example, SQL # 1 is issued at time t21, and the number of SQL issuances is counted. However, SQL # 1 is not executed, and a timeout of SQL # 1 occurs at time t22. As a result, counting of the entire monitoring time starts, and all SQLs are monitored.

  Thereafter, SQL # 2 is issued at time t23, and the number of SQL issued items is counted. However, SQL # 2 is not executed, and a timeout of SQL # 2 occurs at time t24. After this, a plurality of SQL commands are issued, and the number of SQL issuances is counted each time. However, all issued SQL commands are not executed, and a timeout occurs for all issued SQL commands.

  Here, it is assumed that 50 SQL commands are issued in total including SQL # 1 and # 2. In addition, among the 50 SQLs, SQL # 50, which is the 50th SQL command, is finally issued at time t25 and is not executed, and a timeout of SQL # 50 occurs at time t26. The number of SQL issuances when SQL # 50 timeout occurs is “50”.

  For example, it is assumed that the overall monitoring time when counting is started from time t22 reaches a predetermined time, for example, 600 seconds at time t27. Therefore, the time-out of all issued SQL commands has occurred by time t27 when the overall monitoring time has elapsed. In other words, the entire monitoring time has elapsed, and timeouts of all issued SQL commands have occurred.

  However, in this case, the database server 2 is not restarted even though the overall monitoring time has elapsed and all the issued SQL commands have timed out. This is because the number of SQL issuances when the overall monitoring time has elapsed is “50”, which is smaller than the lower limit number of SQL issuances “60”.

  As a result, when the number of SQL issuances is small when the overall monitoring time has elapsed, these timeouts do not cause the database server 2 to restart. Thereby, it is possible to prevent the restart of the database server 2 from being frequently executed.

  FIG. 8 is an explanatory diagram of the SQL monitoring process in the application server.

  For example, the monitoring processing unit 13 includes a lower limit SQL issue interval. The lower limit SQL issuance interval defines a time interval for issuing SQL commands issued by a plurality of application processing units 11 as an exception condition excluded from the case of restarting the database server 2.

  As shown in FIG. 8, when the time interval for issuing the SQL commands issued by the plurality of application processing units 11 is longer than the lower limit SQL issuing interval, the monitoring processing unit 13 uses the SQL issued up to that point. The command is excluded from the monitoring target, the total monitoring time count at that time is reset, and the restart is performed from that time.

  In FIG. 8, for example, SQL # 1 is issued at time t31, and the current time t31 is set as the previous SQL execution time. However, SQL # 1 is not executed, and a timeout of SQL # 1 occurs at time t32. As a result, counting of the entire monitoring time starts, and all SQLs are monitored.

  Thereafter, SQL # 2 is issued at time t33, and the current time t33 is set as the previous SQL execution time. However, SQL # 2 is not executed, and a timeout of SQL # 2 occurs at time t34.

  Thereafter, after 70 seconds elapse from the previous SQL execution time t33, SQL # 3 is issued at time t35.

  At this time, counting of the entire monitoring time started from time t32 is stopped based on the time interval for issuing SQL. This is because the time from the previous SQL execution time t33 to the current time t35 is “70 seconds” at the latest SQL # 3 issuance time t35, which is larger than the lower limit SQL issuance interval “60 seconds”. Note that the current time t35 is set as a new previous SQL execution time after the count of the entire monitoring time is stopped.

  As a result, the SQL # 1 timeout at time t32 and the SQL # 2 timeout at time t34 do not cause the database server 2 to restart. Thereby, it is possible to prevent the restart of the database server 2 from being frequently executed.

  FIG. 9 is an SQL process flowchart in the application server.

  As described above, the application processing unit 11 requests the monitoring processing unit 13 to monitor the SQL command, in other words, to register the SQL command in the monitoring table 151. In response to this, the monitoring processing unit 13 executes an SQL registration process for registering the SQL command in the monitoring table 151 (step S11). The SQL registration process will be described later with reference to FIG.

  Thereafter, the application processing unit 11 requests the execution processing unit 12 to execute the SQL command as described above. In response to this, the execution processing unit 12 executes the registered SQL command (step S12).

  Thereafter, the monitoring processing unit 13 executes an SQL deregistration process for deleting the executed SQL command from the monitoring table 151 (step S13). The SQL registration cancellation process will be described later with reference to FIG.

  FIG. 10 is an SQL registration process flowchart.

  The monitoring processing unit 13 sets the monitoring table 151 to the exclusive state (step S21), and registers the thread, DB command, and start time in the monitoring table 151 (step S22). The monitoring processing unit 13 sets the number of SQL issuances stored in the SQL issuance number area 149 to “SQL issuance number + 1” (step S23). Thereafter, the monitoring processing unit 13 sets the monitoring table 151 to the exclusive release state (step S24).

  FIG. 11 is an SQL registration release processing flowchart.

  The monitoring processing unit 13 makes the monitoring table 151 in an exclusive state (step S31), and deletes SQL information that is information about the executed SQL from the monitoring table 151 (step S32). The monitoring processing unit 13 sets the number of SQL issuances stored in the SQL issuance number area 149 to be “the number of SQL issuances−1” (Step S33). Thereafter, the monitoring processing unit 13 sets the monitoring table 151 to the exclusive release state (step S34).

  FIG. 12 is an SQL monitoring process flowchart.

  The monitoring processing unit 13 sets the monitoring table 151 in an exclusive state (step S41), executes database monitoring processing (step S42), and further executes timeout monitoring processing (step S43). The database monitoring process will be described later with reference to FIG. The timeout monitoring process will be described later with reference to FIG. Thereafter, the monitoring processing unit 13 sets the monitoring table 151 to the exclusive release state (step S44), enters the sleep state (step S45), and waits for the monitoring interval stored in the monitoring interval area 142 to elapse. Step S41 is repeated.

  FIG. 13 is a flowchart of the database monitoring process.

  The monitoring processing unit 13 determines whether or not the overall monitoring start date / time is set in the overall monitoring start time area 146 (step S51). If the overall monitoring start date / time is not set, the monitoring processing unit 13 ends the process.

  When the overall monitoring start date and time is set, the monitoring processing unit 13 calculates the monitoring end time by adding the overall monitoring time to the overall monitoring start time (step S52). Thereafter, the monitoring processing unit 13 determines whether or not the calculated monitoring end time has reached the current time (step S53). If the current time has not reached the monitoring end time, the monitoring processing unit 13 ends the process.

  If the current time has reached the monitoring end time, the monitoring processing unit 13 further determines whether or not the number of SQL issuances is equal to or greater than the lower limit number of SQL issuances stored in the lower limit SQL issuance number area 144 (step) S54). If the number of SQL issuances is not greater than or equal to the lower limit number of SQL issuances, the monitoring processor 13 ends the process.

  When the number of SQL issuances is equal to or greater than the lower limit number of SQL issuances, the monitoring processing unit 13 sets the lower limit SQL issuance interval stored in the lower limit SQL issuance interval area 145 at the previous SQL execution time stored in the previous SQL execution time area 148. The restartable time is calculated by adding (step S55). Thereafter, the monitoring processing unit 13 determines whether or not the calculated restartable time has reached the current time (step S56). If the current time has not reached the restartable time, the monitoring processing unit 13 ends the process without executing the next step S57.

  If the current time has reached the restartable time, the monitoring processing unit 13 calls the restart script (step S57) and ends the process.

  In step S54, when the number of SQL issuances is smaller than the lower limit number of SQL issuances stored in the lower limit SQL issuance number area 144, the monitoring processing unit 13 sets the overall monitoring start date and time in the overall monitoring start time area 146 to “NULL”. (Step S58), the process is terminated.

  FIG. 14 is a flowchart of timeout monitoring processing.

  The monitoring processing unit 13 extracts one SQL command from the monitoring table 151 in the order of issue (step S61). The extracted SQL command is referred to as “SQLn”. The monitoring processor 13 calculates the end date and time by adding the lower limit SQL time to the SQLn start time (step S62). Thereafter, the monitoring processing unit 13 determines whether or not the calculated end date / time has reached the current SQLn time (step S63). If the end date / time has not reached the current SQLn time, the monitoring processing unit 13 executes step S611 described later.

  When the end date / time has reached the current SQLn time, the monitoring processing unit 13 further determines whether or not the SQLn cancellation time that is the cancellation time of “SQLn” is set in the monitoring table 151 (step S64). ). When the SQLn cancel time is set, the monitoring processing unit 13 executes step S611 described later.

  When the SQLn cancel time is not set, the monitoring processing unit 13 sets the current time as the SQLn cancel time (step S65) and calls the cancel processing execution unit (step S66).

  Thereafter, the monitoring processing unit 13 determines whether or not the overall monitoring start time is set in the overall monitoring start time area 146 (step S67). If the overall monitoring start time is not set, the monitoring processing unit 13 sets the current date and time as the overall monitoring start time in the overall monitoring start time area 146 (step S68), and sets the number of SQL issuances in the SQL issuance number area 149 to “ 1 "(step S69). When the entire monitoring start date and time is set, the monitoring processing unit 13 sets the number of SQL issuances in the SQL issuance number area 149 to “SQL issuance number + 1” (step S610).

  Thereafter, the monitoring processing unit 13 determines whether or not SQLn is the last SQL (step S611). If SQLn is not the last SQL, the monitoring processing unit 13 repeats Step S61. When SQLn is the last SQL, the monitoring processing unit 13 ends the process.

DESCRIPTION OF SYMBOLS 1 Application server 2 Database server 3 Client 4, 5 Network 11 Application processing part 12 Execution processing part 13 Monitoring processing part 14 Monitoring information storage part 15 Monitoring table storage part

Claims (8)

A program for operating a computer as an application server,
The program is stored in a computer.
A query step for issuing a query to the database server;
An execution step of executing the query issued in the query step with respect to the database server;
Time to monitor the query issued in the query step, monitoring the timeout that is not executed in the execution step in a predetermined period with the query issued in the query step being monitored And a monitoring step of restarting the database server when a predetermined number of inquiries have timed out within an overall monitoring time that is a predetermined time. 2. The program according to claim 1, wherein in the monitoring step, counting of the total monitoring time is started when the inquiry issued in the inquiry step first times out. In the monitoring step, when a time interval of issuing the query issued in the query step is longer than a lower limit query issuing interval that defines a time interval of issuing the query issued in the query step, the time point 3. The program according to claim 2, wherein the query issued until then is excluded from the monitoring target, the count of the total monitoring time at that time is reset, and restarted from that time. In the monitoring step, when the predetermined number of the queries timed out within the overall monitoring time, the number of the queries that timed out determines the lower limit query that determines the number of the queries issued in the querying step. The case where the number of issued cases is smaller than a case where the predetermined number of the queries time out within the overall monitoring time is excluded from a case where the database server is restarted. Program. The program according to claim 1, wherein, in the monitoring step, even if the inquiry issued in the inquiry step is canceled, the canceled inquiry is the monitoring target. In the inquiry step, a thread generated according to each of a plurality of clients connected to the application server issues the inquiry in response to an inquiry request to the database server received from the corresponding client. The program according to claim 1. Issuing a query to the database server;
Executing the issued query against the database server;
Monitoring the time-out in which the query is not executed in a predetermined period, with the issued query as a monitoring target;
Restarting the database server when a predetermined number of the queries time out within an overall monitoring time, which is a time for determining a time for monitoring the issued queries. Control method of application server device. Multiple application processing units that issue queries to the database server;
An execution processing unit that executes the query issued by the plurality of application processing units to the database server;
As a monitoring target issued the query by the plurality of application processing unit, the query monitors the timeout is a state that is not executed by the execution unit in a predetermined period, monitoring the query is issued An application server device comprising: a monitoring processing unit that restarts the database server when a predetermined number of the queries time out within an overall monitoring time that is a time for determining a time.

Images