JP3335697B2 - Checkpoint journal acquisition method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a checkpoint system for speeding up the collection of a checkpoint journal in an online transaction system, and quickly searching for a recovery start position in the event of a system failure to prevent a delay in system processing. Regarding the journal acquisition method.

[0002]

2. Description of the Related Art For example, an automatic teller machine at a bank or a POS
In a record related to input / output at a terminal, an online transaction system is used for recovery for performing recovery processing when a failure occurs. For example, in a case where cash cannot be withdrawn due to a cashier, the online transaction system performs a commit process (e.g., when cash cannot be withdrawn due to a cashier's instruction, if information received from the cashier is not output, despite an instruction from the cashier to withdraw cash). COMMIT processing) cannot be performed.

[0003] Acquisition of a checkpoint journal in the recovery of the online transaction system is performed by acquiring a journal and detecting a transaction with the oldest write position of a journal file among transactions currently being executed. Therefore, in obtaining the checkpoint journal, it is necessary to check all control blocks (control data) of all transactions.

In addition, the write cycle of the checkpoint journal is based on the COMM of some transactions.
In general, a system is set based on the number of transactions to be acquired when the IT processing is completed.

[0005]

However, conventionally, since the control blocks of all transactions are checked for checkpoint acquisition, it takes a long time to acquire the checkpoints. A large amount of time is required for the recovery time when a system failure occurs, and there is a problem that the processing of the system recovery is delayed.

[0006] The present invention has been made in view of such a conventional problem, and an object of the present invention is to increase the speed of collecting checkpoint journals and to reduce system downtime.
System recovery to prevent an increase in the amount of journals used
It is an object of the present invention to provide a checkpoint journal acquisition method for preventing a delay in the processing of data.

[0007]

In order to achieve the above object, the present invention provides a checkpoint journal writing frequently.
Cycle value and jar, which are the setting values of the size indicating the degree
Each time a null write is made, depending on the write journal size
A journal control information section having an accumulated size to be added , a transaction control block having control information of a transaction being executed, and a transaction control block having the oldest journal file information by connecting the transaction control block to the bottom when an input message journal is obtained. and a journal queue coupled in the order of the transaction control block, against the input message journal
Output message or transaction commit
When a transaction comes out of the journal queue
Remove journal control block and output journal file information
Message journal write position or transaction
Means for changing the position of the update image journal of the journal and re-associating it with the bottom of the journal queue; and, when the integrated size of the journal control information section becomes equal to the cycle, the transaction control block coupled to the head of the journal queue Take out the relevant transac
Jar included in the control information of the application control block
And a checkpoint journal writing means for writing null file information as a checkpoint journal .

[0008]

With the above arrangement, when the input message journal is acquired, the transaction control blocks having the journal control information of the transaction being executed are joined to the bottom, and the transaction control blocks having the oldest journal in the journal queue are joined in order. Is done. Then, the output message of the input message journal
When a message or commit is input, the transaction control block is removed from the journal queue and the journal is deleted.
Writes journal file information to output message journal
Update image jar of the position or transaction
Null position and rejoin to the bottom of the journal queue. By checkpoint write means, it retrieves the transaction control block integration size of the journal control information unit having an accumulated size of the cycle and the journal writing indicating the journal writing quantity is coupled to the top of the equal and the journal queue cycle Since checkpoints are written, the collection of checkpoint journals can be performed at high speed, and the journals used for recovery in the event of a system failure
Prevent delay in system recovery processing
it can.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of a checkpoint journal acquisition method according to the present invention.

In FIG. 1, a transaction control information table 1 includes a journal control information (JLDT) 3, an execution queue (ACT) 5, and a journal queue (TRJ).
Q) 7 to manage various information necessary for transaction processing.

The journal control information 3 is a checkpoint
This is the size setting that indicates the frequency of journal writing
Cycle 9 and each time a journal is written
This is the total size added according to the journal size
It has a SUM 11 and stores various types of control information of the journal system. Journal Write Task 13
Is the message, journal of all transactions
This task gets the journal of a file .

The execution queue 5 has information on all transactions being executed, and when a transaction is entered, a transaction control block, which will be described later, is linked to the bottom.

In the journal queue 7, transaction control blocks to be described later are linked in the order of transactions having the oldest journal. The transaction control blocks 15a to 15d linked to the journal queue 7 and the execution queue 5 have control information of the transaction being executed, and have the transaction control block 15a linked to the head of the queue.
Is set to the journal information 17 which is information at the head of the journal queue 7 by the journal writing task 13. The journal files 19a and 19b are actually updated by the journal writing task 13.

Next, the flow of a transaction will be described with reference to the configuration diagram of FIG. 2 for the entire checkpoint journal acquisition method.

First, in the input message journal (IMJ) journal acquisition and transaction (TX) message input shown in A in the figure, when a message is input, the journal writing task 13 activates the business program 21 to start the journal file. 25 stores the input message at a predetermined address. When the execution of the business program 21 is completed, the TX COMM shown in FIG.
IT processing is performed. By the above COMMIT processing, the journal writing task 13 sets the journal file 2
5, a journal is acquired (data lock is released), and a request is made to actually update the terminal and database shown in C in FIG.
When an actual update is requested from the journal writing task 13, the database writing task (DWTR) 27 actually updates the databases 29a to 29c, and the terminal control (TCOM) 31 sends an output message to the terminal.
I do. When the actual update and transmission by the database writing task 27 and the terminal control 31 are completed, the databases 29a to 29c and the terminal 33 shown in FIG.
Will output a clean point.

Next, the operation of this embodiment will be described with reference to the flowcharts shown in FIGS. Note that FIG.
The flowcharts shown in (a) to (c) operate asynchronously.

First, the flow of transaction (TX) execution will be described with reference to the flowchart of FIG. When the transaction is input, the journal writing task 13 creates a transaction control block and associates it with the bottom of the execution queue 5 (steps 100 to 100).
110). Journal writing task 13 when the transaction input message Journal (IMJ) acquisition mode, writing a journal of the journal information
The serial number and logical position of the file are linked to the transaction control block (steps 120 to 130). After that, when the transaction is executed and the journal is acquired by the COMMIT processing to acquire the output message journal ( OMJ ), the transaction control block is reconnected to the bottom of the journal queue 7. On the other hand, if the input message journal has not been acquired , it is linked to the bottom of the journal queue 7 (steps 140-1).
80).

Next, the clean point acquisition process is shown in FIG.
This will be described with reference to the flowchart of FIG.

After the execution of the transaction is completed, the journal writing task 13 executes the database 29a-29.
c, and the confirmation of the message to the user of the terminal 33 is obtained. Thereafter, when all the clean points are output, the journal writing task 13 removes the transaction control block from the execution queue 5 and the journal queue 7 (Steps 200 to 22).
0).

Further, the checkpoint acquisition processing is shown in FIG.
This will be described with reference to the flowchart of FIG.

The journal write task 13 counts the write journal size into the SUM 11 of the integrated size of the journal control information every time the journal is written, and the integrated size is used as the check point write cycle 9
When the value becomes equal to the value of the journal control block, the journal control block is removed from the head of the journal queue 7 (steps 300 to 31).
0). After the removal, the checkpoint writing is requested to the journal writing task 13 (step 32).
0).

[0023] because it tied the oldest transaction control block to the bottom of di Yanarukyu 7, as well as to the high speed the collection of checkpoints journal, journal to be used for recovery in the event of a system down
An increase in the volume can be prevented, and a delay in the recovery processing of the system can be prevented.

[0024]

As described above, according to the present invention, the output message of the input message journal or the COMM
When the IT is issued, the transaction control block is removed from the journal queue and reconnected to the bottom of the journal queue. When the integrated size becomes equal to the cycle, the transaction control block connected to the head of the journal queue is taken out. Since the checkpoint is written, the speed of collecting the checkpoint journal can be increased.

[Brief description of the drawings]

FIG. 1 is a configuration diagram illustrating an embodiment of a checkpoint journal acquisition method according to the present invention.

FIG. 2 is a configuration diagram showing a flow of a transaction in a checkpoint journal acquisition method.

FIG. 3 is a flowchart showing the operation of the present invention.

[Explanation of symbols]

 1 Transaction Control Information Table 3 Journal Control Information 5 Execution Queue 7 Journal Queue 9 CYCLE 11 SUM 13 Journal Write Task 15a to 15d Transaction Control Block

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-62-121555 (JP, A) JP-A-64-28750 (JP, A) JP-A-4-241644 (JP, A) JP-A-5-241 2517 (JP, A) JP-A-5-12087 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06F 11/34 G06F 12/00

Claims (1)

(57) [Claims] Claims: 1. Checkpoint journal write frequency
Cycle value and jar, which are the setting values of the size indicating the degree
Each time a null write is made, depending on the write journal size
A journal control information section having an accumulated size to be added ; a transaction control block having control information of a transaction being executed; and an oldest jar by connecting the transaction control block to the bottom when the input message journal is obtained.
A journal queue joined in the order of transaction control blocks having null file information, and an output message for the input message journal
Or when the transaction commits
Transaction control block from journal queue
Remove and output journal file information Message jar
Update position of null write position or transaction
Over change in the position of the di-journals, extraction means for re-coupled to the bottom of the journal queue, the transaction control block being coupled to the top of the the cumulative size of the journal control information unit is equal to the cycle the journal queue, the tiger
Included in the control information of the transaction control block
Checkpoint journal with journal file information
A checkpoint journal writing means for writing as a checkpoint journal;