JPS61139847A - Trouble range localizing method of program - Google Patents

Abstract

PURPOSE:To facilitate explicating trouble causes by managing the execution of an application program for every transaction and preserving the state at the start time of each transaction. CONSTITUTION:When an application program 2 is executed, a transaction management part 3 discriminates the first transaction 6 of the application program 2 and starts it, and register values, etc. at the start time of the transaction 6 are stored in a buffer 10 to preserve the start at the start time of the transaction 6. If the transaction 6 is terminated normally, this normal end is reported to the management part 3. Data, updated data, etc. to be delivered to a transaction 7 to be started next are stored in the buffer 10. If the transaction is terminated abnormally on the way, a transaction recovery part 4 is started, and register values, etc. at the start time which are stored in the buffer 10 are transferred, and required data are transferred form an auxiliary storage device 12, and the program is restored to the state at the start time.

Description

[Detailed Description of the Invention] (Industrial Application Field) The present invention relates to a method for localizing the range of a fault in a program, and in particular, it enables easy reproduction of the process that led to the abnormal termination of an application program, and a method for tracing the fault. The present invention relates to an improvement in a management method for an information processing system so that data can be collected effectively and quickly.

[Conventional technology]

When an application program being executed in an information processing system terminates abnormally, the contents of the memory at the time of the abnormal termination have been acquired, but the extent of the failure cannot be localized.

Therefore, if you try to track down the process that led to the abnormal termination, the only way is to re-run the application program from the beginning, which requires a lot of effort. is extremely difficult.

Furthermore, since there is currently no function to perform a retry, even if a job ends abnormally, it may be possible to return to normal operation by retrying, but the opportunity to do so is being missed.

[Problem that the invention seeks to solve]

As described above, conventional information processing systems do not have effective means for dealing with abnormal termination.

[Means for solving problems]

The present invention aims to solve the above problems. Therefore, in the present invention, first, each application program is divided into a plurality of transactions. A delimiter command is added to the beginning and end of each transaction. The operating system provides transaction management means, transaction recovery means, and transaction retry means. When executing the application program, the transaction management means identifies each transaction and saves the data to be delivered to the transaction to be executed and the register values at the start of the transaction. When an abnormal termination occurs, the transaction recovery means restores the abnormally terminated transaction to its starting state using the above-mentioned saved data and register values.

Thereafter, the transaction retry means retries the transaction from the beginning, and stores trace data such as data updated within the transaction in a predetermined buffer.

[Effect]

Manage application program execution for each transaction,
By saving the state at the start of each transaction, in the event of an abnormal termination, it is possible to localize the failure range, and moreover, it is possible to restore the abnormally terminated transaction to the state at the start. Therefore, only the transaction that ended abnormally can be retried without having to re-execute the transaction that ended normally. By identifying and localizing the fault range to be retried in this way, trace data can be acquired quickly and effectively, and the cause of the fault can be easily clarified.

〔Example〕

The figure is a block diagram of main parts showing the configuration of an information processing system used in an embodiment of the present invention. In the same figure, ■ is the operating system, 2 is the application program, and 3 to 5 are
are newly provided parts in this embodiment, 3 is a transaction management section, 4 is a transaction recovery section, 5 is a transaction retry section, 6 to 9 are transactions constituting the above application program 2, and 10 and 11 are buffers. Memory 12 indicates an auxiliary storage device (DASD).

In executing the application program 2 in the system configured as described above, the transaction management unit 3
identifies transaction 6 at the beginning of application program 2, starts it [arrow ■ in the figure], and first stores register values etc. at the start of transaction 6 in buffer 10 [arrow ■ in figure].
Save the starting state of .

If data is updated during the execution of transaction 6, this updated data is also sequentially stored in buffer 10 [arrow ■ in the figure].

If transaction 6 ends normally.

The transaction management unit 3 is notified of this [arrow ■]. At this time, the data, update data, etc. handed over to the next activated transaction 7 are as follows.

It is already stored in the buffer 10.

The transaction management unit 3 transfers the updated data stored in the buffer 10 to the D ASD 12 [arrow ■
], then activates the next transaction 7 [arrow ■], stores the register value in buffer 10 [arrow ■], and transfers the data to be transferred from transaction 6 from buffer 10 [arrow ■] .

Assume that this transaction 7 ends abnormally in the middle. In this case, the transaction recovery unit 4 is activated, and according to a future command [arrow ■], the previously stored register values at the start from the buffer 10 for transaction 7 are restored to D A S D Necessary data is transferred from 12 [arrow [phase], ■], and transaction 7 is restored to its starting state.

On the other hand, the data updated and stored in buffer 10 until transaction 7 terminates abnormally is still DA
It has not been transferred to SD12. Therefore, D.A.S.
D12 maintains the state at the start of transaction 7.

Therefore, the transaction retry unit 5 starts the transaction 7 again [arrow @], executes it from the beginning, and obtains necessary data such as trace data. The acquired data is stored in the buffer 11 [arrow 0].

In this way, reproduce the abnormal termination of transaction 7,
By examining the data acquired during that time, the cause of the abnormality can be easily determined.

If transaction 7 ends normally as a result of the retry, it may be assumed that the transaction has been automatically re-executed.

As described above, according to this embodiment, the range to be retried is specified to be a narrow range. It is easy to acquire trace data and the like within a localized range in this way, and therefore it becomes possible to collect detailed data, and it becomes easy to clarify the cause of the failure.

〔Effect of the invention〕

As described above, according to the present invention, it is possible to retry only a transaction in which an abnormal path has occurred, without having to re-execute a transaction that has terminated normally. Therefore, data for elucidating the cause of the failure can be collected quickly, effectively, and easily.

[Brief explanation of drawings]

The figure is a block diagram of main parts showing the configuration of an information processing system used in an embodiment of the present invention. In the figure, 1 is an operating system, 2 is an application program, 3, 4.5 are transaction management section, transaction recovery section, and transaction retry section, 6 to 9 are transactions that constitute application program 2, respectively, and 10 and 11 are transactions that constitute application program 2. Buffer memory, 1
2 indicates DASD.

Claims (1)

[Claims] An operating system for managing the execution of an application program consisting of a plurality of sequential transactions is configured to include a transaction management means, a transaction recovery means, and a transaction retry means, and when executing the application program, The transaction management means detects the end of each transaction, stores the state at the start of the detected transaction, and when the application program terminates abnormally, the transaction that was being executed at that time is recovered. A method for localizing a failure range of a program, characterized in that the program restores the saved starting state by the means, and then causes the transaction retry means to re-execute the transaction to obtain desired trace data. .