JPH07219828A - Journal block restoring system and method thereof - Google Patents

Abstract

PURPOSE:To improve the restoration processing efficiency of a defective journal block. CONSTITUTION:In a normal state, a data base access means 12 receives an updating request of a data base 13 from a data base updating program 11 and outputs a journal to a journal file 19 via a journal file access means 14 before the base 13 is updated. In this case, the journal is written in two continuous blocks (master and copy blocks). A final journal block positioning means 16 sets a position to the final master block in a journal file 19 when a fault of a journal block is detected. A defective journal block detector means 17 detects the fault of the master block by a prescribed method. A journal block restoring means 15 restores the defective master block based on the contents of a copy block following the master block.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a journal block recovery method and a journal block recovery method when a failure occurs in a journal block during writing to a journal file storing a database update history.

[0002]

2. Description of the Related Art Journal records must be output to a journal file before actual updating of a database. Therefore, I / O for the journal file
The journal file may be forcibly written even if the block is not full. In this way, I /
If the O block is written to the journal file without being full, a free space is created in the journal block, which is wasteful. In order to prevent such waste of journal block, even if the journal block that has been written once is not full, a journal record is added to the empty area of the same block and the journal block is written again. . However, in this method, since the block that has been written once is written again, if a failure such as host down occurs during the second and subsequent writing, the block that has been written once may disappear.

Therefore, in the prior art, journal records of the same image are written in two consecutive blocks (hereinafter, the first block is referred to as a master block and the second block is referred to as a copy block) to fill the blocks. The output is moved to the next block (the copy block and the next block) at the time when (there is no failure due to rewriting). With this method, the image is guaranteed in the other block even when the above-mentioned failure occurs.

[0004]

In the above-mentioned conventional journal block recovery method, after recognizing a failure in a journal block, a user provides means for recovering a failed master block with a copy block in which no failure has occurred. It had to be started up and restored, which took time to restore.

An object of the present invention is to detect a journal block in which a failure has occurred when the journal file is opened and to automatically recover the journal block, so that the user recognizes the failure of the journal file and activates the recovery program. There is no need to enable the use of journal files.

[0006]

According to a first aspect of the present invention, prior to updating a database, when outputting a journal record to a journal file, the same journal record is two consecutive journal blocks and is the first journal. The block is a master block. The journal block having the next journal block as a copy block is output to the journal block, and when the master block is full, the next copy block is the master block and the next journal block is the copy. In the journal block recovery method when a failure occurs in the journal block in the database system that outputs the same journal record as a block, a journal block that detects the failure of the journal block File access means, a final journal block positioning means for positioning the final master block in the journal file when the failure is detected, and a failed journal block for detecting a failure of the master block by a predetermined method The present invention is characterized by including a detecting unit and a journal block restoring unit for restoring the failed master block according to the contents of the copy block next to the failed master block.

In addition, in the second invention, prior to updating the database, when the journal record is output to a journal file, the same journal record is two consecutive journal blocks, and the first journal block is the master block. The next journal block is output as the copy block to the journal block, and when the master block is full, the next copy block is the master block, and the next journal block is the copy block. In the method of recovering a journal block when a failure occurs in the journal block in the database system that outputs the journal record of the The failure of the final master block in the file is detected by a predetermined method, and the failed master block is restored by the contents of the copy block next to the failed master block. And

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of a journal block recovery system of the present invention.

In this embodiment, a database access means 12 for inputting / outputting a database, a database updating program 11 for updating a database, a journal file 19 for storing the update history of the database, and an input / output for the journal file are provided. The journal file access means 14 to perform and the continuous 2 of journal files
A journal block double-writing unit 18 for outputting the same image to a block, a fault journal block detecting unit 17 for detecting a fault in a journal block, and a final journal block positioning unit 16 for positioning the final journal block in a journal file. The journal block recovery unit 15 recovers the failed journal block.

The feature of the present invention is that, in FIG. 1, in the final journal block (master block ... See the prior art) locating process when a journal file is opened, it is checked whether or not a fault has occurred in the journal block. If it occurs, it means to automatically recover from one normally written journal block (copy block ... See the prior art). With this method,
The user can use the journal file without being aware of the journal block failure. It should be noted that the failures here are mainly assumed to be failures such as host down and intermittent hard failures, excluding fixed failures in hardware.

During normal operation, the database update program 1
In response to the update request for updating the database 13 by 1, the database access unit 12 precedes the database update.
The journal is output to the journal file 19 via the journal file access means 14. In this case, the journal output method is similar to the conventional technique in that the journal of the same image is written in two consecutive blocks (master block and copy block), and when the master block is full, the next block (copy block and the next block) is written. Block) to the output. Journal file 19
After outputting the journal to the database access means 12, the database access means 12 updates the database 13.

Next, the operation of the present invention when an abnormality occurs when outputting the master block to the journal file 19 will be described with reference to FIG.

First, the final journal block locating means 16 activated by the journal file access means 14 that has detected an abnormality locates the journal block output at the very end (step 21).

Next, the fault journal block detecting means 1
7 is activated, and it is checked whether or not there is a failure in the located final block (step 22).
Whether or not a block has a failure is determined by, for example, matching the block header with the trailer value.

If a failure has occurred, the journal block recovery means 15 is activated and the recovery is performed by updating the failed block or master block from the recovery next block or copy block (step 23).

After the recovery of the faulty block, the database file access means 14 determines the journal record in the middle of updating by the backward read of the journal, and the database access means 12 restores the database.

[0018]

As described above, according to the present invention, when it becomes necessary to recover a database by using a journal, it is automatically performed without being aware of whether a journal block has failed in the previous journal output. Since the recovery is performed on the first time, there is an effect that the recovery processing of the failed journal block is made efficient.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a flow chart showing the flow of processing of this embodiment.

[Explanation of symbols]

 11 Database Update Program 12 Database Access Means 13 Database 14 Journal File Access Means 15 Journal Block Restoring Means 16 Final Journal Block Positioning Means 17 Fault Journal Block Detecting Means 18 Journal Block Double Writing Means

Claims (2)

[Claims] 1. When outputting a journal record to a journal file prior to updating a database, the first journal block is a master block and the next journal block is the first two journal blocks which are consecutive journal records of the same journal record. Output to the journal block to be a copy block, and when the master block is full, the next copy block is the master block, and the next journal block is the copy block, and the output of the same journal record is made. In the journal block recovery method when a failure occurs in the journal block in the database system, a journal file access means for detecting a failure in the journal block, and the failure Is detected, the final journal block positioning means for positioning the final master block in the journal file, the failure journal block detection means for detecting a failure of the master block by a predetermined method, and the occurrence of a failure A journal block restoration method comprising: journal block restoration means for restoring the master block according to the contents of the copy block next to the failed master block. 2. Prior to updating the database, when outputting a journal record to a journal file, the first journal block is a master block and the next journal block is the two consecutive journal blocks of the same journal record. Output to the journal block to be a copy block, and when the master block is full, the next copy block is the master block, and the next journal block is the copy block, and the output of the same journal record is made. In the method of recovering a journal block when a failure occurs in the journal block in the database system, when the failure of the journal block is detected, the last file in the journal file is deleted. A journal block recovery method, wherein a failure of a star block is detected by a predetermined method, and the failed master block is recovered by the content of the copy block next to the failed master block. .