JP4141680B2 - Database management system and database system management method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention includes two magnetic disks, such as the database software SolsticeDiskSiute manufactured by Sun, and the next time the normal StateDB is less than ½ of the total among the StateDBs on these magnetic disks. The present invention relates to a database management system that uses software that prevents the system from starting up at the time of rebooting and a database system management method.
[0002]
[Prior art]
SolsticeDiskSiute, the standard software for built-in disk mirroring on Sun's UNIX workstation, has 3 or more StateDBs, and when the number of StateDBs in error becomes 1/2 or more (including 1/2), the next time It is designed to prevent the system from starting up at reboot.
[0003]
This database management software SolsticeDiskSuite, when operating a database while mirroring two large-capacity hard disks, each disk has a total of six StateDBs, but if one of the disks fails, StateDB , The number of normal StateDB is only three of the other disk StateDB, and the ratio of normal StateDB is less than 1/2, so even if the remaining disk is healthy The system will not start up at the next reboot.
[0004]
Therefore, if such a failure occurs, the database administrator goes to the installation location of the system, deletes the StateDB where the error is found, and performs an operation to restart only the normal StateDB.
[0005]
Also, conventionally, if a disk failure occurs in the boot area of the current boot disk, it will not be able to start up at the next reboot, but in such a case, the administrator goes to the system installation location and a failure occurs in the boot area. The boot disk is switched from one disk to the other and the operation to restart is performed.
[0006]
[Problems to be solved by the invention]
In this way, in a conventional database management system, when one hard disk fails and the next reboot cannot be performed, the system administrator can go to the installation site of the system and perform necessary operations to reboot the system. Like that.
[0007]
In the database system using Sun's SolsticeDiskSuite, if the system cannot be rebooted due to a hard disk failure on one side as described above, use the other healthy disk as the boot disk and continue until the next periodic hardware check. It is the actual situation to operate. Therefore, as described above, the system administrator has to go to the installation site of the system to work.
[0008]
The present invention has been made in view of such a conventional technical problem. For example, a Sun State SolsticeDiskSiute is provided with two magnetic disks, and among the StateDBs on these magnetic disks, a normal StateDB is present. When operating the database system with software that prevents the system from starting at the next reboot when the total becomes less than half of the total, even if an error occurs in the StateDB on one disk, the next reboot can be performed and managed The purpose is to provide a database management technology that can reduce the work required of a person as much as possible.
[0010]
[Means for Solving the Problems]
The invention of claim 1 includes two magnetic disks, and among the StateDBs having a plurality of areas on these magnetic disks, when the normal StateDB becomes less than half of the whole, the system does not start up at the next reboot. in database management systems employing software for the operation of the database for a predetermined check Syslog every period, sign of a fault of the Syslog either StateDB region if it contains errors to the OS area of the magnetic disk There and determines a fault monitoring means for determining that there is failure of StateDB region if it contains errors for the Syslog to StateDB region, when said failure monitoring means has found the sign of the failure or disorder, the appropriate And StateDB deletion means for deleting all StateDBs on the magnetic disk.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Fig. 1 is provided with two magnetic disks. Among the StateDBs on these magnetic disks, if the normal StateDB is less than half of the total, the software made by Sun Corp. will prevent the system from starting at the next reboot. It shows a database server that uses SolsticeDiskSuite as operational software.
[0017]
This server has two magnetic disks DISK # 1 and DISK # 2. Each disk has an OS area (operating system area) including a Boot area (boot area), and three StateDBs for each disk according to the specifications (DISK # 1 is DB1, DB2, DB3, DISK # 2 is DB4, DB5, DB6) Has a StateDB area in which is recorded, a middleware (MW), an MW for recording data, and a data area.
[0018]
SolsticeDiskSuite, which is database management software, performs mirroring with two magnetic disks, but it is a specification that cannot maintain RAID if the normal StateDB becomes 1/2 or less in the system.
[0019]
The database management software of this server checks the Syslog with a C shell command at a predetermined period, for example, every 5 minutes, and if there is a sign that an error will occur in the StateDB area, or if there is an error in the StateDB area, A program that executes the command to delete (metadb -d 3 “device name”) on the StateDB area of the corresponding magnetic disk has been added.
[0020]
As a result, the database management system according to the first embodiment operates so that it can be started at the next reboot even if a failure occurs in the StateDB area of one magnetic disk.
[0021]
First, if a failure occurs in the OS area or the StateDB area, a message “errorlevel Fatal” is written with a Syslog command, for example, with a write error in a file of / var / adm / message.
[0022]
Therefore, as shown in the flowchart shown in FIG. 2, (step S11) is executed by the C shell command at a predetermined period, if I an error message is written to the Syslog (step S 12), either in the error message If an error is included in the OS area of the current disk (step S13), it is assumed that a failure will spread to the StateDB early and late, and if an error is included in the StateDB area, the StateDB area of the corresponding disk is directly set. By executing the command “metadb -d3 DISK # 1” to be deleted, the StateDB area of the corresponding disk is deleted (step S14).
[0023]
Thereby, in the database management system and the database system management method according to the first embodiment, when a failure or a sign of failure of the StateDB of two magnetic disks is monitored and the sign of the failure or failure is found, By deleting the three StateDBs on the magnetic disk, the three Disks on both disks DISK # 1 and DISK # 2 are reduced to a total of six StateDBs on one disk, and the normal StateDB is 3/3 In other words, since it becomes 1/2 or more, the system starts up normally at the next reboot, and the RAID operation of the database can be continued.
[0024]
Next, a database management system and a database system management method according to the second embodiment of the present invention will be described with reference to FIGS. The feature of the second embodiment is that a database management system that uses Sun's SolsticeDiskSuite as database management software periodically checks Syslog error messages with the C shell command, and boot disk (for example, DISK # 1 and When the failure occurs in the boot area, the boot disk is automatically switched to DISK # 2 so that the system can be started up normally at the next reboot.
[0025]
Also in the second embodiment, the hardware configuration is the same as that of the first embodiment shown in FIG. 1, and the server has two built-in disks, DISK # 1 and DISK # 2. Assume that DISK # 1 is currently operating as a boot disk.
[0026]
In this situation, by the processing shown in the flowchart of FIG. 3, the C shell command, it performs Syslog error checking periodically (step S11, S 12). As in the first embodiment, when an error has occurred in the boot disk DISK # 1 (branch to YES in step S13), an error has occurred in the non-boot disk DISK # 2 (step S13). (S13: NO branch) If an error occurs in the StateDB area, or there is an error or a sign, that is, if it is found that an error has occurred in the OS area or the StateDB area, the StateDB of the corresponding disk The three StateDBs in the area are forcibly deleted so that the next reboot can be normally performed when there is no other failure (steps S14, S15 or S18, S19).
[0027]
However, if the error that occurred in the boot disk DISK # 1 includes the Boot area, it is automatically executed by the boot disk switch command of “eeprom boot-device DISK # 2” added to the command line of C shell. The boot disk is switched to DISK # 2 (steps S16 and S17). Thus, at the next reboot, DISK # 2 becomes the boot disk, and reboots from DISK # 2 based on the information in the Boot area, and the system is started.
[0028]
As described above, in the database management system and the database system management method according to the second embodiment, when the failure of the boot area of the two magnetic disks is monitored and the failure of the boot area of the current boot disk is found, It becomes possible to automatically switch the boot disk to another magnetic disk and continue to operate the database, and it is necessary for the administrator to go to the system installation place for switching the boot disk as before. It is possible to reduce the trouble of performing the switching operation by itself.
[0029]
In both the first embodiment and the second embodiment, if a failure occurs in any area of the magnetic disk, the failure will eventually spread to the entire disk due to the characteristics of the magnetic disk. Since there is a high possibility, the failed disk will be replaced during regular inspection.
[0030]
【The invention's effect】
As described above, according to the present invention, a database is provided with software that includes two magnetic disks and prevents the system from starting at the next reboot when the normal StateDB of these magnetic disks is ½ or less of the whole. When operating the system, even if an error occurs in the StateDB on one side of the disk, the next reboot is possible and the work required by the administrator can be reduced as much as possible.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a hardware configuration of a first embodiment of the present invention.
FIG. 2 is a flowchart of processing according to the first embodiment.
FIG. 3 is a flowchart of processing according to the second embodiment of the present invention.
[Explanation of symbols]
DISK # 1 magnetic disk
DISK # 2 magnetic disk
DB1 to DB6 StateDB

Claims (1)

The software that prevents the system from starting up at the next reboot when the normal StateDB becomes less than half of the entire StateDB, which has two magnetic disks and multiple areas on each of these magnetic disks. In the database management system that is adopted,
Check the Syslog every predetermined period, if it contains an error to the OS area of either of the magnetic disk to the Syslog determines that there sign of a fault in the StateDB area, contains an error with respect StateDB region in the Syslog a fault monitoring means for determining that there is failure in the StateDB region if Re,
A database management system comprising: StateDB deletion means for deleting all StateDBs of a corresponding magnetic disk when the failure monitoring means detects the failure or a sign of failure.

Images