JP2002149413A - Method for distributing system definition information of cluster software in cluster type computer system, and system for the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for distributing the system definition information of cluster software by making it possible to set system definition information from one host computer in a cluster type computer system constituted of a plurality of host computers, and to add a host computer to the cluster type computer system at a low cost with a simple mechanism without stopping the operation of the system. SOLUTION: System definition information is stored in a shared disk 16 by either a system definition information registering means 22a or 22b installed in one of a plurality of host computers 12a and 12b of a cluster type computer system 10. Then, cluster software 24a and 24b to be started by the respective host computers 12a and 12b reads the system definition information stored in the shared disk 16, and operates based on the system definition information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing apparatus, and more particularly, to a plurality of host computers each having an independent magnetic disk device (system disk) and starting up a system using the magnetic disk device, and a plurality of host computers. In a cluster-type computer system including a communication device interconnected by a plurality of host computers and a magnetic disk device (shared disk) shared by the plurality of host computers, the host computers can operate the shared disk equally. The present invention relates to a method of distributing system definition information of cluster software and a system therefor.

[0002]

2. Description of the Related Art In the above-mentioned cluster type computer system, software for enabling all the host computers constituting the same to operate the above-mentioned shared disk is called "cluster software". Definition information that is an operation requirement of software or a system is called “system definition information”. In the above-mentioned cluster type computer system, all the host computers constituting the cluster computer system have no master-slave relationship and are all equal, and the operations of the host computers are mutually monitored.

Conventionally, each host computer on which the cluster software starts individually has network addresses for identifying other host computers to be monitored by each host computer, and system definition information of cluster software such as monitoring intervals. It was set to a magnetic disk device (system disk). Therefore, when it is desired to set or change the system definition information, it is necessary to individually execute the operation for setting or changing the system definition information for all the host computers constituting the cluster type computer system.
There is a problem that the setting and changing operations are complicated.

A system which solves this problem and enables setting and changing of system definition information by any one host computer constituting a cluster type computer system is proposed in, for example, Japanese Patent Application Laid-Open No. Hei 5-81036. ing. In the above publication, a method of transmitting environment definition processing information unique to a terminal system from the management system to each terminal system with respect to an installation method in a computer network system and making the environment definition processing work of each terminal system unmanned is disclosed. Proposed. Also, JP-A-6-599
In Japanese Unexamined Patent Publication No. 94, the installation information and the installation target
Simultaneous connection communication by data link control is performed between a primary station host computer having a network address of the next station host computer and a plurality of secondary station host computers to be installed, and a plurality of secondary station host computers are connected. A method has been proposed for transferring installation information at the same time.

The above-mentioned JP-A-5-81036 and JP-A-6-81036
In any of the methods described in JP-A-59994, a definition information management mechanism to be distributed to a remote host computer in a network system, a unit for setting and changing the definition information, and a network address information of a distribution destination remote host computer. It has a management mechanism and a management host computer having these management mechanisms and setting change means.

In the above cluster type computer system, a system definition information management mechanism in which network addresses of all the host computers constituting the cluster type computer system are set is provided in a system disk of all the host computers, and each of the host computers operates. The cluster software monitors the operation of the host computer having all the remaining network addresses excluding the network address of the host computer on which the cluster software operates from the network address of the system definition information via the communication device. Perform Further, in the above cluster type computer system,
By having a means for setting and changing the system definition information together with the system definition information management mechanism in all the host computers constituting the same, all the host computers can equally function as the management host computer, When the system definition information is set or changed by any one of the host computers constituting the system, all the host computers having the network address set in the system definition information management mechanism are communicated through the communication device. Distribute system definition information.

[0007]

However, the conventional methods described in JP-A-5-81036 and JP-A-6-59994 have the following problems.

That is, in a cluster-type computer system including a communication device interconnecting a plurality of host computers and a magnetic disk device (shared disk) shared by the plurality of host computers,
The method described in JP-A-5-81036 and JP-A-6-59994 is described in order to allow any one of the host computers constituting the cluster type computer system to set and change the system definition information. When applying a new host computer to the clustered computer system, the operation of the clustered computer system must be performed in order to enable the added host computer as a host computer constituting the clustered computer system. Must be temporarily stopped, and the system must be restarted after the system definition information is changed. In other words, operations using the cluster computer system must be interrupted.

The reason is that the method of unmanning the environment definition processing work of each terminal system described in the above-mentioned Japanese Patent Application Laid-Open No. 5-81036 and the plural methods described in the Japanese Patent Application Laid-Open No. 6-59994 are disclosed. The method for simultaneously transferring the installation information to one host computer is to transfer the installation information to the host computer using the network system and to set the definition information in the system disk of the transfer destination host computer.

Due to the mechanism of operation of the cluster type computer system, it is necessary to set the same system definition information including the network address information of each host computer to all the host computers constituting the same. The set system definition information becomes effective only when the cluster software is started after the system definition information is set equally for all the host computers.

For this reason, when a new host computer is added to the cluster type computer system, the network address of the host computer to be added is additionally set in the system definition information, and the operation of the cluster type computer system is temporarily stopped. There must be. Therefore, there is a method of adding a host computer which is not activated in the system but apparently added to the configuration to the system definition information before operating the cluster type computer system. In the method of distributing system definition information, simultaneous distribution according to the setting of system definition information on a specific host computer cannot be performed to host computers that do not exist on the network. A method of redistributing definitions is required. In addition, even in the case of distribution to an existing host computer, there is a case where the distribution may fail due to some kind of failure, so it is necessary to judge that the distribution fails because there is no existing distribution method. However, it requires a large amount of additional functions, which is expensive. In any case, a method of redistribution to the added host computer is required, and the distribution procedure becomes complicated, so that consistency of the system definition information cannot be maintained.

SUMMARY OF THE INVENTION Accordingly, the present invention has been made to solve the above problems, and has as its object to provide independent magnetic disk devices (system disks),
A plurality of host computers that use the magnetic disk device to start up the system, a communication device that connects the plurality of host computers to each other, and a magnetic disk device (shared disk) that is shared by the plurality of host computers. In a combined disk sharing system (cluster type computer system), any one of the host computers configuring the cluster type computer system can set and change the system definition information, and the system is not started but the host computer is not configured. By making it possible to add a host computer that is clearly added to the system definition information with a simple mechanism at low cost, the host computer can be added to the cluster computer system without stopping the operation of the cluster computer system. It made it possible to add a computer to provide a method of distributing the system definition information of the cluster software.

Another object of the present invention is to provide an independent magnetic disk device (system disk) suitable for implementing the method of distributing system definition information of cluster software, and to provide the magnetic disk device with A loosely-coupled disk sharing system comprising a plurality of host computers performing system startup, a communication device interconnecting the plurality of host computers, and a magnetic disk device (shared disk) shared by the plurality of host computers ( Cluster type computer system).

[0014]

In order to achieve the above object, a method for distributing system definition information of cluster software according to the present invention is directed to a magnetic disk device (shared disk) shared by a plurality of host computers in a cluster type computer system. ) Stores the system definition information, enables the system definition information in the shared disk to be set and changed from each of the plurality of host computers, and sets the shared disk if cluster software is started in each of the plurality of host computers. Is read out, and the cluster software operates based on the system definition information.

Further, the cluster type computer system according to the present invention includes a magnetic disk device (shared disk) shared by a plurality of host computers in the system and for storing system definition information. Each of: a system definition information registration unit for setting and changing the system definition information in the shared disk; and cluster software operating based on the system definition information by reading the system definition information in the shared disk. When the cluster software operates on each of the plurality of host computers, the system definition information in the shared disk is read, and the cluster software operates based on the system definition information. Set Characterized in that to allow the distribution of information.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A preferred embodiment of the present invention will be described. In a cluster type computer system according to a preferred embodiment of the present invention, if the system definition information is set in any of a plurality of host computers constituting the cluster type computer system, the setting content is changed to a plurality of host computers. The system definition information of the cluster type computer system is registered or updated in a magnetic disk device (shared disk) shared by the computers.

The cluster software started by each host computer operates based on the system definition information stored in the shared disk. For this reason, the setting of the system definition information performed by one host computer becomes effective in the cluster software of all the host computers thereafter.

More specifically, the present invention, in a preferred embodiment thereof, comprises a plurality of host computers (14a, 14b in FIG. 1) each of which has an independent system disk and uses the system disk to start up the system. A communication device (18 in FIG. 1) to which the host computer connects to each other, and a shared disk (FIG. 1) shared by the host computer
16), the cluster-type computer system (10 in FIG. 1) stores the network address information of a plurality of host computers in the system definition information management mechanism (28 in FIG. 1) in the shared disk, and Cluster software presence / absence information indicating whether or not the corresponding cluster software is running, and operation definition information common to the cluster software such as the monitoring interval between cluster software on each host computer and the allowable number of monitoring failures, Disk address management mechanism in the system disk (26a, 2 in FIG. 1)
6b) stores the shared disk address information indicating the position of the system definition information management mechanism in the shared disk.

Each host computer reads the shared disk address information from the disk address management mechanism, and reads the shared disk address information from the disk address management mechanism (20a, 20b in FIG. 1). System definition information registration means (22a, 22 in FIG. 1) for registering system definition information in the system definition information management mechanism indicated by the address information.
b) reading the shared disk address information from the disk address management mechanism at the time of startup, further reading the system definition information registered in the system definition information management mechanism indicated by the read shared disk address information, and based on the read system definition information. The provision of operating cluster software (24a and 24b in FIG. 1) enables distribution of system definition information of the cluster software.

Next, in order to describe the embodiments of the present invention in more detail, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of one embodiment of the present invention. Referring to FIG. 1, the present embodiment applies the present invention to a cluster-type computer system 10 including a plurality of (two in this example) host computers 12a and 12b operating on a common resource. Applied. Although FIG. 1 shows a configuration example of two host computers for simplicity, the present invention is not limited to such a configuration.

The cluster type computer system 10 includes:
All host computers 12 belonging to the same system
a and 12b are provided with a communication device 18 and a shared disk 16 which are shared. In the shared disk 16, a system definition information management mechanism 2 for storing system definition information
8 have been built.

In this embodiment, the system definition information is
The cluster software on each host computer monitors, via the communication device 18, the network address of the host computer that identifies the host computer on which the other cluster software is running, the monitoring interval between each cluster software, and the allowable number of monitoring failures. This is operation definition information common to cluster software and information indicating whether or not cluster software has been activated on each host computer belonging to the same system for each host computer.

FIG. 2 shows an example of the configuration of the system definition information stored in the system definition information management mechanism 28 in the shared disk 16 in one embodiment of the present invention.

Referring to FIG. 2, the system definition information management mechanism stores a column for storing the network address of each host computer (the leftmost column in the table), and cluster software presence / absence information indicating whether cluster software has been activated. (A column on the left side of the center of the table), a column for storing monitoring interval information between cluster software (a column on the right side of the table), and a column for storing information on the allowable number of times of failure in monitoring between cluster software (table Rightmost column).

Each host computer 12a, 12b
Each has system disks 14a and 14b, and disk address management mechanisms 26a and 26b for storing shared disk addresses are built in each system disk.

Each host computer 12a, 12b
Disk address registration means 20a and 20b are provided for updating and registering shared disk addresses of the disk address management mechanisms 26a and 26b provided in the system disks 14a and 14b. The disk address registration means 20a and 20b are constructed by software.

Each of the host computers 12a, 12a
b includes system definition information registration means 22a and 22b, which reads the shared disk addresses of the disk address management mechanisms 26a and 26b in the system disks 14a and 14b, and reads the position indicated by the shared disk address on the shared disk 16. This is for updating and registering the system definition information of the system definition information management mechanism 28 provided in. The system definition information registration means 22a and 22b are constructed by software.

Further, each host computer 12a, 1
2b comprises cluster software 24a, 24b,
This is because the shared disk addresses of the disk address management mechanisms 26a and 26b in the system disks 14a and 14b are read out at the time of startup, and the system definition information management mechanism 28 provided at the position indicated by the shared disk address on the shared disk 16 is read. Read the system definition information,
Based on this system definition information, each host computer operates as a cluster-type computer system for shared resources.

The cluster software 24a, 24b is constructed by software.

The operation of the embodiment of the present invention will be described. First, a specific example of an operation of setting a shared disk address to be executed as a pre-process of an operation of setting system definition information in the cluster computer system 10 will be described with reference to a flowchart of FIG.
The setting of the shared disk address is performed by all the host computers in the system in order for the cluster software to operate. Here, as an example, the disk address registration means 20a is set in the host computer 12a.
It is assumed that the command to start
0). Activated disk address registration means (step 3
2) read out the disk address management mechanism 26a in the system disk 14a and, when there is a change input of the read-out shared disk address, write the changed shared disk address to the disk address management mechanism 26a. Is updated (step 34).

Next, the cluster type computer system 1
A specific example of the operation of setting the system definition information at 0 will be described with reference to the flowchart shown in FIG. Here, it is assumed that a command for starting the system definition information registration means has been input in the host computer 12a (step 40). When the system definition information registration means 22a is started by inputting a command (step 41),
The system definition information registration means 22a first reads the shared disk address stored in the disk address management mechanism 26a on the system disk 14a in the host computer 12a (step 42), and the stored shared disk address is correct. It is checked whether the shared disk is pointed (step 43). If the shared disk address is invalid, the system definition information registration unit 22a displays the error message such as "invalid shared disk address" and terminates (step 4).
9).

On the other hand, if the shared disk address is correct, the system definition information registration means 22a reads the system definition information stored in the system definition information management mechanism 28 in the shared disk 16 indicated by the shared disk address. (Step 45) Check whether the cluster software is running on the host computer registered in the system definition information (Step 4).
6). If there is a column in which “Presence” is stored in the presence / absence information of the cluster software in the system definition information, register the system definition information in order to maintain the consistency of the system definition of the cluster software and the consistency of operation. The means 22a displays the error message such as the error message "Cluster software is running" and terminates (step 49).

On the other hand, if "none" is stored in the column of the presence / absence information of cluster software in the system definition information, the system definition information registration means 22a
When the system definition information management mechanism 28 in the shared disk 16 is read, and when there is a change input of the read system definition information, the system definition information after the change is written in the system definition information management mechanism 28, whereby the system definition of the cluster software is obtained. Update registration of information is performed (step 34).

Since the update registration of the system definition information does not depend on the operation of the host computer to be added, an inactive host computer can be additionally registered in the system definition information. Also, by storing the system definition information of the cluster type computer system 10 on the shared disk 16, the operation of updating the system definition information by any host computer in the system is the same as the operation of the host computer 12a. It will be exactly the same. Therefore, in the cluster type computer system 10, only a command for executing the system definition information registration means is input from any one of the host computers in the system, and the change is thereafter applied to all the host computers in the system. The change becomes effective in the computer, and in the change, by adding a host computer whose system has not been started to the system definition, the host computer can be added without stopping the system thereafter.

[0036]

As described above, according to the present invention,
It is possible to add a host computer to a cluster computer system at a low cost with a simple mechanism without stopping the operation of the cluster computer system, and to add the host computer to any one of a plurality of host computers in the system. The effect is that the system definition information can be executed only by performing the addition operation of the system definition information by one host computer. Therefore, the operability of the system is particularly improved.

[Brief description of the drawings]

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

FIG. 2 is a diagram illustrating an example of a configuration of system definition information stored in a system definition information management mechanism in a shared disk according to an embodiment of the present invention.

FIG. 3 is a flowchart for explaining an operation of setting shared disk address information in the operation of one embodiment of the present invention.

FIG. 4 is a flowchart for explaining an operation of setting system definition information in the operation of one embodiment of the present invention.

[Explanation of symbols]

10 cluster computer system, 12a, 12
b: Host computer, 14a, 14b: System disk, 16: Shared disk, 18: Communication device, 20
a, 20b... disk address registration means, 22a, 22
b: system definition information registration means, 24a, 24b: cluster software, 26a, 26b: disk address management mechanism, 28: system definition information management mechanism.

Claims (2)

[Claims] 1. A plurality of host computers each having an independent magnetic disk device (system disk), performing system startup using the magnetic disk device, a communication device interconnecting the plurality of host computers, and In a loosely coupled disk sharing system (cluster type computer system) composed of magnetic disk devices (shared disks) shared by a plurality of host computers, each of the host computers equally stores the shared disk in a system disk of each of the host computers. A disk address management mechanism for managing a disk address (shared disk address) of a shared disk that stores definition information (system definition information) which is required by software (cluster software) for enabling operation (cluster software); A disk address registration unit for registering and updating the shared disk address in the disk address management unit; a system definition information management unit for storing and managing system definition information of the cluster software in the shared disk; System definition information registration means for registering and updating the system definition information of the cluster software in the system definition information management mechanism, wherein the cluster software started from the system disk of each of the host computers is shared from the disk address management mechanism by the shared disk. The shared disk address is read from the system disk by means for reading the address, and the cluster software sends the system definition information from the system definition information management mechanism indicated by the read shared disk address. The system definition information is read by a reading means, and the cluster software operates based on the read system definition information while storing the read system definition information in a system disk of the host computer. Method of distributing system definition information. 2. A disk address management mechanism for managing a disk address (shared disk address) of a shared disk storing system definition information of the cluster software in a system disk of each of the host computers in the cluster type computer system. Disk address registration means for registering and updating the shared disk address in the disk address management mechanism; and system definition information of the cluster software (that is, definition information that the cluster software has as operating requirements) in the shared disk. A system definition information management mechanism for storing and managing, and system definition information registration means for registering and updating system definition information of cluster software in the system definition information management mechanism; The system definition information registering means started in each host computer registers the system definition information in the system definition information management mechanism indicated by the shared disk address stored in the disk address management mechanism of each host computer, and the system disk of each host computer described above. The cluster software read from the system disk having means for reading the shared disk address from the disk address management mechanism, and reads the shared disk address from the system disk. The cluster software reads the shared disk address from the system definition information management mechanism indicated by the read shared disk address. Means for reading definition information, reads the system definition information, and the cluster software stores the read system definition information in the system disk of its own host computer. The reluctant the readout system definition information by operating as an initial value,
A method of distributing system definition information, wherein setting and changing of system definition information are enabled by any one of the host computers.