JP2001195377A - Isolation judgment system, management method therefor and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To recognize fault generation in respective slave nodes themselves and to reduce burdens for a management job as a master node. SOLUTION: In this management method of the isolation judgment system for which a management computer (master node) and plural computers (slave nodes) to be managed are connected by a network, it is constituted of plural management groups for which the several pieces of the slave nodes are used as one set, the slave nodes inside the management group exchange node information with each other and the respective slave nodes hold the node information inside the management group. The master node manages the node information of all the slave nodes by exchanging the node information with one piece of the slave node of a highest priority order inside each management group.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a management method and an isolation determination system for an isolation determination system in which a management computer (master node) and a plurality of managed computers (slave nodes) are connected via a network, and the management. The present invention relates to a recording medium in which a method is an operation program and the operation program is stored in a computer-readable manner.

[0002]

2. Description of the Related Art As an example of a conventional node information exchange method, there is a method in which a master node exchanges node information with all nodes.

The system of such a system operates as follows.

That is, the master node exchanges node information with all nodes, and the master node manages all node information.

[0005] As an example of this, Japanese Patent Application Laid-Open No. 2-137552 is disclosed.
This publication describes a data communication system for transmitting data between a main device and slave devices divided into a plurality of groups. The master device manages each group code of one or more groups to which each slave device belongs, and a group code for sending data retrieved by searching the management device to communication data to be sent to the slave device. First processing means for adding and creating transmission data;
And first communication means for transmitting the data created by the processing means to the plurality of groups, wherein the plurality of slave devices are:
Setting means for presetting one or more groups to which the slave device belongs, second communication means for receiving data from the first communication means, and extracting a group code from the received data; Second processing means for comparing and analyzing the group code and the group code set in the setting means, and when the two group codes match as a result of the analysis by the second processing means, Describes that the received data is valid, and if they do not match, the received data is discarded.

[0006] Japanese Patent Application Laid-Open No. 4-42632 discloses that
As a system management method, in a system management method including a main management device, a sub-management device and a managed device managed by the main management device, and the sub-management device performs a patrol for a health check on the managed device, The main management device is provided with means for dynamically changing the configuration information and patrol monitoring information of the managed device to be patrolled by the sub-management device. Describes that the group of managed devices managed by the slave management device is managed in addition to the group of managed devices managed by the slave management device that can act on behalf of the slave management device. Have been. These documents describe that patrols of health checks are performed at regular intervals to detect injuries, and during the failure period, managed devices are managed by exchanging patrol frame information with other slave management devices. ing.

Japanese Patent Application Laid-Open No. 7-112488 discloses a periodic diagnosis method for a multiprocessor system comprising a master processor and a slave processor connected to the master processor through a bus.
One of a plurality of slave processors in each unit in which a plurality of slave processors are grouped into a plurality of groups is selected as a health check master, and the master processor periodically diagnoses only the health check master, and a failure occurs based on the diagnosis result. It is described that a resumption process is sometimes performed. This health check master performs periodic diagnosis of slave processors in the same unit, transmits the diagnosis result to the master processor, and describes a two-stage management method.

[0008]

However, the above-mentioned Japanese Patent Application Laid-Open No.
In JP-A-42632, there is a subordinate management device for each management group, and only the subordinate management device manages the managed devices in the management group. However, there is no determination of the priority order in the management group. However, there is a problem that the main management device has to scan and manage the managed device together with all the slave management devices, and the burden of managing the main management device is large.

Japanese Patent Laid-Open Publication No. 7-112488 discloses that information in a management group is collected only in a health check master. When the health check master changes, the master receives a request from the master and knows the scope of management for the first time and starts collecting information. Unable to find their own injuries.

As described above, these conventional techniques have the following problems.

The first problem is that the master node or the main unit or the master processor sends the node information and the health check patrol to the managed devices and the slave processors managed by all the slave nodes or the slave management units. This means that the load is placed on the master node to perform the replacement.

The second problem is that if the health check process fails, is the cause on the master node side?
Because it is not possible to determine whether it is on another node side,
This means that the reliability of exchanging node information is low.

Accordingly, an object of the present invention is to make it possible for each slave node itself to know the occurrence of a fault, and to reduce the burden for management work as a master node.

[0014]

According to the present invention, there is provided a management computer (master node) for solving the above problems.
A plurality of managed computers (slave nodes) are connected via a network to determine whether or not each node is isolated. It comprises a management group, the slave nodes in the management group exchange node information with each other, each slave node holds node information in the management group, and the master node By exchanging node information with one of the high priority slave nodes, the node information of all the slave nodes is managed.

Further, according to the present invention, in an isolation determination system in which a management computer (master node) and a plurality of managed computers (slave nodes) are connected via a network, the master node includes several slave nodes. A management group priority list provided for each of a plurality of management groups, and a transmission / reception unit for transmitting / receiving node information to / from the network, wherein the slave nodes in the management group include the master node and the management node. A transmission / reception unit that transmits / receives node information to / from other slave nodes in the group via the network, and the management group priority list in the management group, wherein the master node has a priority order of each of the management groups. Executes the node information acquisition request to one of the management groups In response to a node acquisition request from a slave node in the management group, the node information of the management group priority list to which the slave node that has requested the acquisition belongs is transmitted, and the slave node is configured to transmit at least another node in the management group. The management node priority list is updated by mutually executing a node information acquisition request with a slave node.

The present invention also provides an isolation determination system in which a plurality of management computers (master nodes) and a plurality of managed computers (slave nodes) managed by the master nodes are connected via a network. The node holds a master management group priority list for storing node information by executing a request for acquiring node information mutually with other master nodes, and the master node sets several slave nodes as one set. A management group priority list provided for each of the plurality of management groups, and a transmission / reception unit for transmitting / receiving node information to / from the network, wherein the slave nodes in the management group are other than the master node and the slave nodes in the management group. Send and receive node information with other slave nodes via the network The transmitting and receiving unit, comprising the management group priority list in the management group, the master node executes the node information acquisition request to one of each management group according to the priority of each management group, In response to a node acquisition request from a slave node in the management group, transmits node information of a management group priority list to which the slave node that has requested the acquisition belongs, and the slave node communicates with at least another slave node in the management group. The management group priority order list is updated by mutually executing a node information acquisition request.

According to the present invention, a management computer (master node) and a plurality of managed computers (slave nodes) are connected via a network to manage an isolation determination system for determining whether each node is isolated. In a recording medium in which the method is an operation program and the operation program is stored in a computer-readable manner, the management method of the isolation determination system comprises a plurality of management groups each including a set of the slave nodes. The slave nodes in the group exchange node information with each other, each slave node holds the node information in the management group, and the master node has one of the higher priority in each management group. By exchanging node information with slave nodes, the master node can manage the nodes. To as the operation program to manage the node information of all the slave nodes, characterized in that the said operating program computer has stored readably.

Further, according to the present invention, in a system in which a management computer (master node) and a plurality of managed computers (slave nodes) are connected via a network, a plurality of management groups each including several slave nodes are set. The slave nodes in the management group exchange node information with each other, each slave node holds the node information in the group, and the master node sends one node to each slave node in each group. In this method, the node information of all slave nodes can be managed by exchanging the node information, and a function is provided in which the slave nodes in the management group determine an isolated state by themselves and stop the system.

Therefore, if a fault occurs in each slave node, if the fault is recovered in isolation, the other slave nodes are not affected, and if a fault occurs in the master node, fault detection is performed. Since each slave node that does not receive the operation information operates in isolation, the other slave nodes can execute processing in isolation until the failure of the master node is recovered, minimizing the effect of the node where the failure occurred. it can.

Further, according to the present invention, referring to FIG. 1, each of the managed slave nodes 11 to 19 constitutes one set of management groups 111 to 113.
Each of the slave nodes 11 to 19 exchanges node information between the slave nodes in the management group via the communication device 120, and stores the node information in the management group in the main storage device 21 and the auxiliary storage device of the own node. To hold.

The master node 1 periodically exchanges node information with the high-priority slave nodes 11 in the management groups 111 to 113 via the communication device 120 as a server for managing the state of each slave node. To do.
The master node 1 manages the node information of all the managed slave nodes 11 to 19 by holding the node information in each of the management groups 111 to 113 in the main storage device 101 or the auxiliary storage device of the own node. Can be. If the slave nodes 11 to 19 cannot exchange node information with other slave nodes in the group due to some kind of failure, the slave nodes 11 to 19 determine that they are isolated in the system and notify the master node 1 of the node information. By performing the exchange, the node itself can determine whether it is a failure of its own node or a failure of another node. Similarly, the master node 1 can determine whether or not the master node 1 is isolated from other slave nodes of the management system based on a response when detecting node information via the communication device 120.

In this way, the slave node itself and the master slave itself can determine whether the node itself has failed or another node has failed, and the master node exchanges node information with all slave nodes. Since this is not performed, the reliability of the exchange of the node information can be improved and the load on the master node can be reduced as compared with the conventional method in which the master node exchanges the node information with all the slave nodes.

[0023]

Embodiments of the present invention will be described in detail with reference to the drawings.

First Embodiment (1) Description of Configuration Referring to FIG. 1, in a first embodiment of the present invention, in a transaction system constructed by a plurality of computers, a master node is used for management and a slave is used. Assuming that the nodes are computers that execute transactions, the node information exchange method by grouping includes a master node 1, slave nodes 11 to 19, and a communication device 120.

As the managed computer 100, the slave nodes 11 to 19 each operate as a managed computer in the same manner as a so-called personal computer, and perform accounting work processing, word processing, distributed computer processing such as CAD and design work, and master processing. It can communicate with a node, another slave node, or the like to receive data.

The master node 1 is provided with a database and the like (not shown) together with a task for managing the managed computer based on the priority list of each management group.
For example, tasks such as transmitting various data in response to a request from the slave node, receiving data of the slave node and storing the data in a database can be performed.

Here, the slave nodes 11 to 13 mutually request the node information and share the node information. The management group 111 shares the node information. The slave nodes 14 to 16 mutually request the node information and share the node information. The management group 112 and the slave nodes 17 to 19 form a management group 113 that requests acquisition of node information mutually and shares the node information. Each of the slave nodes 11 to 19 has a priority list of the management group to which the own node belongs.

The master node 1 has a management group 11
1, 112, and 113 in each group. Each management group priority list 101-103
Is a list of node names sorted in descending order of priority in the order of inquiries in each management group. Each slave node name and OK, OK, There are areas 121 to 123 for storing the status of NG and the update time of the status.

The master node 1 is a management computer 1
00, a node inquiring function for inquiring node information of a higher priority node of each group in the system, a switching function of switching an inquiry to a node having the next priority when node information of the group cannot be acquired, It has a reply function for replying to an inquiry about node information from a node.

Each slave node in each of the management groups 1 to 3 has an information inquiry function for inquiring other slave nodes in each group for node information, and own node information in response to an inquiry from another slave node. A reply function that replies together with the node information of the other slave nodes that are present, an update function that updates the received node information if it is newer node information compared to the stored node information, and an own node. A master inquiry function for making an inquiry to the master node 1 when there is no response to an inquiry to another slave node to which the management group to which the master node belongs, and a case where communication with the master node and the slave node known to the own node cannot be performed. Has a stop function of stopping the system by determining that it is isolated.

(2) Description of Operation Next, the overall operation of the present embodiment will be described in detail with reference to FIG. 1, FIG. 2, and FIG. First, in FIG.
The node information collecting operation of the management computer (master node 1) will be described.

The master node 1 acquires a node name of a slave node having a higher priority in each management group to be collected for each predetermined management time (S1). A node information acquisition request is made to the slave node corresponding to the node name acquired for each of the management groups 111 to 113 via the communication device 120 (S2). If there is a response to the request, the node information (state, time) is stored in the storage device (S3, 4). The master node 1 counts the time, and if there is no response for a predetermined time, acquires a node name in the next priority order (S1). If the slave node has been acquired within a predetermined time, a node information acquisition request is made to the slave node (S2). This processing is repeated until node information of the management group can be obtained or a node information acquisition request operation is performed for all slave nodes in the priority list of the management group. This process is performed for all management groups of the system (S5). If node information has not been collected for all management groups, the process proceeds to step S2, and if node information has been collected for all management groups, the process ends.

Next, an operation when the master node 1 receives a node information request will be described with reference to FIG.

Upon receiving the node information request, the master node 1 searches each management group priority list for the node information of the slave node specified by the node information request from the node information of each management group collected so far. And read (S6). The read node information is transmitted to the requesting slave node (S7). In this case, the node information of the other slave node can be transmitted to the requesting slave node to notify the requesting slave node whether or not the other slave node has failed. Also, in consideration of security, it is possible to restrictively transmit node information of other slave nodes in the same management group. The operation of the master node 1 is restricted by the restriction of the node information on the slave nodes in the same management group, so that the management work of the master node can be reduced.

Next, referring to FIG.
1, 112, 113 each slave node 11-
The nineteenth node information collection operation will be described.

Each of the slave nodes 11 to 19 acquires another node name in its own management group from its own management group priority list (S8). For example, in the management group 111, a list of the slave nodes 11 to 13 is displayed on a display (not shown), and the node name is acquired by clicking the node name of the slave node that the user wants to know. A node information acquisition request is sent to the slave node corresponding to the acquired node name (S9). It is determined whether a response from the slave node has been received within a predetermined time (S10), and if node information cannot be obtained (S1).
1) Assuming that the status of the corresponding node information is abnormal, for example, “NG” and the time at that time are stored in the status item of the management group priority list, and the process proceeds to the next node information acquisition (S).
8). If the node information can be obtained, the state of the corresponding node information is stored as normal “OK”, and the node information other than the corresponding node transmitted at the same time is compared with the corresponding corresponding node information, and the comparison result is obtained. , The node information whose update time is newer than the information in the storage device is updated, and the process proceeds to the next node information acquisition (S12).

If all the node information in the management group has been collected, there is no slave node to be selected from the management group priority list in the own group. Is determined (S13), and if not abnormal, the node information collecting operation is terminated (STOP). If all the node information in the management group is abnormal, an inquiry is made to the master node 1 via the communication device 120 (S13,
14). Here, the determination as to whether or not there is an abnormality is made when there is no response for a predetermined time after requesting the acquisition information of the node information, or when the own acquisition request is not transmitted to the destination slave node, the communication device is damaged. If it has occurred, there is no response in any case, so it is determined to be abnormal.

If it is determined in step S13 that all the information is abnormal, the process ends if node information can be collected from the response from the master node 1 (S15). When there is no response from the master node 1 within a predetermined time, it is determined that the own node is isolated from the network, and the system of the own slave node is stopped (S16).

Next, an operation when the slave node receives a node information acquisition request will be described with reference to FIG. This operation may be called a health check. This health check should be checked regularly, in principle, once a day.
Health check may be performed once every 10 minutes, or optionally, according to the frequency of needing to know node information of other slave nodes in the group necessary for business processing of the management group. I do.

When receiving a node information acquisition request from another slave node or the master node 1, the slave nodes 1 to 9 read their own node information (S17), and collects and retains the own node information up to that point in the management group. Is read from the management group priority list stored in the own slave node (S1).
8) Transmit both the own node information and the other node information in the own management group held as request information to the request source (S19).

[Second Embodiment] Next, a second embodiment of the present invention will be described in detail with reference to the drawings. Referring to FIG. 6, the master nodes 202, 203, 20
4 are mutually connected to the network 201. Master nodes 202, 203, 20
By making each of the four perform the same operation as the slave node similar to the management group of the above-described first embodiment, it is possible to mutually monitor the systems.

In FIG. 6, a master node 202 manages a management group 211, a management group 212, and a management group 213 as masters via transmission paths of communication devices. The master node 203 manages the management group 214 and the management group 215 as masters. The master node 204 is a management group 216
And the management group 217 as a master through the transmission path of the communication device. Each of the master nodes may be a management server that exclusively performs the management task of the slave node, or may execute the management task of the slave node as a part of the multitask.

Each master node 202, 203, 204
Uses this master node group as one group system, holds a mutual master management group priority list, assigns priorities randomly, or assigns priorities according to the size of the system, or a plurality of slave nodes. Is assigned as a management group, and priorities are assigned according to the number of management groups, or priorities are assigned in descending order of the number of all managed slave nodes instead of the management groups, and group system priorities according to the respective priorities The lists are mutually stored, and a health check is performed between other master nodes at predetermined time intervals.

[Third Embodiment] Still referring to FIG. 7, the third embodiment includes a storage medium 130 storing an operation program of a system for exchanging node information by grouping and an isolation determination. The storage medium 130 is mounted on the driver 131 of the master node 1 and the operation program is written or read. The storage medium 130 is a magnetic disk such as a hard disk or a floppy disk, a semiconductor memory such as an IC card or a stick memory having a flash memory, and other storage media such as a CD-R, a CD-RW, and a DVD-RAM. May be.

The installation from the storage medium is performed by the master node, and an instruction is given to each slave node to perform the installation. Further, in order to individually determine whether or not the slave node is isolated in the management system, the communication device 1
The storage medium may be installed in its own storage means through the storage medium 20 or inserted into a drive of its own recording medium and packaged.

The operation program of the system for exchanging node information and isolating by grouping is stored in the storage medium 130.
To a master node 1 and a slave node 1 of each computer (central processing unit; processor; data processing unit)
1 to 19, control of exchange of node information by grouping of slave nodes and isolation determination. Each slave node performs the same processing as the processing shown in the first embodiment by exchanging the node information by the read grouping and by the operation program of the isolation determination system.

In this operation program, an operation program for the entire isolation determination system, an operation program for the master node 1, and an operation program for each of the slave nodes 11 to 19 may be individually provided. In the case where the data is managed and updated under management, it is preferable to update the data for each slave node in accordance with an instruction from the master node 1 in terms of system management unification.

As described in the first embodiment, the operation program sets a priority order for each management group, and sets the master node to one of the management groups.
First, a slave computer in each management group requests node information from a response, and the slave node in each management group requests node information from another slave node in the same management group and waits for a response.

In each of the above embodiments, the management group priority order list 101 of the master node,
Since the management group priority list of the slave nodes 11 to 13 has the same contents, if any node fails, if the failed node recovers, the master node By copying the management group priority list of the same management group or the management group priority list of any other slave node of the management group, the same management group priority list can be obtained.
The isolation determination system makes the recovery process extremely easy.

[0050]

According to the present invention, the slave nodes in the group perform a health check, and if the slave nodes in the group cannot perform a health check, the health check is performed on the master node. The reliability is increased because the health check is performed by two systems, and the master node performs a health check with only one of each group. Therefore, the load can be reduced compared to the method of performing a health check on all nodes. Thus, the reliability of the health check can be improved and the load on the master node can be reduced.

If there is no health check response from another node in the group, the information from the non-responding slave nodes includes the node information of the non-responding slave node, and the time of the information is referred to. Can be determined before querying the master node, and finally by querying the master node,
It is possible to determine the cause of a failure in its own node or a failure in another node in the group.Therefore, if a failure has occurred in its own node or another node in the group has failed. Can be separated.

When a failure occurs in a node for which the master node inquires about a node state, other nodes in the group to which the node belongs also have node information in the group. Therefore, the master node can perform management of all nodes without any problem by making a node information request to other nodes in the management group. Can also manage the state of all nodes.

[Brief description of the drawings]

FIG. 1 is a configuration block diagram of an isolation determination system according to the present invention.

FIG. 2 is an operation flowchart of a master node of the isolation determination system according to the present invention.

FIG. 3 is an operation flowchart of a master node of the isolation determination system according to the present invention.

FIG. 4 is an operation flowchart of a slave node of the isolation determination system according to the present invention.

FIG. 5 is an operation flowchart of a slave node of the isolation determination system according to the present invention.

FIG. 6 is a configuration block diagram of a network of an isolation determination system according to the present invention.

FIG. 7 is a configuration block diagram of an isolation determination system according to the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Master node 11-19 Slave node 21 Management group priority list 100 Management computer 101-103 Management group priority list 111-113 Management group 120 Communication equipment 121-123 Contents of management group priority list 130 Storage medium

Claims (8)

[Claims] 1. Management computer (master node)
A plurality of managed computers (slave nodes) are connected via a network, and each of the nodes determines an isolated state. And the slave nodes in the management group exchange node information with each other, each slave node holds node information in the management group, and the master node is a node in each management group. The node information of all the slave nodes managed by the master node is exchanged by exchanging node information with the slave node having a higher priority order. 2. The management method according to claim 1, wherein each of the slave nodes determines that the slave node is isolated from other slave nodes in the same management group, and determines whether the slave node is in the isolated state. A management method for an isolation determination system, comprising a function of stopping an operation function of the slave node for recovery of the slave node. 3. The management method for an isolation determination system according to claim 1, wherein the master node has a management group priority list for each of the management groups, and the master node has a management group priority list at predetermined time intervals. A management method for an isolation determination system, wherein a health check is performed on a slave node having a priority. 4. The management method according to claim 1, wherein the slave node has a management group priority list in the management group to which the slave node belongs, and the slave node has a management group priority list at predetermined time intervals. Send a node information acquisition request to other slave nodes from the list, and if there is no response, send the node information acquisition request to other slave nodes, and update their own management group priority list according to the response. A management method for an isolation determination system characterized by the following. 5. A management computer (master node)
And an isolation determination system in which a plurality of managed computers (slave nodes) are connected via a network.
A management group priority list provided for each of the plurality of management groups, and a transmission / reception unit for transmitting / receiving node information to / from the network, wherein the slave nodes in the management group are in the master node and in the management group. A transmission / reception unit that transmits / receives node information to / from another slave node via the network, and the management group priority list in the management group, wherein the master node manages each management group according to the priority of each management group. Executing the acquisition request of the node information to one of the above, in response to the node acquisition request from the slave nodes in the management group, transmitting the node information of the management group priority list to which the slave node that has requested the acquisition belongs, The slave node has at least other slave nodes in the management group. Isolation determination system and updates the management group priority list by performing the acquisition request of the slave node and mutually node information. 6. The isolation determination system according to claim 5, wherein the slave node executes a request for acquiring the node information to another slave node in the management group and does not respond to the management group. An NG flag is set to the node status for the other slave nodes in the priority list, and if the master node issues a request to acquire the node information and there is no response, it is determined that the own device has been isolated. An isolation determination system characterized in that: 7. An isolation determination system in which a plurality of management computers (master nodes) and a plurality of managed computers (slave nodes) respectively managed by the master node are connected via a network. The master node has a master management group priority list for storing node information by executing a node information acquisition request with the master node.
A management group priority list provided for each of the plurality of management groups, and a transmission / reception unit for transmitting / receiving node information to / from the network, wherein the slave nodes in the management group are in the master node and in the management group. A transmission / reception unit that transmits / receives node information to / from another slave node via the network, and the management group priority list in the management group, wherein the master node manages each management group according to the priority of each management group. Executing the acquisition request of the node information to one of the above, in response to the node acquisition request from the slave nodes in the management group, transmitting the node information of the management group priority list to which the slave node that has requested the acquisition belongs, The slave node has at least other slave nodes in the management group. Isolation determination system and updates the management group priority list by performing the acquisition request of the slave node and mutually node information. 8. A management computer (master node)
A management method of an isolation determination system in which a plurality of managed computers (slave nodes) are connected via a network and each node determines an isolated state;
In a recording medium storing the operation program in a computer-readable manner, the management method of the isolation determination system comprises a plurality of management groups each including a set of the slave nodes, and the slave nodes in the management group. The nodes exchange node information with each other, each slave node holds node information in the management group, and the master node sends a node to one high-priority slave node in each management group. By exchanging information,
A recording medium, wherein the operation program is configured to manage node information of all slave nodes managed by the master node as the operation program, and the operation program is stored in a computer-readable manner.