JP2009128987A - Computer management system, computer management method and computer management control program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a computer management system and method, and a computer management control program for quickly changing, if a manager of a computer group connected to a network fails, the manager to a new manager managing an agent. <P>SOLUTION: A management node determination section 211 of a manager 201 determines an agent which can serve as a manager between agents 202<SB>1</SB>and 202<SB>2</SB>before the manager 201 fails and stores it in a management node storage section 227 of the current agent. When a manager monitor section 221 detects a failure in the manager 201, a management request section 225 requests the agent 202 thereof to serve as a manager. Consequently, the manager 201 can speedily be switched. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a computer management system, a computer management method, and a computer management control program for managing a plurality of computers connected to a network. Particularly, a computer management system and a computer suitable for a failure in a managing computer. The present invention relates to a management method and a computer management control program.

  Many personal computers, servers or facsimile machines, mobile phones, PDA (Personal Digital Assistants) built-in processors (hereinafter simply referred to as computers) are connected to networks such as the Internet and LAN (Local Area Network). It is becoming. These computers are connected via a network, so that one computer among them can manage other computers among specific groups.

  In this specification, a managing computer is called a manager, and a managed computer is called an agent. For example, it is assumed that a plurality of personal computers are connected as one group via the Internet, and one of them or one server arranged in the same group is a manager. Then, the manager monitors agents composed of other personal computers in the group. And when a failure occurs in one of them, this is notified to the user, or if the setting of the personal computer permits, an appropriate process such as a restart process is performed to recover from the failure Action can be taken.

  When such a computer management system is adopted, information of each agent is concentrated on the manager. Therefore, the manager can easily grasp the entire system. However, when the functions for system management are concentrated on the manager, if the manager itself fails or if communication between the manager and the agent is interrupted due to a network failure, the node that manages the agent No longer exists. As a result, when such a failure occurs in the manager, the reliability of the computer management system itself is significantly reduced.

  Therefore, it has been proposed as a first related technique of the present invention to duplicate a node serving as a manager (see, for example, Patent Document 1). Further, in the second related technology of the present invention, the node serving as the manager is duplexed, and when one of these nodes fails, the agent is set as the manager and the manager node is always duplexed. (For example, refer to Patent Document 2).

  In these first and second related technologies, the manager node is duplicated into an active system and a standby system. However, it cannot be denied that there is a possibility of simultaneous failure of these duplicated manager nodes. When such a failure occurs, these technologies cannot cope with the continuity of communication. There is also a computer management system in which it is impossible to duplicate manager nodes due to resource constraints.

Therefore, a third related technique of the present invention in which a network monitoring server and a management server are provided in addition to a client terminal management server that manages a plurality of client terminals has been proposed (see, for example, Patent Document 3). ). The third related technique has been devised for the purpose of reducing the burden on the management server, and addresses of a plurality of network monitoring servers are registered in the client terminal management server. Thus, even when a failure occurs in one of the network monitoring servers, the client terminal management server can access another network monitoring server and acquire the address of the desired network monitoring server.
Japanese Patent Laid-Open No. 06-197112 (paragraph 0013, FIG. 1) Japanese Patent Laying-Open No. 2006-235837 (paragraph 0010, FIG. 1) JP 2003-256303 A (paragraphs 0029, 0045, 0060, FIG. 4)

  However, the third related technique is not a technique in which, when a failure occurs in the client terminal management server, the network monitoring server takes over the management of the client terminal instead. That is, since the network monitoring server is connected via the only client terminal management server, if a failure occurs in the client terminal management server, management of the client terminal becomes impossible at this point.

  In the third related technology, a server organization having a three-layer structure of a client terminal management server, a network monitoring server, and a management server is provided. In the case where a failure occurs in all of the network monitoring servers, the client terminal management server cannot access the list itself.

  Accordingly, an object of the present invention is to provide a computer management system, a computer management method, and a computer management control program capable of quickly switching to a new manager for managing an agent when a failure occurs on the manager side in a group of computers connected to a network. Is to provide.

  In the present invention, (b) any number of managed computers, and (b) any number of managed computers connected via a network, respectively, Computer management of the managing computer with failure response determining means for determining the content of the failure response processing executed by the managed computer at the time of the failure before the management failure occurs Provide in the system.

  In the present invention, (a) a computer suitable for managing each of these computers on behalf of the own device by communicating with an arbitrary number of computers under the control of the own device via a network. The computer management method includes a management computer discrimination step for discriminating whether or not there is a computer.

  Furthermore, according to the present invention, as a computer management control program, a computer management control program connected to an arbitrary number of managed computers connected via a network can be used as a computer management control program. (B) a determination process for sequentially determining the processing contents to be executed by the managed computer at the time when a failure occurs in the apparatus; and (b) a determination result by the determination process is transmitted to the managed computer. It is characterized by executing notification processing for notification.

  As described above, according to the present invention, the content of the failure handling process executed by the managed computer is determined by communicating with an arbitrary number of computers placed under the management of the own apparatus via the network. It was to be. Therefore, the other computers that have been managed acquire the determination result as a new manager candidate, so that the reliability of the system can be quickly recovered between the managed computers when a failure occurs. In addition, it is possible to shift to a system in which managed computers are managed in an autonomous and distributed manner.

  Next, the best mode for carrying out the present invention will be described in detail.

FIG. 1 shows the basic configuration of the computer management system of this embodiment. The computer management system 100 according to the present embodiment is a first as a specific computer or computer group (hereinafter referred to as a manager) 101 and another computer (hereinafter referred to as an agent) managed by the manager 101. And second agents 102 ₁ and 102 ₂ . The manager 101 and the first and second agents 102 ₁ and 102 ₂ (the number of agents is an arbitrary positive integer) are connected by the network 103.

The manager 101 includes a failure handling determination unit 111 and a failure handling notification unit 112 as processing parts for dealing with the occurrence of a failure. Here, the failure handling determination unit 111 is a processing part that determines the management content for the first or second agent 102 ₁ , 102 ₂ while the manager 101 has not failed. Specifically, it is a processing unit that predetermines when the current manager 101 is alive, assuming that another optimal agent that each agent newly requests to manage when the manager 101 fails becomes the manager. The failure handling notification unit 112 is a processing unit that performs notification 114 of the determination content 113 of the failure handling determination unit 111 to the first and second agents 102 ₁ and 102 ₂ through the network 103 when the manager 101 fails. It is.

With this configuration, the computer management system 100 according to the present embodiment allows each of the agents 102 ₁ and 102 ₂ to be determined other agents (not shown) when a failure occurs in the manager 101. On the other hand, management can be requested immediately. Thus, the management mode with the manager 101 according to the first and second agents 102 _1, 102 ₂ configuration, that you immediately shifts to autonomous decentralized management of the first and second agents 102 _1, 102 ₂ to each other Therefore, the reliability of the system can be quickly recovered in the event of a manager failure.

FIG. 2 specifically shows the configuration of the computer management system of this embodiment. In FIG. 2, the same parts as those in FIG. 1 are denoted by the same reference numerals, and description thereof will be omitted as appropriate. The computer management system 200 has a configuration in which a manager 201 and first and second agents 202 ₁ and 202 ₂ are connected via a network 203.

Among these, the manager 201 includes a management node determination unit 211 corresponding to the failure handling determination unit 111 in FIG. 1 and a management node notification unit 212 corresponding to the failure handling notification unit 112. Here, the management node determination unit 211 determines the optimal management node candidate to be managed on behalf of the first and second agents 202 ₁ , 202 ₂ as each agent as a response to the failure of the manager 201. Part. That is, the optimal management node determined by the management node determination unit 211 indicates not the optimal node managed by a certain agent but the optimal node that manages a certain agent. The management node notification unit 212 is a processing unit that transmits the determination items of the management node determination unit 211 to the first and second agents 202 ₁ and 202 ₂ . In this figure, the first and second agents 202 ₁ and 202 ₂ are shown, but the number of agents constituting the computer management system 200 is not limited to these two.

The first and second agents 202 ₁ and 202 ₂ have the same configuration. Therefore, the configuration of the first agent 202 ₁ will be mainly described, and the description of the second agent 202 ₂ will be omitted as appropriate. For the second agent 202 ₂ , the numerical subscript “1” attached to the reference numerals used to represent the components of the first agent 202 ₁ is replaced with “2”.

The first agent 202 ₁ includes a manager monitoring unit 221 ₁ that monitors the manager 201 and a management node receiving unit 223 ₁ that receives management node information 222 ₁ notified from the manager 201. Monitoring result information 224 ₁ of a manager monitoring unit 221 ₁ is adapted to be inputted to the management request unit 225 _1. When the management request unit 225 ₁ detects the failure state of the manager 201 as a monitoring result, it issues a management request 226 ₁ for its own node to the second agent 202 ₂ as another node.

The first agent 202 _{1 also} includes a management node storage unit 227 ₁ and a management determination unit 228 ₁ . The management node storage unit 227 ₁ stores the reception information 229 ₁ received by the management node reception unit 223 ₁ . The management determination unit 228 ₁ determines this when there is a management request 226 ₁ from the management request unit 225 ₂ of the second agent 202 ₂ as another node.

The manager 201 and the first and second agents 202 ₁ and 202 ₂ described above are naturally computers, and are of course provided with a CPU (Central Processing Unit) and a storage medium. The storage medium stores a control program for the CPU to realize the computer management system 200. Of course, a computer as another node that changes its role when a failure occurs in the manager 201 is similarly provided with a storage medium storing a CPU and a predetermined control program.

FIG. 3 and FIG. 4 show an outline of a system operation that enables managers to transfer management in the computer management system configured as described above. Of these, FIG. 3, as the corresponding fault current manager 201 shown in FIG. 2, the first and second agents 202 _1, 202 of these agents 202 ₁ information optimal other nodes to request the management of _2, 202 ₂ shows pre-preparation operations for dealing with failures until storage. FIG. 4 shows an operation at the time of occurrence of a failure in which when a failure occurs in the current manager 201, the management is requested to another agent.

First, a preparatory operation for the computer management system 200 as a whole to cope with a failure of the manager 201 will be described with reference to FIG. This will be described with reference to FIG. First, the manager 201 uses the management node determination unit 211 to determine an optimal management node other than the own device for the first and second agents 202 ₁ and 202 ₂ when a failure occurs in the own device (step S301). . This determination is performed when the computer management system 200 is activated for the first time or when the configuration of the agent 202 such as the first and second agents 202 ₁ and 202 ₂ is changed. This is because if the configuration of the agent 202 is changed, the optimum management node may also change. The manager 201 may perform this process periodically until a failure occurs.

The manager 201 sends this decision 231 as management node information 222 ₁ , 222 ₂ from the management node notification unit 212 in the own apparatus via the network 203 and notifies the first and second agents 202 ₁ , 202 _2. (Step S302). The notification regarding this optimal management node (candidate) is received by the management node receivers 223 ₁ and 223 ₂ in the first and second agents 202 ₁ and 202 ₂ (step S303). The received information 229 ₁ and 229 ₂ are stored in the corresponding management node storage units 227 ₁ and 227 ₂ (step S304).

As described above, when the configuration of the agent 202 is changed or as a periodic process, when the process from step S301 in FIG. 3 is performed again at a predetermined timing, the optimum management node ( The received information 229 ₁ , 229 ₂ regarding the candidate) is overwritten and saved in the management node storage units 227 ₁ , 227 ₂ .

Next, the process of FIG. 4 will be described with reference to FIG. The manager monitoring units 221 ₁ and 221 ₂ in the first and second agents 202 ₁ and 202 ₂ shown in FIG. 2 always monitor whether a failure occurs in the manager 201 (step S321). When a failure occurs in the manager 201 (Y), the management requesting units 225 ₁ , 225 ₂ of the first and second agents 202 ₁ , 202 ₂ are notified from the management node storage units 227 ₁ , 227 ₂ of the own device. Node information 233 ₁ and 233 ₂ for requesting node management is read (step S322).

If the node information 233 ₁ , 233 ₂ has been successfully acquired (step S323: Y), the management request units 225 ₁ , 225 _{2 make} management requests 226 ₁ , 226 ₂ to the acquired nodes (step S324). . In this case, the management determination unit 228 of the device that is the agent 202 at the present time that is the target of the management request determines whether to perform management (step S325). As a result, when the management determination unit 228 of the device determines (or agrees) that it can be managed (step S326: Y), after notifying the determination result to the requesting side, the agent 202 has a new one. The manager 201 is reached (step S327), and a series of processing ends (end).

For example, it is stored in the management node storage unit 227 ₁ that it is optimal that the second agent 202 ₂ becomes the manager 201 of the first agent 202 ₁ , and the management determination unit 228 _{2 of} the second agent 202 ₂ Suppose that it is determined that this is acceptable. In this case, after the first agent 202 ₁ detects the failure of the manager 201, the second agent 202 ₂ becomes the manager 201 of the first agent 202 ₁ (step S327). In the case of this example, the manager 201 is quickly replaced.

On the other hand, in step S323, the optimal management node candidate is not stored in the management node storage unit 227 ₁ (or the management node storage unit 227 ₂ ) and may not be acquired (step S323: N). In such a case, a management node to be the manager 201 is searched for by a general method conventionally used (step S328). The same applies to the case where it is determined that the candidate agent 202 cannot be managed (step S326: N). In this case, management may be disabled for a relatively long time until a new manager 201 is determined.

  As described above, according to the present embodiment, at the time of failure of the manager 201, it is possible to quickly switch from manager-led management to autonomous distributed management by using the information of the optimal management node that has already been determined. Therefore, there is an effect that the time during which the reliability of the system is lowered can be shortened. Even if the manager 201 determines a candidate to become a new manager, the agent 202 determined as the candidate determines (or agrees) whether to become a manager. It becomes possible.

  FIG. 5 shows the configuration of the computer management system in one embodiment of the present invention. As shown in this simplified diagram, in the computer management system 400 of this embodiment, the first to third nodes 401 to 403 are connected to each other by a common network 404 such as the Internet.

  FIG. 6 shows a common configuration of each node in this embodiment. A computer 411 as a node that can be a manager or an agent includes a manager unit 412 that functions as a manager, an agent unit 413 that functions as an agent, a function switching unit 414 that switches these functions, and the network illustrated in FIG. A network communication unit 415 that communicates with 404 is provided. The computer 411 becomes a manager when the function switching unit 414 causes the manager unit 412 to function. Also, when the function switching unit 414 causes the agent unit 413 to function, it becomes an agent.

  The manager unit 412 includes the management node determination unit 211 and the management node notification unit 212 shown in FIG. The management node notification unit 212 is connected to the network communication unit 415 via the function switching unit 414. Since the management node determination unit 211 and the management node notification unit 212 have been specifically described, description thereof will be omitted.

  The agent unit 413 includes an agent communication unit 421 connected to the network communication unit 415 via the function switching unit 414. The agent communication unit 421 is a part that communicates with the network communication unit 415, and is connected to the management node reception unit 223, the manager monitoring unit 221, the management request unit 225, and the management determination unit 228. The management node storage unit 227 constitutes a partial area of a nonvolatile memory (not shown) in the computer 411 and is connected to the manager monitoring unit 221 and the management request unit 225. The manager monitoring unit 221 and the management request unit 225 are directly connected. Since the components other than the agent communication unit 421 in the agent unit 413 are the same as the respective processing units illustrated in FIG. 2, detailed description thereof is omitted.

  The function switching unit 414 is connected to a function sharing table 416 configured by other areas of the nonvolatile memory described above, and depending on the contents, either of the manager unit 412 or the agent unit 413 is related to the counterpart node. Either one is set to work.

FIG. 7 shows the contents of the function sharing table in the initial state before the failure occurs. This time is defined as a first time t ₁ . The function sharing table 416, at a first time point t _1, e.g. administrator of the computer management system 400 shown in FIG. 5, and assigns node managing a node to be managed respectively. This will be described with reference to FIG.

  In the function sharing table 416, the managed node is a node as an agent, and the managed node is a node as a manager. In this example, the former is an agent and the latter is a manager between the first node 401 and the third node 403. Between the second node 402 and the first node 401, the former is an agent and the latter is a manager. Between the third node 403 and the second node 402, the former is an agent and the latter is a manager.

Therefore, as an example, the relationship between the first node 401 and the third node 403 will be specifically considered. In this example, it is assumed that the third node 403 as a manager generates a failure at a _second time t ₂ after the _first time t ₁ . It is assumed that the third node 403 determines that the second node 402 is the most suitable node for managing the first node 401 as an agent at a time before the _second time t _2. .

  FIG. 8 shows the state of processing of a node serving as a manager on the management side in relation to the partner node. This will be described with reference to FIGS. Since the third node 403 has a function as a manager in relation to the first node 401 in the function sharing table 416, first, a failure occurs in the own device in relation to the first node 401 as an agent. Then, a node that can play a role as a manager is determined (step S501).

  First, the state of node determination will be described. The management node determination unit 211 of the third node 403 as a manager can appropriately communicate with each agent such as the first node 401 as an agent and collect various types of information. As a result, based on the obtained information such as the CPU load and memory usage of each agent, the management node determination unit 211 of the third node 403 first node 401 when a failure occurs in the own device. The most suitable node is determined.

  The timing of this determination may be when monitoring information is collected, or when the load of the third node 403 as a manager is low and it is easy to perform monitoring, and the CPU load of each agent increases rapidly. It may be when the state has changed drastically as in the case of. Of course, various other cases may be used. The timing at which the management node determination unit 211 determines can be changed according to system requirements. For example, when determining when monitoring information is collected, since the latest monitoring information can be used, the information of the management node in the latest state can always be maintained. On the other hand, if the decision is made when the manager load is low, the influence on the original processing operation of the third node 403 can be minimized.

  Returning to FIG. If the third node 403 determines that there is a node that is the optimal manager for the first node 401 (step S502: Y), the node that is the optimal manager is the first agent as the corresponding agent. The node 401 is notified (step S503). And it waits until the timing for the next determination arrives (step S504: N), and if the timing demonstrated previously comes (Y), it will return to the process of step S501 (return). In this way, the node serving as the optimum manager can be changed as appropriate in response to changes in the environment.

  If it is determined in step S502 that there is no optimal manager node (N), an error indicating that the corresponding node does not exist is notified to the first node 401 as the corresponding agent (step S505). . Also in this case, the process proceeds to step S504, and the determination in step S501 is performed at the next timing.

  FIG. 9 shows the flow of processing on the agent side. This will be described with reference to FIGS. In this example, the first node 401 serving as an agent waits for the notification of step S503 or step S505 in FIG. 8 to be received from the third node 403 serving as a manager (step S521). When this notification is received (Y), the management node storage unit 227 of the own apparatus is overwritten with the notification content and updated (step S522). For example, if the third node 403 is notified that the second node 402 is the optimum manager node, the management node storage unit 227 is overwritten with “second node”. If an error notification is received, the management node storage unit 227 is overwritten with “no corresponding node”. For example, the second node 402 is checked by the third node 403 when it is overloaded.

  In addition, the first node 401 as an agent may receive a management request as a manager from another manager (step S523: Y). In this case, the first node 401 determines whether or not the management determination unit 228 of the own device can manage the node designated by the own device as a manager (step S524). This determination is made using various situations such as prediction of the future load state of the own device and whether the own device is already a manager of another node.

  If it is determined that management as a manager is possible (step S525: Y), the management request destination is notified that management is possible (step S526). In this case, since management is executed immediately, the description in the column corresponding to the function sharing table 416 of the own apparatus is changed. For example, when the third node 403 cannot be managed as a manager due to the relationship between the third node 403 and the first node 401, the first node 401 manages the second node 402 as a manager. Suppose that In this case, when the second node 402 receives the management request, the corresponding “node to be managed” in the function sharing table 416 of the own device is rewritten to “second node” (step S527).

  FIG. 10 shows a state where the function assignment table is rewritten from the state of FIG. As can be seen from comparison with FIG. 7, the “node that manages” the first node 401 is changed from the third node 403 to the second node 402. Accordingly, the second node 402 becomes a manager that manages both the first and third nodes 401 and 403. Of course, for example, in the case where the second node 402 lacks eligibility as a manager in relation to the third node 403, the first node 401 is replaced with the manager that manages the third node 403. Such changes may occur in the future.

  Returning to FIG. 9, the description will be continued. If it is determined in step S525 that the first node 401 cannot be managed (N), the management request destination node is notified that the management is impossible (step S528).

  By the way, in the state shown in FIG. 7 in which the third node 403 manages the first node 401, the first node 401 as an agent determines whether the third node 403 functions as a manager. Monitoring is performed by the monitoring unit 221. Specifically, when the manager's monitoring time comes (step S529: Y), for example, the existence of the manager is confirmed in relation to the own device (step S530). For example, if survival is confirmed by responding to a survival confirmation message (step S531: Y), the processing is terminated without doing anything (return).

  On the other hand, if survival is not confirmed (step S531: N), it is necessary to have the manager take the place of another node. Therefore, an optimum management node candidate is read from the management node storage unit 227 of the own apparatus (step S532). At this time, if an error that a node serving as a manager cannot be read does not occur (step S533: N), a management request is notified to the read node (step S534). If a failure occurs in the third node 403 in the previous example, data specifying the second node 402 is read from the management node storage unit 227, and a management request is notified to this.

  On the other hand, if an error has occurred in step S533 (Y), the manager node cannot be determined. Therefore, in this case, a general management node search process is executed (step S535). In the present embodiment, only the first to third nodes 401 to 403 constitute the computer management system 400 for the sake of simplification. However, if more nodes than this constitute the system, the rest This node is also checked whether it can become a managing node. Similarly, when a message indicating that management is impossible is received from the previous node (the second node 402 in this example) that has made the management request in step S534 (step S536: Y), step S535 is also performed. Then, a general management node search process is executed.

  As described above, according to this embodiment, since the node that becomes the manager selects the device that replaces the own device as the node that becomes the next manager, as compared with the case where each agent selects the manager. Adjustment such as selecting one manager among a plurality of candidates becomes unnecessary. In addition, since the agent side node is also freed from the task of manager selection, there is an advantage that the load on the own device and the network can be reduced.

  In the embodiment, it is decided to send a survival confirmation message to the manager. However, it is possible to measure the response time delay from the manager or to receive a report on the manager load status. Of course, other agents may be replaced as managers.

  Further, when determining a node to be a manager from among a plurality of agents, past results such as the ratio of each agent becoming a manager may be stored as data and used as a reference.

  Furthermore, in the embodiment, each node has been described as having a manager function and an agent function as shown in FIG. 6, but a specific node is configured as a node having only one of these functions. May be.

  Furthermore, in the embodiment, the optimum agent that can become a new manager is selected as the candidate, but two or more candidates may be notified with priorities. In this case, the agent with the highest priority is first requested to the manager, and if the consent is not obtained, the request with a lower rank is made. Thereby, the risk when the management request is rejected can be reduced.

It is a principle figure of the computer management system of this Embodiment. It is a block diagram showing concretely the composition of the computer management system of this embodiment. It is a flowchart which shows the manager's advance preparation operation | movement for responding to a failure in this Embodiment. 6 is a flowchart showing an operation on the agent side that requests another agent to perform management when a failure occurs in the present embodiment. 1 is a system configuration diagram showing an outline of a configuration of a computer management system in an embodiment of the present invention. It is the block diagram which showed the common structure of each node of a present Example. It is explanatory drawing showing the content of the function allocation table in the initial state before a failure generate | occur | produces. It is a flowchart showing the mode of processing of the node which becomes the manager of the side which manages in relation to the partner node in the present embodiment. It is a flowchart showing the flow of processing on the agent side in the present embodiment. It is explanatory drawing which showed the state by which the function allocation table was rewritten from the state of FIG.

Explanation of symbols

100, 200, 400 Computer management system 101, 201 Manager 102 ₁ , 202 ₁ First agent 102 ₂ , 202 ₂ Second agent 103, 203, 404 Network 111 Failure response determination unit 112 Failure response notification unit 211 Management node determination Unit 212 management node notification unit 221 manager monitoring unit 225 management request unit 227 management node storage unit 228 management determination unit 401 first node 402 second node 403 third node 411 computer 412 manager unit 413 agent unit 414 function switching unit 416 Function sharing table

Claims (10)

Any number of managed computers, and
A computer on the managed side at the time when a failure occurs in the stage before a failure occurs in the management of the managed computer connected to any number of these managed computers via a network. And a managing computer having a failure response determining means for determining the content of the failure response process executed by the computer.   2. The computer management system according to claim 1, wherein the managing computer includes notification means for notifying the managed computer of the determination result of the failure handling determining means.   The failure response determining means of the managing computer repeats the process of determining a computer that can replace the own device at an interval from the time when the own device becomes the managing computer. 3. The computer management system according to claim 2, wherein the notification means notifies the determination result to the corresponding managed computer every time the failure response determination means determines.   4. The failure handling determining means selects the managing computer from the managed computers as a computer that can replace the managing computer. The computer management system described in any one.   The replaceable computer determined by the failure response determining means receives the request from the managed computer when a failure occurs in the management of the managed computer, and is managed by the own apparatus. The computer management system according to claim 1, wherein the migration is performed by consenting to the migration to the computer on the side.   The replaceable computers determined by the failure handling determining means are a plurality of prioritized computers, and the consent to move to the managing computer is obtained from the one with the higher priority. The computer management system according to claim 5. It comprises a failure response determining step for communicating with an arbitrary number of computers placed under the management of its own device via a network and determining the processing contents to be executed when a failure occurs in each of these computers. Computer management method.   8. The notification step of notifying, via the network, the processing contents to be executed when a failure occurs in a management computer in the failure handling determination step. Computer management method.   The management computer determining step is characterized in that the managing computer repeats the process of determining the content of the failure handling process at intervals, and the determination result is obtained in the notification step each time the determination process is performed. 9. The computer management method according to claim 8, wherein the computer is notified to the corresponding managed computer. A computer that manages these managed computers connected to any number of managed computers via the network,
A determination process for sequentially determining the processing contents to be executed by the managed computer at the time when a failure occurs in the own device; and
A computer management control program for executing notification processing for notifying the managed computer of the determination result of the determination processing.

Images