JP2001325241A - Distributed processing system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique by which a distributed processing system for easily activating a host can be constructed. SOLUTION: Each host is provided with: an automatic connection processing module 34 for performing the processing for starting connection of the host to another host of this distributed processing system and notifying other hosts that the connection is to be started; a connection state reporting/recognition module 33 for notifying the change of a connection state further to the other hosts at the time of being notified from the other optional host within the distributed processing system that the connection state between the hosts within the distributed processing system is changed; and a connection state holding module 35 for holding connection state information indicating the connection state between the plurality of hosts. Also, at least one of the hosts within the distributed processing system is provided with a connection state investigation module 31 for gathering information on the connection state between the hosts from the individual hosts and investigating the connection state.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed processing system comprising a plurality of hosts, and more particularly, to a distributed processing in which messages between hosts are exchanged in accordance with a connection state of a message path established between the plurality of hosts. About the system.

[0002]

2. Description of the Related Art In general, a distributed processing system is constructed so that a plurality of hosts (computers connected to a network) cooperate with each other in order to perform complicated processing in a distributed manner. In order to operate a distributed processing system, a plurality of hosts must usually be started according to a predetermined procedure.

[0003]

However, when the number of hosts is large, there is a problem that it is not always easy to start up according to the start-up procedure.

The present invention has been made to solve the above-mentioned conventional problems, and has as its object to provide a technique capable of constructing a distributed processing system in which a host can be easily started.

[0005]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a distributed communication system in which exchange of messages between hosts is performed in accordance with the connection state of a message path established between a plurality of hosts. Provide a processing system. Each of the plurality of hosts is a connection processing unit that performs a process for the host to start a connection to another host of the distributed processing system and notifies the other host to start a connection. And a connection for notifying another host of the change in the connection state when notified from any other host in the distributed processing system that the connection state between the hosts in the distributed processing system has changed. A status reporting unit; and a connection status holding unit that holds connection status information indicating a connection status between the plurality of hosts. Further, at least one of the plurality of hosts includes a connection state investigating unit that collects a connection state between the plurality of hosts from each host and investigates the connection state.

In this distributed processing system, even if the order of activation of each host is set arbitrarily, the connection status information reflecting the connection status of the activated hosts is updated, so that messages are exchanged between the activated hosts. It is possible to exchange. Therefore, each host can be started in an arbitrary order.

[0007] When the host is disconnected from the distributed processing system, the connection status reporting unit may notify another host of the disconnection. Further, the connection processing unit, when receiving a notification that the connection reporting unit is disconnected from another host, determines whether or not the connection state reorganization is necessary, and if necessary, The connection state may be reorganized.

In this way, even when the host is disconnected from the distributed processing system, messages can be exchanged between the hosts according to the connection state reorganized as necessary. For example, maintenance (maintenance) of a certain host
If it is necessary, the host can be stopped and maintenance can be performed while the entire distributed processing system is kept operating.

The connection processing unit determines whether or not the connection state needs to be reorganized when the connection state reporting unit recognizes a communication abnormality with the connection destination host. The connection state may be reorganized.

In this way, even when an unexpected communication error occurs, messages can be exchanged between the hosts according to the connection state reorganized as necessary.

A management unit for managing generation and maintenance of the connection state checking unit, wherein the generation management unit checks whether or not a plurality of connection state checking units exist in the distributed processing system; When there is a connection state investigating unit, the other connection state investigating unit may be deleted so that only one connection state investigating unit remains in the distributed processing system.

With this configuration, since only one connection state checking unit remains in the distributed processing system, the processing for checking the current connection state can be simplified.

The present invention can be implemented in various modes. For example, a distributed processing system and a method of constructing the same, a host used in the distributed processing system,
The present invention can be realized in the form of a computer program for realizing the functions of those methods or apparatuses, a recording medium on which the computer program is recorded, a data signal including the computer program and embodied in a carrier wave, and the like.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, embodiments of the present invention will be described based on examples in the following order. A. Configuration of distributed processing system: B. Processing at host startup: Processing when host disconnects: Processing at the time of communication failure: Modification:

A. Configuration of Distributed Processing System: FIG. 1 is a conceptual diagram showing the configuration of a distributed processing system as one embodiment of the present invention. This distributed processing system has a configuration in which a plurality of hosts 20A to 20D are connected to each other via a network line 10. As the network line 10, Ethernet (trademark of Fuji Xerox), I
SDN or the like can be used. Further, TCP / IP or the like can be used as a network protocol.

FIG. 2 is a block diagram showing the internal configuration of each host. One host includes an agent 22, a factory distributed object (FDO) 24, a distributed object for communication (CDO) 25, an object connection management unit (OCM) 26, and a remote message communication unit (R).
MC) 28. The agent 22 of each host executes a part of various processes for plate making data used for printing plate making, for example. The communication distributed object 25 has a function of exchanging messages between agents via the remote message communication unit 28. That is, the communication distributed object 25 has the function of transferring a message received from the agent 22 in the same host to the communication distributed object of another host, and the function of transferring the message received from the communication distributed object of the other host to itself. Agent 22 in the host
And a function to pass it on. The factory distributed object 24 is for generating a distributed object 25 for communication as needed. Agent 22 of each host
A message can be exchanged with an agent in another host via the communication distributed object 25 and the remote message communication unit 28. The object connection management unit 26 has a function of managing the connection state of the communication distributed object 25.

Each unit in the host is realized by a CPU (not shown) executing a computer program stored in a memory (not shown) in each host. Such a computer program is provided to the computer in a form stored in a recording medium such as a CD-ROM, or is downloaded to the computer via a computer network such as the Internet.

In this specification, the term "host" means a unit of a unit including an agent 22, distributed objects 24 and 25, an object connection management unit 26, and a remote message communication unit 28. Usually, it is realized on one computer hardware. It is also possible to configure a plurality of hosts on one computer hardware. Agent 2
Instead of 2, a normal client or server can be used.

FIG. 3 is an explanatory diagram showing a message path (connection state) between hosts. In this example, the host 20A
The other three hosts 20 </ b> B to 20 </ b> D are connected in a tree-like manner. The host 20A includes three hosts 20
Three communication distributed objects CDO have been generated to communicate with B, 20C, and 20D. Also, a communication distributed object CDO for performing communication with the host 20A is generated in each of the three hosts 20B, 20C, and 20D. In FIG. 3, the factory distributed object FDO and the object connection management unit OCM
And the remote message communication unit RMC are not shown.

The exchange of messages between agents is as follows.
This is performed according to the route shown in FIG. For example, the second host 2
When a message is sent from the agent B of the first host 20A to the agent D of the fourth host 20D, the message is first transmitted to the communication distributed object CDO (BA) of the second host 20B and the message of the first host 20A. It is sent to the agent A via the distributed object for communication CDO (AB). This message further includes a communication distributed object CDO (A-D) of the first host 20A and a communication distributed object CDO of the fourth host 20D.
(DA) to the agent D. Note that the communication distributed objects CDO that communicate with each other
Have the IP addresses of the hosts of each other and the object names of the distributed objects for communication with each other, and transmit messages using these.

FIG. 4 shows that a new host 20E is a host 20E.
4 shows a message path when connected to B. At this time, a communication distributed object CDO (BE) for communicating with the host 20E is generated in the host 20B, and a communication distributed object CDO (EB) for communicating with the host 20B is also generated in the host 20E. ) Is generated.

FIG. 5 shows a message path when the host 20A is disconnected. "Disconnect Host"
Means disconnecting (logging off) the computer constituting the host from the network. At this time, a distributed object for communication CDO (BC), C for connecting to the hosts 20C and 20D is provided to the host 20B.
DO (BD) is created. Also, the host 20C, 2
A corresponding communication distributed object is also created in 0D.

In the examples shown in FIGS. 3 to 5, one distributed object for communication CDO only exchanges messages between two agents, but instead, one distributed object for communication CDO is used. It is also possible to exchange messages between three or more agents. For example, in the configuration of FIG.
A may have one common distributed object for communication CDO for exchanging messages with the three lower hosts 20B, 20C and 20D.

The connection status of each host is managed by the object connection management unit 26 of each host. FIG. 6 is a block diagram showing an internal configuration of the first object connection management unit 26a in the first host 20A and the second object connection management unit 26b in the second host 20A. The first object connection management unit 26a includes a connection state investigation module 31, an investigation module generation management module 32, a connection state report / cognition module 33, an automatic connection processing module 34, and a connection state holding module 35. ing. Each module executes the following processing.

(1) Connection status check module 31: This module collects information indicating the connection status of each host from each host and checks the connection status. It is sufficient that one connection state investigation module 31 exists in the distributed processing system. Therefore, when the connection state investigation module 31 exists in the first host 20A, the connection state investigation module 31 is not generated for other hosts. Note that the connection state investigation module 31 is disconnected by disconnection of the host 20A.
Is no longer present in the distributed processing system,
One connection state investigation module 31 is generated in another host (for example, the host 20B). Thereafter, if the first host 20A is reconnected, the first and second hosts 2A
The connection state investigation modules 31 are generated at 0A and 20B, respectively. In this case, one of them disappears according to a preset rule. That is, the connection state investigation module 31 stores 1
Only one is allowed to exist. The other modules 32 to 35 described below are each generated for each host.

(2) Investigation module generation management module 32: This module examines whether or not the connection state investigation module 31 exists in the distributed processing system.
When the connection state investigation module 31 does not exist in the distributed processing system, the connection state investigation module 31 is generated in any of the hosts. When a plurality of connection state investigation modules 31 exist in the distributed processing system,
It also has the function of leaving only one and extinguishing the other.

(3) Connection status report / cognition module 3
3: When this module is notified from another host that the connection state in the distributed processing system has changed,
The change of the connection state is notified to another host. Also,
It also has a function of notifying other hosts when the host is disconnected.

(4) Automatic connection processing module 34: This module performs processing for the host to start a connection to another host in the distributed processing system, and notifies the host of the connection destination to start the connection. Has functions. "Start connection" or "Connect"
The term means to create a communication distributed object CDO for exchanging messages and establish a message path.

(5) Connection state holding module 35: This module requests the latest connection state list (described later) indicating the connection state checked by the connection state check module 31, and has a function of holding the received list. Have. FIG.
FIG. 4 is an explanatory diagram showing a connection state between hosts and a connection state list corresponding thereto. Here, the first host 2
A plurality of hosts are connected so as to form a three-layer tree structure with 0A at the top.

The connection status list is a list in which, for each host, the host name of the connection source and the host name of the connection destination are registered. Here, the “host name” is an ID for identifying the host from other hosts. “Connection source” means a host existing in a lower layer in the tree structure, and “connection destination” means a host existing in a higher layer. Note that the host name (host ID) is the IP address of the host (computer) and the factory distributed object F in the host.
DO name. The host IP address is
This is a fixed value assigned in advance. Further, the name of the factory distributed object FDO is common to all hosts in this distributed processing system. That is, the connection state list in FIG. 7 represents the connection state between hosts as shown in FIG.

By the way, the communication distributed object CDO
Is dynamically assigned by the factory distributed object FDO when the communication distributed object CDO is created by the factory distributed object FDO. Therefore, the communication distributed object CDO is:
As shown in FIGS. 3 to 5, it is possible to generate a plurality in a single host. In contrast, there is only one factory distributed object FDO in one host. The name of the factory distributed object FDO is used as the connection state list indicating the connection state between the hosts.

Each host can know the connection state of the distributed processing system at any time by referring to the connection state list held in the connection state holding module 35 in each host.

B. FIG. 8 is an explanatory diagram showing the connection state before and after the connection of the host 20B. In the following, as shown in FIGS. 8A and 8B, the host 20B
Will be described when the PC is activated and connected to the host 20A.

FIG. 9 shows that when the host 20B starts up,
FIG. 14 is a sequence diagram illustrating a process performed by an automatic connection processing module of the host. When the host 20B starts up (step S1), the automatic connection processing module 34 of the host 20B connects the connection state holding module of the connection destination host (hereinafter, referred to as “previous connection host”) to which the host 20B was connected at the time of previous operation. A request is made to 35 to send a connection status list (step S2). Note that the host name (IP address + factory distributed object name) of the “previous connection host” is held in the connection state holding module 35 (or the automatic connection processing module 34) of the host 20B, and this host name is used. Step S
The list request in 2 is executed.

Automatic connection processing module 3 of host 20B
4 judges whether there is a response from the previously connected host (step S3), and if there is no response, selects the next host (step S4). Here, various methods can be adopted as a method of selecting the “next host”. For example, another host connected to the “previous connection host” can be selected from the connection state list at the time of the previous operation. Alternatively, a host having an address closest to the ID address of the “previous connection host” may be selected as the “next host”. In the example of FIG. 9, there is shown a case where there is no response from the host connected to the host 20B last time, and the host 20A is selected as the next host.

Automatic connection processing module 3 of host 20B
4 starts connection processing to the discovered host 20A (step S5). At this time, the factory distributed objects FDO of both hosts respectively generate communication distributed objects CDO (step S6).
The object names of the generated communication distributed objects CDO are mutually notified (step S7). Then, communication is established between these hosts (step S8), and messages can be exchanged.

As described above, when an arbitrary host is started, the automatic connection processing module 34 of that host performs processing for automatically connecting to another host. It is possible to initiate a connection. Therefore, when activating the distributed processing system, it is not necessary to activate each host in a certain order, and the procedure for activating the system can be simplified. Further, the service of the distributed processing can be started when the minimum number of hosts required for the distributed processing is started, so that the productivity can be improved.

FIG. 10 is a diagram showing a sequence of the generation management process of the connection status examination module 31 when the host 20B is connected. When a new host 20B is connected according to the procedure shown in FIG. 9, the investigation module generation management module 32 of this host 20B causes the host 20B
The connection state investigation module 31 is generated in the inside. However, in this example, the connection state investigation module 31 also exists in the host 20A in advance. Therefore, the investigation module generation management module 32 causes one of these two connection state investigation modules 31 to disappear and the other to continue according to the following procedure.

Automatic connection processing module 3 of host 20B
4, when the automatic connection processing to the host 20A is performed (step S21), the host 20A notifies its own connection state report / cognition module 33 of the connection (step S22). The connection status report / cognition module 33 of the host 20A reports the connection to another host (Step S23), and reports the connection to all the generation management modules 32 (Step S24). Upon receiving this report, the generation management module 32 temporarily evacuates its own connection state investigation module 31 and stops its activity (steps S25 and S26). While the connection state investigation module 31 is temporarily evacuated, even if there is a connection state investigation request (list collection request) from another module, the connection state investigation module 31 does not accept the request.

The connection state investigation module 31 of the existing host 20A makes an inquiry by broadcast (simultaneous communication) as to whether or not another generation host has a generation management module (step S27). When there is a response of the existence from the connection state investigation module 31 of the host 20B (step S28), the generation management module 32 of the host 20A
Recognizes that the generation management module 32 exists in another host (step S29). Then, it compares the superiority of its own connection state investigation module 31 with that of another host's connection state investigation module 31 according to a predetermined rule (step S30). Although any rule can be used as the rule, for example, a host having a larger number of times the connection state investigation module 31 has existed in the past may have a higher advantage. In the example of FIG. 10, it is assumed that the connection state investigation module 31 of the host 20A has higher dominance.

At this time, the generation management module 32 of the host 20A is connected to the generation management module 3 of the other host 20B.
2 is issued (step S31). In response to this request, the connection state investigation module 31 of the host 20B disappears (step S32), and the host 20A
Connection state investigation module 31 returns (step S
33).

As described above, the investigation module generation management module 32 performs the connection state investigation so that only one connection state investigation module 31 exists in the distributed processing system even when a new host starts connection. The module 31 is managed. Also, as can be understood from the description of FIG. 10, when the connection state investigation module 31 does not exist in the distributed processing system, the connection state investigation module 31 is generated in one host.

FIG. 11 shows another processing sequence when the host 20B is connected. When the host 20B starts connection to the host 20A according to the procedure of FIG. 9 described above (step S41), the host 20A of the connection destination
The connection status report / cognition module 33 of the host 20B
Is recognized as a connection source, and the host 20A is recognized as having a connection destination (step S42). Host 20A
The connection status report / recognition module 33 notifies the change in the connection status to the connection status report / recognition modules 33 of all the other hosts connected to itself (step S43). The connection status report / recognition module 33 of the host 20C that has received the notification requests the connection status investigation module 31 (present in the host 20A in this example) in the distributed processing system to collect a connection status list ( Step S44). The connection status report of each host /
Since the recognition module 33 does not know on which host the connection state investigation module 31 exists, the recognition module 33 proceeds to step S
Forty-five list collection requests are broadcast to all hosts. Connection status investigation module 31
Executes the connection state list collection process described later in response to this request (step S45).

When the collection of the connection state list is completed, the connection state investigation module 31 returns the list to the connection state report / recognition module 33 of the host 20C that has issued the list collection request (step S46). The connection status report / recognition module 33 that has returned the list causes the connection status holding module 35 to hold the connection status list (step S47).

Similarly, the connection status report / recognition module 33 of the host 20A requests the connection status investigation module 31 to collect a connection status list (step S4).
8) When the list is returned (step S49), the list is held in the connection state holding module 35 (step S50). When the connection status report / recognition module 33 of the host 20B that has started the connection also receives notification from the automatic connection processing module 34 that the connection has been completed (step S51), the connection status list is collected by the connection status investigation module 31. Is requested (step S52), and when the list is returned (step S53), the list is held in the connection state holding module 35 (step S5).
4).

As described above, when the connection of an arbitrary host 20B is started, the connection state list held in the connection state holding module 35 of each host connected to the distributed processing system is updated. Hosts can always exchange messages according to the current connection status (message path) of each host.

FIG. 12 is a diagram showing the sequence of the connection state list collection process executed in step S45 of FIG. Here, a case where a list collection request is sent from the connection status report / recognition module 33 of the host 20C to the connection status investigation module 31 is shown. Host 2
The connection status report / cognition module 33 of 0C first makes an inquiry to the generation management modules 32 of all the hosts by broadcast whether or not the connection status investigation module 31 exists (step S51). The generation management module 32 of each host reports whether or not the connection state investigation module 31 exists in its own host (Step S62). In this example, it is reported that the connection state investigation module 31 exists in the host 20A.

The connection status report / recognition module 33 of the host 20C transmits a list collection request to the connection status investigation module 31 (step S63). Upon receiving this request, the connection state investigating module 31 requests the connection state to the automatic connection processing module 34 of each host in the distributed processing system by broadcasting (step 64), and each automatic connection processing module 34 The current connection information (connection source host name and connection destination host name) of the host is reported (step S65). The connection state investigation module 31 creates and updates a connection state list from the connection information from each host (step S66). And
The connection state investigating module 31 sends the list collection request to the connection state report / recognition module 33 of the host 20C.
, The latest connection status list is returned (step S6).
7). Step S67 in FIG. 12 is the same as step S45 in FIG.

As described above, the connection state investigation module 31
Collects connection information (connection source host name and connection destination host name) indicating the latest connection state, updates the connection state list (connection state information), and distributes the list to each host. Therefore,
Each host can exchange messages according to the connection status list.

In the above description, only one connection state investigation module 31 is generated in the distributed processing system. However, it is assumed that a plurality of connection state investigation modules 31 are generated in the distributed processing system. Is also good.
However, if there is only one connection state examination module 31 in the distributed processing system, there is an advantage that the number of messages exchanged when examining the connection state can be reduced. If there is no change in the connection state, the load of message exchange can be reduced by reusing the existing connection state list.

C. Processing when disconnecting host: FIG.
FIG. 6 is a sequence diagram showing processing when the host 20A is disconnected from the connection state of FIG. When the termination processing of the host 20A (logoff from the network) is performed (step S71), all the hosts 20B to which the host 20A is connected, from the host 20A,
A notification is sent to the connection status report / recognition modules 33 of 20C and 20D to perform disconnection processing (step S72). FIG. 13 shows the operation of the host 20C at this time.

The connection state investigation module 3 of the host 20C
Upon receiving the notification of the disconnection processing from the host 20A, 1 instructs the automatic connection processing module 34 to execute the disconnection processing (step S73). At this time, the automatic connection processing module 34
5 is requested for the connection status list (step S7).
4), the connection state holding module 35 returns a list (step S75). The automatic connection processing module 34 refers to the connection state list and confirms whether or not the connection of the host 20C needs to be reorganized (reconstruction of the connection state) when the host 20A is disconnected (step). S76).
For example, when the host 20A is disconnected from the connection state shown in FIG. 3, the other three hosts 20B, 20C, and 20D become disconnected from each other.
It is determined that C and 20D need to reconfigure the connection state. On the other hand, when the host 20B is disconnected from the connection state of FIG. 3, the hosts 20A, 20C,
Since the connection state of 20D does not change, it is determined that reorganization of the connection state is unnecessary.

When it is determined that the connection state needs to be reorganized, the connection state holding module 35 of the host 20C executes a connection process for the host 20B (step S7).
7). This connection process is substantially the same as that described with reference to FIGS. 9 to 12, and thus the description thereof is omitted here.

Thus, when a host is disconnected,
Since the connection status between hosts is automatically reorganized as needed, even if a host at any level of the tree structure is disconnected, other hosts can exchange messages according to the reorganized connection status. It is possible. Therefore, there is an advantage that the function of the distributed processing system can be maintained to some extent by the remaining host. For example, even when some hosts are disconnected for maintenance (maintenance), there is no need to stop the entire distributed processing system, and there is an advantage that productivity is not excessively reduced.

D. FIG. 14 is a sequence diagram showing a process when the host 20A stops for some reason from the connection state of FIG. In this case, substantially the same processing as that at the time of cutting described with reference to FIG. 13 is performed.

First, when the connection status report / recognition module 33 of the host 20C recognizes a communication abnormality with the host 20A (step S81), the automatic connection processing module 34
Is instructed to execute processing at the time of disconnection (step S82). The processing in steps S82 to S85 is the same as the processing in steps S74 to S77 described with reference to FIG. 13, and a description thereof will be omitted here.

As described above, even when a communication error occurs between the hosts, the connection state between the hosts is automatically re-organized as necessary, so that each host can send a message according to the re-organized connection state. It is possible to exchange.
Therefore, it is possible to prevent the entire system from stopping due to an abnormality in some of the hosts. Further, when the host in which the abnormality has occurred is recovered, it is possible to reconnect the host by the above-described procedure of FIG. 9 to FIG. 12 and return to the original connection state. As described above, in this distributed processing system, even when some of the hosts are stopped or disconnected, the connection state between the other hosts is automatically reorganized, so that the utilization rate of the distributed processing system is increased as much as possible. It is possible.

E. Modifications: The present invention is not limited to the above-described examples and embodiments, and can be carried out in various modes without departing from the gist thereof. For example, the following modifications are also possible. is there.

E1. Modification Example 1 As described with reference to FIG. 6, in the above embodiment, each module in the connection management unit 26 is configured, but the connection management unit 26 is not necessarily classified into such modules. Alternatively, the connection management unit 26 may be configured to realize each function as described above.

E2. Modification 2 In the above embodiment, the agents exchange messages with each other via the distributed object for communication CDO. However, the present invention can be applied to other distributed processing systems.

E3. Modification 3 In the above embodiment, a plurality of hosts are connected so as to form a tree structure. However, the connection state is not limited to this. For example, a plurality of hosts may be sequentially connected in series. . INDUSTRIAL APPLICABILITY The present invention is generally applicable to a distributed processing system in which a plurality of hosts are connected to each other and messages between the hosts are transmitted according to the connection state.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram showing a configuration of a distributed processing system as one embodiment of the present invention.

FIG. 2 is a block diagram showing the internal configuration of each host.

FIG. 3 is an explanatory diagram showing a message path between hosts.

FIG. 4 is an explanatory diagram showing a message path when a new host 20E is added.

FIG. 5 is an explanatory diagram showing a message path when a host 20A disconnects.

FIG. 6 is a block diagram showing an internal configuration of an object connection management unit.

FIG. 7 is an explanatory diagram showing a connection state between hosts and a connection state list corresponding thereto.

FIG. 8 is an explanatory diagram showing a connection state before and after connection of a host 20B.

FIG. 9 is a sequence diagram showing an automatic connection process when a new host 20B is started.

FIG. 10 is a sequence diagram showing a generation management process of the connection state investigation module 31 when the host 20B is connected.

FIG. 11 is a sequence diagram illustrating another process when the host 20B is connected.

FIG. 12 is a sequence diagram showing a connection state list collection process.

FIG. 13 is a sequence diagram showing processing when the host 20A disconnects.

FIG. 14 is a sequence diagram showing processing when a communication failure occurs in the host 20A.

[Explanation of symbols]

 20A-20E: Host 22: Agent 24: Factory distributed object 25: Communication distributed object 26: Object connection management unit 28: Remote message communication unit 31: Connection status investigation module 32 ... Investigation module generation management module 33: Connection status report / Recognition module 34 Automatic connection processing module 35 Connection state holding module

Claims (6)

[Claims] 1. A distributed processing system in which exchange of messages between hosts is performed in accordance with a connection state of a message path concluded between a plurality of hosts, wherein each of the plurality of hosts is such that the host is A connection processing unit that performs a process of starting a connection to another host in the distributed processing system, and notifies the other host that a connection is started, and a connection processing unit from any other host in the distributed processing system. A connection status reporting unit that, when notified of a change in the connection status between the hosts in the distributed processing system, notifies the other host of the change in the connection status, and indicates a connection status between the plurality of hosts. A connection state holding unit that holds connection state information, wherein at least one of the plurality of hosts determines a connection state between the plurality of hosts by each host. Collected from preparative includes a connection status examination unit for performing a survey of the connection state, the distributed processing system. 2. A host for constructing a distributed processing system in which messages between hosts are exchanged in accordance with a connection state of a message path concluded between a plurality of hosts, wherein the host is another host of the distributed processing system. A connection processing unit that performs a process of starting a connection to the host and notifies the other host that a connection is to be started, and a connection processing unit in the distributed processing system from any other host in the distributed processing system. When notified that the connection status between the hosts has changed, a connection status reporting unit that further notifies the other host of the change in the connection status, and collects the connection status between the plurality of hosts from each host. A connection state investigation unit that investigates the connection state; and a connection state holding unit that holds connection state information indicating a connection state between the plurality of hosts. Host, characterized in that. 3. The host according to claim 2, wherein, when the host is disconnected from the distributed processing system, the connection status report unit notifies another host that the host is disconnected. The processing unit, when receiving a notification that the connection reporting unit is disconnected from another host, determines whether or not the connection state needs to be reorganized, and if necessary, changes the connection state. The host that performs the reorganization. 4. The host according to claim 2, wherein the connection processing unit determines whether or not the connection status needs to be reorganized when the connection status reporting unit recognizes a communication error with the connection destination host. A host that determines whether the connection has been established and, if necessary, performs a reorganization of the connection state. 5. The host according to claim 2, further comprising: a management unit that manages generation and maintenance of the connection state investigation unit, wherein the generation management unit includes a plurality of connection states in the distributed processing system. In addition to checking whether there is an investigating unit, if there are a plurality of connecting state investigating units, the other connecting state investigating units are arranged such that only one connecting state inspecting unit remains in the distributed processing system. Annihilate the host. 6. A computer readable recording computer program for realizing a host for constructing a distributed processing system in which a message exchange between hosts is performed in accordance with a connection state of a message path concluded between a plurality of hosts. A recording medium, wherein the computer program performs a process of starting connection to another host of the distributed processing system and a connection process of notifying the other host that connection is to be started. A function, when any other host in the distributed processing system notifies that the connection state between the hosts in the distributed processing system has changed, notifies the other host of the change in the connection state. A connection status reporting function, collecting connection statuses between the plurality of hosts from each host, and A connection status examination function for the status of the survey, a computer readable recording medium having a program for actualizing a connection state holding function of holding the connection state information indicating a connection state between the plurality of hosts computer.