JPH08314875A - Cooperative distributed processing method, distributed shared memory monitoring device, distributed shared memory network tracking device and distributed shared memory network setting supporting device - Google Patents

Abstract

PURPOSE: To provide a cooperative distributed processing method for backing up the lost function of a failed distributed node with a healthy distributed node in the case where a fault is detected in the distributed node in a distributed processing system, and a supporting device for constructing a system. CONSTITUTION: On a distributed shared memory 13, a state monitoring table 15 where distributed system management data are recorded and shared data 14 shared and utilized by the respective distributed nodes 10a to 10d are place. The respective distributed nodes perform a normal processing while performing mutual monitoring by the distributed system management data, and when the fault of the other distributed node is detected, lost function replacement is performed among the normal distribured nodes corresponding to the priority of the execution right of an application task 16 lost by the fault.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a cooperative distributed processing method effectively applied to a distributed processing system using a distributed shared memory system that requires real-time property and fault tolerance, a distributed shared memory monitoring device, and a distributed shared system. The present invention relates to a memory network tracking device and a distributed shared memory network setting support device.

[0002]

2. Description of the Related Art FIG. 25 shows, for example, JP-A-3-19460.
It is a block diagram which shows the conventional cooperative distributed processing system shown by the 1st publication. In the figure, 1a to 1e are control devices as distributed nodes, and 2 is these control devices 1a to 1e.
It is a plant to be controlled by 1e. An input / output unit 3 includes a detector that acquires process information of the plant 2 and an operation unit that operates the plant 2 according to control information. Reference numeral 4 is a transmission path for the input / output unit 3 to exchange arbitrary process information, and 5 is each of the control devices 1a-1.
It is a transmission path for performing local communication between e.

Next, the operation will be described. Control device 1a
1 to 1e periodically execute a plurality of plant control tasks and control the plant 2 via the input / output unit 3. Further, each of the control devices 1a to 1e periodically communicates with each other using the transmission path 5, and the adjacent control devices perform state monitoring. If a time-out is detected in this status monitoring communication, the control device (for example, the control device 1c) of the communication source that has timed out is determined to be in failure, and the adjacent healthy control devices 1b and 1d execute the failed control device 1c. Distributed backup of tasks according to the load balancing method. As a result, the load of the control devices 1b and 1d adjacent to the failed control device 1c increases, so that in the next cycle, the distributed backup of the tasks according to the load distribution method in the control devices 1a and 1e further adjacent to them. I do. In other words, for a failure, the task is shared by the adjacent node first, but the task is gradually spread to nodes further away,
Eventually, the entire system keeps executing the tasks of the failed node in a load-balanced state. Such a system is called a cooperative distributed processing system.

FIG. 26 shows, for example, Japanese Patent Laid-Open No. 6-680.
It is a block diagram for demonstrating the storage method of the distributed shared memory in such a conventional distributed processing system shown by the 47 publication. In the figure, 6a to 6d are nodes such as computers and terminal devices, 7 is a network to which these nodes 6a to 6d are respectively connected to constitute a distributed processing system, and 8a to 8e are circulating in the network 7. It is a data unit to be used.

Next, the operation will be described. Memory information composed of a set of addressed data is made to circulate among the nodes 6a to 6d in a desired order as data units 8a to 8e corresponding to the addresses. In FIG. 26, the data unit sent from the node 6a to the node 6b is 8a,
The data unit sent from the node 6b is 8b, and the node 6 is
The data unit received by c is 8c, the data unit sent from the node 6c to the node 6d is 8d, and the data unit sent from the node 6d to the node 6a is 8e. On the side of each of the nodes 6a to 6d, the data unit that matches the request is identified from the data units 8a to 8e circulating in the network 7 based on the memory access request from the node 6a, and the data unit corresponding to the request is read out. Is transferred to the node of the access request source, and in the write request, the content of the corresponding data unit is rewritten and relayed. Thus, the storage method of the distributed shared memory type system using the network, which can shorten the access time to the distributed shared memory from each of the nodes 6a to 6d, is provided.

Here, the distributed shared memory type system is a system in which the extended memory of each node is connected to the network, and each node holds duplicate information of the distributed shared memory network. Therefore, it is possible to construct a distributed processing system having excellent distributed transparency and fault tolerance. Also, since the value update of the distributed shared memory is transferred to other nodes almost instantly, the speed of distributed processing is increased,
There is an advantage that it is easy to secure real-time responsiveness.

Regarding the shared distributed memory of this type, SCRAMNet of the US company SYSRAN, V of the US
MIC's reflective memory etc. have already been released as products. Such a distributed shared memory card or board can be used for distributed shared memory network settings and CP.
Setting functions such as physical address setting for U can be performed by setting switches, jumper wires and registers. For example, the SCRAMNet has a control register accessible from the CPU, which allows network arbitration setting and interrupt notification setting. You can do it.

[0008]

Since the conventional cooperative distributed processing system is configured as described above, mutual monitoring is performed only locally, and each node can see the status of the entire system. There wasn't. Also, because the shared processing data and the takeover data are communicated periodically one by one,
There was a problem that the processing was heavy. In order to solve these problems, it is possible to introduce a data sharing method using a distributed shared memory network so that the shared data and the monitoring data can be effectively accessed and managed. However, the conventional distributed system using the distributed shared memory lacks the system management and data management functions necessary for constructing the cooperative distributed processing system.

Further, since the conventional distributed shared memory type system only provides basic functions as a distributed shared memory such as data management, when an actual application system is constructed, system debugging, trouble analysis, or There were problems such as lack of means for detecting a failure and a great deal of time and effort required for system construction.

The present invention has been made to solve the above-mentioned problems, and uses distributed shared memory to share processing data and system management data between distributed nodes of a distributed system, and Realize a cooperative distributed processing method of a distributed processing system in which nodes perform normal processing while mutually monitoring the operating status, and when abnormalities occur, substitute functions of abnormal nodes among healthy distributed nodes, and further construct such a distributed processing system. The purpose is to provide various functions required for.

Further, according to the present invention, in a distributed processing system using a distributed shared memory, a distributed shared memory monitoring device for efficiently supporting debugging, trouble analysis, etc. of the application system when constructing the application system, A distributed shared memory network tracking device and a distributed shared memory network setting support device are provided.

[0012]

A cooperative distributed processing method according to a first aspect of the present invention distributes a state management table on a distributed shared memory in which processing data shared by each distributed node is stored. When system management data is recorded, each distributed node performs normal processing while performing mutual monitoring by the distributed system management data, and when a failure of another distributed node is detected, priority is given to the execution right of the application task lost due to the failure. According to the degree, the loss function substitution is performed between the normal distributed nodes.

In the cooperative distributed processing method according to the present invention, a distributed shared memory management table is placed on the distributed shared memory for keyword management, and shared data access management is performed between the distributed nodes via this keyword. Is to do.

In the cooperative distributed processing method according to the present invention, a record management table is placed on a distributed shared memory, and when accessing a record, sharing is performed among distributed nodes based on the state of the semaphore in the record management table. Exclusive control of data record access is performed.

According to a fourth aspect of the present invention, there is provided a cooperative distributed processing method in which a node status monitoring table is placed on the distributed shared memory, and each distributed node writes the operating status symbol of its own node in the node status monitoring table. The distributed nodes refer to the node status monitoring table to mutually monitor the status and detect anomalies.

According to a fifth aspect of the present invention, in the cooperative distributed processing method, a task execution management table is placed on the distributed shared memory to centrally manage application tasks of all distributed nodes, and when a certain distributed node fails. The function of a failed node can be replaced, and a distributed node with a high priority activates the function as an application task.

In the cooperative distributed processing method according to the present invention, a node state monitoring table is placed on the distributed shared memory, and the own node performs mode transition according to the node state monitoring table and a state transition sequence defined in advance. It determines whether or not it is possible and notifies the application task of the determination result.

In the cooperative distributed processing method according to the seventh aspect of the present invention, when a failed distributed node is restored or a new distributed node joins the system, the distributed distributed node is distributed from a healthy distributed node to the distributed node. The contents of the shared memory are transferred to equalize the distributed shared memory of all distributed nodes.

According to another aspect of the present invention, there is provided a distributed shared memory monitoring apparatus which displays a memory reading unit which periodically reads a value of a predetermined area of the distributed shared memory and a value which is read by the memory reading unit. The apparatus is provided with a display control section for displaying periodically.

In the distributed shared memory monitoring apparatus according to the present invention, the display control unit calculates the difference between the value read by the memory reading unit in the previous cycle and the value read in the current cycle. It also has a difference display control unit to be passed to.

According to a tenth aspect of the present invention, there is provided a distributed shared memory network tracing device, wherein a packet acquisition unit for acquiring packet contents on the distributed shared memory network detected by the packet reception detection unit for each reception, and a packet acquisition unit A display control unit for displaying on a display device in the order of reception is provided.

The distributed shared memory network tracking device according to the invention of claim 11 also comprises a packet related information reading unit for passing information not included in the packet related to the memory update to the display control unit. .

The distributed shared memory network tracking device according to the invention of claim 12 is also provided with a parameter setting unit for acquiring the packet condition designated by the user and passing it to the packet acquisition unit and the display control unit.

In the distributed shared memory network setting support apparatus according to the invention described in claim 13, if the parameter designated by the user is the control function setting of the distributed shared memory network setting register, setting information is created and stored in the setting register. For setting and setting reference, a distributed shared memory network setting register reference setting unit for reading setting information from the distributed shared memory network setting register and passing the setting information to the display control unit for controlling the display of the display device is provided.

The distributed shared memory network setting support device according to the invention of claim 14 has a timer,
The display control unit is activated at fixed time intervals.

According to the fifteenth aspect of the present invention, in the distributed shared memory network setting support device, the value read by the distributed shared memory network setting register reference setting unit in the previous cycle and the value read in the current cycle are set. Is also provided, and a difference display control unit for controlling the display of the display device based on the calculated difference.

[0027]

The cooperative distributed processing method according to the invention described in claim 1 is required for a distributed system by using a distributed shared memory network which is a simple and high-speed data sharing function with little complicated protocol and communication delay. It enables efficient sharing of processing data and management data and management of replication. Further, each distributed node writes the status of its own node in its own entry of the status management table on the distributed shared memory and checks the status of other distributed nodes while executing normal processing, that is, application tasks. Then, if an abnormality of another distributed node is detected by this table monitoring, the application task executed by the abnormal node is shared among the sound distributed nodes that can execute the backup, and the entire distributed processing system performs backup. The task execution status of each distributed node is also placed in the state management table on the distributed shared memory, and the healthy distributed node that has the execution format program and data of the lost task and has a high priority is responsible for the succession execution of the lost task. To do. This sharing result is written in the task execution status table. Thus, by placing the shared data and the distributed system management data on the distributed shared memory, a distributed processing system capable of detecting a failed distributed node and backing up its loss function with a healthy distributed node group is realized.

According to a second aspect of the present invention, there is provided a distributed shared memory management table for managing addresses, sizes, and keywords of the distributed shared memory, and each application task is required to secure an area for the distributed shared memory. By calling the search function of the above table by designating a keyword to, the memory access is performed by obtaining the address of the memory area registered with the keyword. Since the distributed shared memory management table is on the distributed shared memory, it is immediately reflected in the distributed shared memory management table of another distributed node when the table is operated. This enables shared data to be managed by keywords between the distributed nodes.

A coordinated distributed processing method according to a third aspect of the present invention is intended to perform access exclusive control using shared data on a distributed shared memory as a record.
After allocating a record with a semaphore in the distributed shared memory, each task performs semaphore check for read / write access to the record. If you want to exclude access from other distributed nodes, lock the semaphore once,
Make semaphore free after operation. If the semaphore is locked, access is denied. Since this record with semaphore is on the distributed shared memory, it is immediately reflected on the table of other distributed nodes when the table operation is performed. This enables exclusive control of record operations between distributed nodes.

According to a fourth aspect of the present invention, in the cooperative distributed processing method, the state monitoring table of each distributed node of the distributed system is placed on the distributed shared memory, and each host periodically or event-drives its own entry in the table. By running the task that writes the status of its own node, the status of each distributed node can be checked by referring to the status monitoring table. In addition, the same task can detect a sudden failure of the distributed node by monitoring the update of the table for timeout, and the abnormality of the other distributed node detected by the task can be interrupted by any other task. In particular, by notifying the distributed failure coping task, it becomes possible to monitor the status of each distributed node and detect anomalies on the distributed shared memory.

In the cooperative distributed processing method according to the fifth aspect of the present invention, a task management table indicating which application task is being executed in each distributed node is placed in the distributed shared memory, and the failure of other distributed nodes is prevented. When the detection notification is sent, the failure coping unit uses the task management table to
Get the application task that the fault distribution node was executing. The failure coping unit can execute its own node among the application tasks lost due to this failure.
That is, an application task having a program and data and having a high priority of its own node is shared. The task assignment result is reflected in the task management table, and the task execution management unit of each distributed node activates the assigned application task. These series of operations are executed in each distributed node, and operate so as to cooperatively complement the processing of the failed distributed node in the entire distributed processing system.

According to a sixth aspect of the present invention, there is provided a cooperative distributed processing method, which is effective when starting and stopping the entire system and when it is necessary to control the order of start / stop or mode transition of distributed nodes. Each distributed node changes the mode state each time it makes a mode transition of its own entry in the state monitoring table on the distributed shared memory, and each distributed node has an order sequence of distributed nodes that can transition in its own start / stop or mode transition. Whether the mode transition is possible is determined by referring to the node state monitoring table, and the state transition task is notified of the determination result by the inter-task communication function such as signal and message, and the state transition task starts or stops the distributed node. Control mode transitions.

According to the cooperative distributed processing method of the invention described in claim 7, the content of the distributed shared memory is refreshed from a healthy distributed node when the failed distributed node is restored or when a new distributed node newly enters the system. By doing so, the same contents are reflected in the distributed shared memory of the restored or newly entered distributed node via the distributed shared memory network. As a result, the shared data in the distributed shared memory of the restored or newly joined distributed node is equalized with the shared data of the healthy distributed node, and the restored or newly joined distributed node starts executing normal processing.

According to an eighth aspect of the present invention, a memory reading unit periodically sets a value of a specified address or area of a distributed shared memory in a distributed processing system in which each distributed shared memory of a distributed node is connected by a distributed shared memory network. By changing the value of the distributed shared memory (the value of the address or the area) that changes from moment to moment in a predetermined cycle, the user can change the value of the distributed shared memory by the application of the local node or another node. Can be monitored.

According to the ninth aspect of the invention, the difference display control section calculates the difference between the value read by the memory reading section in the previous cycle and the value read in the current cycle, and passes the difference to the display control section. , The display on the display device is changed only for those having a change in the previous period and the value.

According to a tenth aspect of the present invention, the packet acquisition unit is a distributed processing system in which the distributed shared memories of the distributed nodes are connected by a distributed shared memory network, and the packet sharing detection unit detects the distributed sharing each time the packet is received. By acquiring the contents of packets on the memory network, passing them to the display controller, and displaying them in the order of reception, it is possible to track the update order of the distributed shared memory on the distributed shared memory network, and the distributed shared memory Make it possible to confirm the legitimacy of the parallel access operation of.

According to the eleventh aspect of the present invention, the packet-related information reading unit acquires the information related to the memory update, which is not included in the packet, and passes it to the display control unit. The related information can be displayed together with the packet contents read by the.

According to the twelfth aspect of the present invention, the parameter setting unit acquires the packet condition designated by the user and transfers the packet condition to the packet acquisition unit and the display control unit so that the packet reading and display control according to the packet condition are performed. It is possible.

The distributed shared memory network setting register reference setting unit according to the thirteenth aspect of the present invention is provided in a distributed processing system in which each distributed shared memory of a distributed node is connected by a distributed shared memory network, and controls the distributed shared memory network. For the distributed shared memory network setting register where the setting information of the function is set, if the parameter specified by the user is the control function setting, create the setting information and set it, and if it is the setting reference, set the setting information from it. By reading out and displaying on the display device,
It facilitates various settings of the distributed shared memory network and reference to the setting information.

The timer according to the invention of claim 14 periodically displays the setting information of the setting register by activating the setting reference function of the reference setting section at a set time cycle.

According to the fifteenth aspect of the present invention, in the difference display control section, the distributed shared memory network setting register reference setting section calculates the difference between the value read in the previous cycle and the value read in the current cycle, and the difference is calculated. Is passed to the display control unit, the display of the display device is changed only for those having a change in the previous period and the value.

[0042]

【Example】

Example 1. An embodiment of the present invention will be described below with reference to the drawings. 1 is a block diagram showing a distributed processing system using a cooperative distributed processing method according to a first embodiment of the present invention.
In the figure, reference numerals 10a to 10d denote distributed nodes including a controller or a computer that form the system.
Reference numeral 1 is a CPU, 12 is a main memory, and 13 is a distributed shared memory. Each of the distributed nodes 10a to 10d has the CPU 1
1, a main memory 12, and a distributed shared memory 13. Further, 14 is each of the distributed nodes 10a to 10d.
15 is a shared data on the distributed shared memory 13 that is shared and used in the above, and 15 is a state monitoring table on the distributed shared memory 13 in which the distributed system management data is stored. further,
In the CPU 11, 16 is a plurality of application tasks of the CPU 11, 17 is a task execution control unit that controls the execution of each of these application tasks 16, 18 records the state of its own node in the state monitoring table 15, and A state monitoring unit that refers to the monitoring table 15 to detect an abnormality in another distributed node, and 19 requests the task execution control unit 17 to perform necessary application task 16 processing when a failure is detected in another distributed node. This is a failure coping unit.

Reference numeral 20 denotes each of these distributed nodes 10a ...
A distributed shared memory network in which the distributed shared memory 13 of 10d is directly connected, and each distributed node 10a-1
The contents of the distributed shared memory 13 of 0d are changed via the distributed shared memory network 20 by each distributed node 10a ...
Independent of the operation of 10d, each distributed node 10a-10d
And the contents of the memory are reflected. The distributed shared memory network 20
Media may be light, electricity or radio waves,
Further, its topology may be any of bus type, ring type, star type, and wireless type.

Next, the operation will be described. Here, since the distributed nodes 10a to 10d are equal to each other and operate in exactly the same way only by changing the reference numerals, the operation in the distributed node 10a will be described below, and the other distributed nodes 10b to 10d will be described. The description is omitted. Now, when the CPU 11 of the distributed node 10a writes to the distributed shared memory 13, the address and contents thereof are transferred to the distributed shared memory 13 of the other distributed nodes 10b to 10d via the distributed shared memory network 20,
The same contents are written at the same address. That is, the distributed shared memory 13 of each of the distributed nodes 10a to 10d is equivalent except for communication delay. The CPU 11 executes a plurality of application tasks 16, and the execution processing of each application task 16 advances by accessing the local data on the main memory 12 and the shared data 14 on the distributed shared memory 13. The task execution control unit 17 controls which application task 16 is executed.

Here, FIG. 2 shows the state monitoring unit 1 of the CPU 11.
8 is a flowchart showing the flow of processing by the failure handling unit 19, the task handling control unit 17, and the task execution control unit 17. In step ST1, the state monitoring unit 18 first starts the state monitoring table 15 of the distributed shared memory 13 at the end of the task or periodically.
The state of the distributed node 10a itself is recorded in. Next, in step ST2, when step ST1 is executed,
Alternatively, the state monitoring table 15 may be set at a specified timing.
And other distributed nodes 10b to 1 in step ST3.
The occurrence of a fault such as an abnormal state of 0d or occurrence of timeout is checked. As a result, the other distributed nodes 10b-
When an abnormality is detected in one of the 10d, for example, the distributed node 10b, in step ST4, the failure coping unit 19 is notified that the distributed node 10b has failed. On the other hand, when no abnormality is detected in the other distributed nodes 10b to 10d, the process directly branches to step ST8 and the process is passed to the task execution control unit 17.

Upon receipt of the failure notification from the state monitoring section 18, the failure handling section 19 calculates whether or not to back up the failed distributed node 10b according to the failure notification in step ST5. Next, in step ST6, the task execution state of the faulty distributed node 10b is checked by referring to the state monitoring table 15, and the application task 16 executed by the faulty distributed node 10b, that is, the execution right of the lost task due to this fault is confirmed. It is checked whether or not the distributed node 10a has the second runner priority. As a result of the check, if the application task 16 has the second runner priority of the execution right, the task execution control unit 17 is notified of the name or identifier of the application task 16 in step ST7. On the other hand, if it does not have the second priority of the execution right, the process directly branches to step ST8 and the process is passed to the task execution control unit 17. In such a case, the following processing is executed by the distributed node 10c or 10d having the second priority of the execution right of the application task 16 lost due to the failure of the distributed node 10b.

Upon receiving the notification of the name or identifier of the lost application task 16 from the failure handling unit 19, the task execution control unit 17 adds the application task 16 to the currently executing application task 16 in step ST8, Task 1
6 is reorganized, and the reorganized application task 16 is recorded in the state monitoring table 15 of the distributed shared memory 13 in step ST9. Next, step ST10
2, the CPU 11 refers to the status monitoring table 15 to determine the application task 16 to be executed, accesses the local data in the main memory 12 and the shared data 14 in the distributed shared memory 13, and then the application task 16 Process. As a result, the CPU 11 executes the process of the lost task requested by the failure handling unit 19. After that, by returning to step ST1 and repeating this series of processes, the application task 1 lost due to the failure of the distributed node 10b.
6, the sound distributed nodes 10a (10c, 10d) back up in a coordinated distributed manner, and the execution of the application task 16 of the entire distributed processing system can be continued without losing the shared data 14.

Example 2. FIG. 3 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to the second embodiment of the present invention. In the figure, 11 is a CPU, 1
3 is a distributed shared memory, 14 is shared data, 15 is a state monitoring table, 16 is an application task, and 20 is a distributed shared memory network, which are portions corresponding to those in the first embodiment denoted by the same reference numerals in FIG. Since only one distributed node 10 is designated by the reference numerals 10a to 10d in FIG. 1, detailed description thereof will be omitted. Further, 21 is a distributed shared memory management table which is set in a memory area of an arbitrary size on the distributed shared memory 13 and is used for management of keywords for access management of the shared data 14, and 22 is this distributed shared memory management table. 21 is a distributed shared memory management table access control unit that controls access to the memory 21.

Next, the operation will be described. When the application task 16 requests the distributed shared memory management table access control unit 22 to access the distributed shared memory management table 21 provided on the distributed shared memory 13 using the shared data 14 as a keyword, the distributed shared memory management table access The control unit 22 refers to the distributed shared memory management table 21 and returns the address and size of the registered specific data of the shared data 14 to the application task 16. This allows the application task 16 to refer to specific data by address. Also when registering a keyword in the distributed shared memory management table 21, the size and the keyword are specified and the distributed shared memory management table access control unit 22 is called.
If the specified size can be secured, the distributed shared memory management table access control unit 22 registers it in the distributed shared memory management table 21. On the other hand, if the specified size cannot be secured, the application task 16 is notified that the registration has failed. The same applies to the case of deregistration. When the distributed shared memory management table 21 is changed, the distributed shared memory network 20 immediately reflects it in the distributed shared memory management tables of other distributed nodes. Therefore, the distributed shared memory management table 21 is not shared between distributed nodes but via this keyword. It is possible to manage access to the data 14.

Example 3. FIG. 4 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a third embodiment of the present invention. Corresponding parts are designated by the same reference numerals as those in FIG. 3 and their explanations are omitted. In the figure, 23 is a record in the shared data 14 in the distributed shared memory 13, and 24 is a record.
Is a record management table 24 for managing exclusive control of each record 23 of this shared data 14 in record units.
Reference numeral 25 is a record management table access control unit provided in the CPU 11 and used when the application task 16 accesses the record 23.

Next, the operation will be described. In the distributed shared memory 13, a record management table 24 that manages exclusive control for each record 23 in the shared data 14 is prepared. The record management table 24 includes the address, size, and semaphore of the record 23. It is an array of entries. Application task 1
6 calls the record management table access control unit 25 when accessing the record 23 of the shared data 14, and the record management table access control unit 25 refers to the record management table 24 and first checks the semaphore of the record 23 to be accessed. To do. As a result, if the semaphore is in the free state, the semaphore is locked, then the record 23 is read and written, and if it is in the semaphore locked state, it waits until the semaphore becomes free. Here, since the record management table 24 is on the distributed shared memory 13, the semaphore is shared by each distributed node. Therefore, the semaphore lock or free operation is immediately reflected in other distributed nodes, and the access exclusive control between the distributed nodes of the same record information is managed.

Example 4. FIG. 5 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a fourth embodiment of the present invention. Corresponding parts are designated by the same reference numerals as those in FIG. 3 and their explanations are omitted. In the figure, reference numeral 26 is a node state monitoring table prepared on the distributed shared memory 13 and in which the operating states of the respective distributed nodes are recorded, and 27 is the operating state of the distributed node 10 itself written in this node state monitoring table 26. At the same time, it is a node state monitoring table access control unit that refers to the node state monitoring table 26 to check the operating states of other distributed nodes and detect abnormal distributed nodes.

Next, the operation will be described. In the distributed shared memory 13, a node status monitoring table 26 is prepared in which the operating status of each distributed node is recorded. The node status monitoring table 26 shows the node number of each distributed node and each operating status symbol. It is an array that is used as an entry. Node status monitoring table access controller 2
When the application task 16 is called by the application task 16 at any predetermined timing such as when the task ends, the operating state of the distributed node 10 itself is written in the node state monitoring table 26. The node status monitoring table access control unit 27 called by the application task 16 uses the node status monitoring table 26.
The operating state of each distributed node is checked, the distributed node in which an abnormality has occurred is detected, and the abnormality detection result is returned to the application task 16. Furthermore, when the application task 16 registers the interrupt notification in the node status monitoring table access control unit 27,
It is notified to the application task 16 by an interrupt notification mechanism such as a signal or an event. The abnormality detection using the node state monitoring table 26 is performed when the abnormal state symbol is written or when the operating state symbol of the distributed node has not been updated for a certain period of time. In this way, by placing the node state monitoring table 26 on the distributed shared memory 13, the network 2
The node status of each distributed node connected to 0 can be centrally managed in one table.

Example 5. In the distributed processing system using the cooperative distributed processing method according to the fourth embodiment shown in FIG. 5, the fifth embodiment uses the node status monitoring table access control unit 27 of the CPU 11 to set the operation status symbol of the distributed node 10 to the node status monitoring table. 26 is separated into a node state monitoring table updating unit 28 that periodically writes to 26 and a node abnormality detecting unit 29 that performs state monitoring and abnormality detection of other distributed nodes and notifies the application task 16 of the result thereof in advance. It was done. FIG. 6 is a block diagram showing an essential part of such a cooperative distributed processing method according to a fifth embodiment of the present invention. The other parts are given the same reference numerals as the corresponding parts in FIG. 5 and their explanations are omitted.

As described above, in the fifth embodiment, the node state monitoring table access control unit 27 is separated into the node state monitoring table updating unit 28 and the node abnormality detecting unit 29, and the node state monitoring table updating unit 28 is used. Shows the distributed node 10 in the node state monitoring table 26 at regular intervals.
Of the operating status symbol of the node status monitoring table 26 and the node status monitoring table updating section 28.
Checking the write cycle timeout.

Example 6. FIG. 7 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a sixth embodiment of the present invention. Corresponding parts are designated by the same reference numerals as those in FIG. 6 and their explanations are omitted. In the figure, 30 is a task execution control unit that manages execution control such as loading, execution, standby, and termination of the application task 16, and 31 is a task that is referred to when the task execution control unit 30 controls execution of the application task 16. It is an execution management table. When receiving the abnormality notification from the node abnormality detection unit 29, the task execution management table 32 receives the task execution management table 3 as necessary.
1 is a failure coping unit for requesting the task execution control unit 30 to process the required application task 16 by updating 1.

Next, the operation will be described. The task execution control unit 30 controls each application task 16 based on the task control timing such as timers, events, and I / O.
It manages the execution control such as loading, executing, waiting, and ending. The status of task execution control by the task execution control unit 30 includes the identification name or identification number of each application task 16, a symbol indicating the task execution mode of the application task 16, and which distributed node can execute the application task 16. It is centrally managed by using a task execution management table 31 having an entry having an execution priority indicating whether or not The task execution management table 31 manages not only the application tasks 16 of the distributed node 10 itself but also the application tasks of other distributed nodes as a whole.

At the task control timing, the task execution control unit 30 of the distributed node 10 determines that if the distributed node having the highest priority of execution right of the controlled application task 16 is its own node, the application task 1
6 control is executed. Further, the failure coping unit 32 receives the abnormality notification from the node abnormality detection unit 29, and among the lost application tasks that the failed distributed node was in charge of executing, the distributed application node having the second highest execution priority of the controlled application task. If the node is its own node, the task execution management table 31 is updated to be in charge of executing the application task. The task execution control unit 30 refers to the task execution management table 31 and executes the control of the application task to back up the control of the assigned application task.

Example 7. FIG. 8 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a seventh embodiment of the present invention. Corresponding parts are designated by the same reference numerals as those in FIG. 6 and their explanations are omitted. In the figure, 33 is a pre-defined state transition sequence of its own node in the distributed node 10, and 34 is an operating state symbol of another distributed node, causing a state transition of its own node along the state transition sequence 33. This is a state transition task for transitioning the state of the distributed node 10 itself if the conditions are met.

Next, the operation will be described. Distributed node 1
The state transition task 34 of 0 is a distributed state, that is, the operating state symbols of other distributed nodes in the node state monitoring table 26 have the predefined distributed node 1
If the condition that causes the state transition of its own node along the state transition sequence 33 of 0 itself is met, the distributed node 10
Transitions its own state. This state transition may be performed independently of the application task 16 or the application task 16 may control the sequence and timing. At the time of this state transition, the operating state symbol of the own node in the node state monitoring table 26 is also updated. Note that this processing may be performed by the node state monitoring table access control unit 27 in the fourth embodiment. In this way, the node status monitoring table 26
The state transition of each distributed node is managed in cooperation with the distributed state by writing the mode state of each distributed node to the node, determining whether the node can make the mode transition, and notifying the application task 16 of the determination result. It is possible to control, and it becomes possible to manage the node start-up / shutdown sequence at the time of starting and shutting down the entire system.

Example 8. FIG. 9 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to an eighth embodiment of the present invention. Corresponding parts are designated by the same reference numerals as those in FIG. 3 and their explanations are omitted. In the figure, 10n is a distributed node that has newly entered the distributed processing system or recovered from a failure. Reference numeral 35 is data including shared data and a state management table on the distributed shared memory 13 of the new distributed node 10n and the operating distributed node 10, and 36 is the new distributed node 10n and the operating distributed task. This is a distributed shared memory equalization task for equalizing the contents of 10 data 35.

Next, the operation will be described. When the distributed node 10n newly enters the system, the data 35 including the shared data and the state management data on the distributed shared memory 13 of the distributed node 10n does not match the data 35 of other distributed nodes 10. Therefore, distributed node 1
Distributed shared memory equalization task 36 in any of 0
To read the contents of the data 35 in the distributed shared memory 13 of the distributed node 10 in the order of addresses, transfer it to the newly entered distributed node 10n, and rewrite the data 35 in the distributed shared memory 13. To go. As a result, the data 35 of the newly entered distributed node 10n
Is refreshed, and the contents of the other distributed node 10
Is equalized with the contents of the data 35 of The same applies to the case where the faulty distributed node is restored. Further, even if the distributed shared memory equalization task 36 is started manually in the corresponding distributed node 10, a timer, an event, an I / O
It may be performed by using O as a trigger, or may be performed based on the notification from the state transition task 34 of the seventh embodiment.

Example 9. 10 is a block diagram showing a distributed shared memory monitoring device according to a ninth embodiment of the present invention.
In the figure, 10 is a distributed node, 11 is a CPU, 12 is a main memory, 13 is a distributed shared memory, and 20 is a distributed shared memory network, which are equivalent to the parts denoted by the same reference numerals in FIG. It is a thing. Further, 40 is a display device such as a display or printer connected to the CPU 11 of the distributed node 10, and 41 is this display device 4.
Distributed shared memory 13 in which the value displayed in 0 is stored
Is a partial area (specified address or area).
Further, in the CPU 11, 42 is a parameter processing unit for processing user parameters, 43 is a timer for setting a display cycle on the display device 40, 44 is periodically activated by the timer 43, and distributed shared memory 13, a data processing unit as a memory reading unit that reads out the value of the specified address or area, that is, the value of the partial area 41, 45 indicates the value read by the data processing section 44 from the partial area 41 of the distributed shared memory 13. It is a display control unit for displaying.

Next, the operation will be described. Also in this case,
The distributed shared memories 13 of a plurality of distributed nodes 10 are directly connected to the distributed shared memory network 20, and each distributed node 10 operates independently. Note that FIG.
Then, the plurality of distributed nodes 10 are replaced by one distributed node 1
It is represented by 0. Here, the distributed shared memory 1
The contents of 3 are written by another node, and
Its value changes independently of the operation of PU11. Therefore, it is necessary to dynamically monitor the state of the distributed shared memory 13 by means independent of the application of the own node. In the ninth embodiment, in order for the system builder (user) to dynamically monitor the contents of the distributed shared memory 13, the value of the specified address or area of the distributed shared memory 13 is displayed on the display device 40.
Is displayed at the specified cycle. The processing procedure for displaying the contents of the partial area 41 of the distributed shared memory 13 will be described below with reference to the flowchart shown in FIG.

When the processing is started, the parameter processing unit 4
In step ST11, 2 acquires the address of the partial area 41 to be displayed in the distributed shared memory 13 and the display cycle of the operation parameter designated by the user. The acquired address is given to the data processing unit 44, and the display cycle is set in the timer 43. Next, step ST1
In 2, the data processing unit 44 is activated to access the partial area 41 of the distributed shared memory 13 based on the specified address, read the stored value, and pass it to the display control unit 45 together with the address. Next, in step ST13, the display control unit 45 displays the received address of the partial area 41 and its value on the display device 40 such as a display or a printer.

Next, in step ST14, it is determined whether or not the processing of displaying the contents is completed. As a result, if the processing is not completed, the process proceeds to step ST15 to set the timer 43, and at step ST16, the timer wait state is entered. When the timer 43 times out, the process returns to step ST12, the data processing unit 44 is activated again, and the series of processes described above is repeated until it is detected in step ST14 that the process of displaying the contents is completed. As a result, the value of the predetermined partial area 41 of the distributed shared memory 13 is periodically displayed on the display device 40 in accordance with the display cycle designated by the user.

FIG. 12 is an explanatory diagram showing a display example of the display device 40 thus displayed. The 16-byte contents of the partial area 41 of the distributed shared memory 13 is 1 in each row.
It is displayed in hexadecimal, and the first address of 16 bytes is displayed at the beginning of each line. One or more such lines are displayed at once. Any number of bytes can be displayed in one line, and the display method can be selected from bit string, quaternary, octal, ASCII display and the like. Further, the head address does not always have to be displayed.

Thus, the specified distributed shared memory 1
Since the value of the address 3 or the value of the area (partial area 41) is displayed on the display device 40 at a specified cycle, the user can dynamically monitor the value change of the distributed shared memory 13 by the application of the own node or another node. Is possible.

Example 10. FIG. 13 is a first embodiment of the present invention.
FIG. 11 is a block diagram showing a distributed shared memory monitoring device according to 0, corresponding portions are assigned the same reference numerals as in FIG. 10, and description thereof will be omitted. In the figure, 46 is a difference display control unit for supporting the difference display in the display control unit 45, and 47
Is a buffer for storing the value of the partial area 41 in the previous cycle. The tenth embodiment differs from the ninth embodiment in that the difference display control unit 46 and the buffer 47 are provided.

Next, the operation will be described. The data processing unit 44 accesses the partial area 41 of the distributed shared memory 13 and sends the read value to the display control unit 45, and the display control unit 4
5 transfers it to the difference display control unit 46. Here, the buffer 47 stores the value of the partial area 41 read in the previous cycle, and the difference display control unit 46 stores the value sent from the display control unit 45 in the buffer 47. The value of the previous cycle is compared and the difference is returned to the display control unit 45. The display control unit 45 controls the display device 40 so as to change the display of only this difference. Therefore, the display is changed only in the partial region 41 of the distributed shared memory 13 in which the value is changed from the previous period. In this way, by printing or changing only the changed part, the screen can be updated at high speed and the flicker is eliminated, making it easier for the user to read. Also, by performing reverse display or highlighted display, only the changed part can be easily understood. You can also

Example 11. FIG. 14 is a first embodiment of the present invention.
FIG. 11 is a block diagram showing a distributed shared memory network tracking device according to No. 1, and corresponding portions are given the same reference numerals as those in FIG. 10 and description thereof is omitted. In the figure, 48 is a distributed shared memory update packet (packet) flowing through the distributed shared memory network 20, and 49 is a specific area on the distributed shared memory 13 in which the distributed shared memory update packet 48 is stored. Reference numeral 50 is a packet reception detection unit that detects reception of the distributed shared memory update packet 48, and 51 is a packet acquisition unit that reads the distributed shared memory update packet 48 from the specific area 49 on the distributed shared memory 13 and acquires the packet contents. There is a display control unit 52 for displaying the content of the read distributed shared memory update packet 48 on the display device 40.

Next, the operation will be described. Also in this case,
The distributed shared memories 13 of a plurality of distributed nodes 10 are directly connected to the distributed shared memory network 20, and each distributed node operates independently. Here, in the distributed processing system using the distributed shared memory 13, even if each distributed node 10 operates as designed, depending on the memory writing order of each distributed node 10 operating in parallel, the entire system may operate as designed. There are times when you don't. For such trouble analysis, in the eleventh embodiment, the distributed shared memory update packet 48 of the distributed shared memory network 20 is detected for each reception, and the contents of the received packet are displayed on the display device 40 in the order of reception. The processing procedure for displaying the content of the received distributed shared memory update packet 48 will be described below with reference to the flowchart shown in FIG.

First, the distributed shared memory update packet 48 flowing through the distributed shared memory network 20 is sent to the distributed node 1
It is read to the distributed shared memory 13 of 0 and stored in the specific area 49 on the distributed shared memory 13. As a result, the memory value changing process of the distributed shared memory 13 is performed.
The packet reception detection unit 50, in step ST21,
Whether or not the distributed shared memory update packet 48 is received is monitored. When the packet reception detection unit 50 detects the reception of the distributed shared memory update packet 48, the packet acquisition unit 51 proceeds to step ST22 and the distributed shared memory update packet 48 stored in the specific area 49 on the distributed shared memory 13 is detected. Is read and the packet information is passed to the display control unit 52 to wait for the next packet reception. In addition,
This packet information includes the value and address of the distributed shared memory 13. Next, the display control unit 52 performs step ST2.
3, a display device 4 such as a display or a printer
The content of the distributed shared memory update packet 48 is displayed at 0. This series of processes is repeated until the end of the process is detected in step ST24.

FIG. 16 is an explanatory diagram showing a display example of the display device 40 thus displayed. The memory update address and the update value are displayed in one line in each line. Although depending on the packet control information, the packet transmission source node number, the packet checksum, etc. can be displayed as supplementary information, or the user can specify not to display them. In the illustrated example, the packet transmission source node number is also displayed as this additional information. Also, the display method can be selected from bit strings, quaternary, octal, hexadecimal and the like. The packet control information is information indicating priority, interrupt information, and the like.

In this way, the distributed shared memory update packet 48 of the distributed shared memory network 20 is detected each time it is received, and its contents are displayed in the order of reception on the display device 40. The distributed shared memory update order can be traced, and the user can confirm the correctness of the parallel access operation of the distributed shared memory 13.

Example 12 FIG. 17 is a first embodiment of the present invention.
15 is a block diagram showing a distributed shared memory network tracking device according to No. 2, in which the corresponding portions are given the same reference numerals as in FIG. 14 and their description is omitted. In the figure, 53 is the main memory 1
2 is an update time, 54 is an area in the distributed shared memory 13 in which the value before the memory update is stored, 55 is a packet such as the value before the memory update stored in the update time 53 and the area 54 Packet-related information that is not included in the above, but is related to the memory update, adds it to the information read by the packet acquisition unit 51, and adds it to the display control unit 52.
It is a packet related information reading unit to be passed to the.

Next, the operation will be described. The packet related information reading unit 55 uses the distributed shared memory update packet 48.
However, the packet acquisition unit 51 acquires information related to the memory update, such as the value before the memory update stored in the area 54 of the distributed shared memory 13 or the update time 53, which is not included in the distributed shared memory. The information is sent to the display control unit 52 together with the information of the distributed shared memory update packet 48 read from the specific area 49 on 13. The display control unit 52 displays information related to the memory update on the display device 40 in addition to the information of the distributed shared memory update packet 48. Thus, by displaying the information related to the memory update in addition to the information of the distributed shared memory update packet 48,
In the distributed shared memory network 20, when a plurality of distributed nodes 10 are writing data to a specific area on the distributed shared memory 13, if there is a place where the memory update time does not change, the distributed node 10 fails. It is possible to detect a failure, such as determining that there is a failure.

Example 13. FIG. 18 shows a first embodiment of the present invention.
15 is a block diagram showing a distributed shared memory network tracking device according to No. 3, in which the corresponding portions are given the same reference numerals as in FIG. 14 and their description is omitted. In the figure, reference numeral 56 is a parameter setting unit that acquires a packet condition designated by the user and passes it to the packet acquisition unit 51 and the display control unit 52. The packet acquisition unit 51 and the display control unit 52 acquire the parameter setting unit 56. It differs from those shown in FIG. 14 in that it operates according to the above conditions.

Next, the operation will be described. The parameter setting unit 56 acquires a packet condition designated by the user, such as an update packet of a specific address or a packet from a specific node, and passes the parameter value to the packet acquisition unit 51 and the display control unit 52. . The packet acquisition unit 51 reads the distributed shared memory update packet 48 according to the packet condition based on the received parameter value, and sends the packet information to the display control unit 52. The display control unit 52 controls the display of the packet information from the packet acquisition unit 51 on the display device 40 according to the packet condition based on the received parameter value. In this way, by reading the distributed shared memory update packet 48 and controlling the display of the display device 40 in accordance with the packet conditions, in the distributed shared memory network 20, a plurality of distributed nodes 10 can execute specific areas on the distributed shared memory 13. It becomes possible to know the memory update order of a specific distributed node when writing data to.

Example 14. FIG. 19 shows the first embodiment of the present invention.
FIG. 11 is a block diagram showing a distributed shared memory network setting support device according to No. 4, in which the corresponding portions are given the same reference numerals as in FIG. 10 and their description is omitted. In the figure, 57 is a CPU
11 is a distributed shared memory network setting register in which the setting information of the control function of the distributed shared memory network 20 is set, and 58 is also arranged in the CPU 11 and each bit of the distributed shared memory network setting register 57 Is a distributed shared memory network setting register profiler that describes the meaning of and the relationship between bits. Reference numeral 59 denotes a distributed shared memory network hardware control unit including the distributed shared memory network setting register 57 and the distributed shared memory network setting register profiler 58. Also, 60
Refers to the distributed shared memory network setting register profiler 58 by acquiring parameters for control function setting and setting reference designated by the user, and according to the acquired parameters, the distributed shared memory network setting register 5
7 is a distributed shared memory network setting register reference setting unit that performs control function setting and setting reference processing of FIG. 6
Reference numeral 1 denotes a display control unit for displaying on the display device 40 the information sent from the distributed shared memory network setting register reference setting unit 60 at the time of setting reference.

Next, the operation will be described. Also in this case,
The distributed shared memories 13 of a plurality of distributed nodes 10 are directly connected to the distributed shared memory network 20, and each distributed node operates independently. Here, the distributed shared memory card or the distributed shared memory board can perform setting functions such as setting of the distributed shared memory network 20 and setting of a physical address for the CPU 11 by using switches, jumper wires, or registers. For example, SCRAMNet can perform network arbitration setting and interrupt notification setting by a control status register (hereinafter referred to as CSR) accessible from the CPU. This Example 14 is
Of the setting functions, those that can be set from the CPU 11,
That is, the present invention provides a function that allows the user to easily refer to or set the setting status of a function that can be set by software.

The CPU 11 has a distributed shared memory network setting register 57 in which setting information of the control function of the distributed shared memory network 20 is set. By referring to and setting the distributed shared memory network setting register 57, the distributed shared memory 13 and It is possible to set the control function of the distributed shared memory network 20. For example,
In SCRAMNet, the distributed shared memory 13 and the distributed shared memory network 20 are implemented by CSR. More specifically, there are protocol selection and data filtering functions. It should be noted that this data filter function means that when a certain data is written in a specific area on the distributed shared memory 13, if the data is the same as the data written before that, another distributed node This is a function in which packets are not transmitted to 10.

The procedure for referring to and setting the setting information of the control function of the distributed shared memory network 20 will be described below with reference to the flowchart shown in FIG. First,
Distributed shared memory network setting register reference setting unit 6
0 acquires a parameter for setting reference and control function setting designated by the user and passes it to the display control unit 61, and in step ST31, it identifies whether the acquired parameter is setting reference or control function setting. As a result, if the parameter instructed by the user is the setting reference, the process proceeds to step ST32, and the distributed shared memory network setting register reference setting unit 60 follows the distributed shared memory network according to the user-specified parameter (eg register number). The register value is read from the setting register 57. At that time, the distributed shared memory network setting register profiler 58, which describes the meaning of each bit of the distributed shared memory network setting register 57 and the relationship between the bits, is referred to, the comment and definition of the setting state are acquired, and the shared sharing is performed. It is passed to the display control unit 61 together with the value obtained from the memory network setting register reference setting unit 60. Next, in step ST33, the display controller 61
Displays the register value received from the distributed shared memory network setting register reference setting unit 60 on the display device 40 such as a display or printer together with the comment and definition of the setting state.

On the other hand, if the identification result in step ST31 is the setting, the process proceeds to step ST34, where the distributed shared memory network setting register reference setting unit 60 refers to the distributed shared memory network setting register profiler 58 to refer to the distributed shared memory. It is checked where and what value is set in the network setting register 57, and the user parameter is converted into the setting register and its value to create setting information. Next, in step ST35, the distributed shared memory network setting register reference setting unit 60 sets the control function in the distributed shared memory network setting register 57 based on the created setting information. If there is an instruction from the user, the distributed shared memory network setting register reference setting unit 60 reads the result set in the distributed shared memory network setting register 57, and the display control unit 61 controls the distributed shared memory network setting register profiler. Display device 40 with reference to 58
To be displayed.

FIG. 21 is an explanatory diagram showing an example of the display of the display device 40 thus displayed. The setting function items and the current settings are displayed on each line. The display item can be selected by the user's instruction. Further, another display example is shown in FIG. In this case, all register information is displayed regardless of the set function item. The display format can be specified by the user such as a bit string, quaternary, octal, or hexadecimal.

As described above, the setting information of the control function of the distributed shared memory network 20 set in the distributed shared memory network setting register 57 is read according to the instruction of the user, converted and displayed in a format easy for the user to understand, and By setting the control function in the distributed shared memory network setting register 57 according to the instruction, the user can easily confirm the setting state and set the setting function of the distributed shared memory network 20 that can be set by software. A distributed shared memory network setting support device that can be realized can be realized.

Example 15. FIG. 23 is a first embodiment of the present invention.
FIG. 20 is a block diagram showing a distributed shared memory network setting support device according to No. 5, corresponding parts are assigned the same reference numerals as in FIG. 19, and description thereof will be omitted. In the figure, reference numeral 62 is a timer for setting a display cycle when the display controller 61 periodically displays the control device 40. The fifteenth embodiment, except for the register setting function of the fourteenth embodiment,
This timer 62 is added to the one having only the reference function.

Next, the operation will be described. The distributed shared memory network setting register reference setting unit 60 sets, in the timer 62, an activation period for periodically activating the display control unit 61. The display control unit 61 is activated according to the cycle set by the timer 62, and each time the display control unit 61 displays information such as the register value passed from the distributed shared memory network setting register reference setting unit 60 on the display device 40. In this way, the timer 62 periodically causes the display controller 6 to
Since 1 is activated, the register setting status is periodically displayed. When the information is changed according to the user instruction parameter, the information is passed to the display control unit 61 each time the information is changed.

Example 16. FIG. 24 shows the first embodiment of the present invention.
FIG. 24 is a block diagram showing a distributed shared memory network setting support device according to No. 6, in which the corresponding portions are given the same reference numerals as in FIG. 23 and their description is omitted. In the figure, 63 is a difference display control unit for controlling the difference display in the display device 40, and 64 is a buffer for storing the value of the previous cycle. The sixteenth embodiment is a display controller 61 of the fifteenth embodiment.
Is replaced with the difference display control unit 63 and the buffer 64.

Next, the operation will be described. In the cycle display described in the fifteenth embodiment, the difference display control unit 63 stores the value in the previous cycle in the buffer 64, and at the time of display, the value of the previous cycle stored in the buffer 64 and the current cycle value are stored. The amount of change from the cycle value is calculated, and only the changed amount is passed to the display control device 61 and displayed on the display device 40. As described above, only the changed register setting state can be displayed by the changed amount with the value of the previous cycle calculated by the difference display control unit 63, and high-speed and flicker-free screen display can be performed. Further, by displaying and highlighting only the change information according to the change from the difference display control unit 63, a display that is easier to read can be performed.

[0091]

According to the first aspect of the present invention, each distributed node refers to the state management table placed on the distributed shared memory, and normal processing is performed while mutual monitoring is performed by the distributed system management data. When a failure of another distributed node is detected, the loss function substitution is performed between the normal distributed nodes according to the priority of the execution right of the application task lost due to the failure. In a system that requires real-time performance and fault tolerance, it is possible to efficiently perform normal distributed processing and cooperative operation at the time of failure without performing complicated communication management and distributed control.

According to the second aspect of the present invention, a memory area of arbitrary size on the distributed shared memory is used as a distributed shared memory management table for keyword management, and the shared data is referenced by referring to this distributed shared memory management table. Since it is configured to perform the access control of the shared data, there is an effect that the access control of the shared data can be performed through the keyword between the distributed nodes regardless of the address.

According to the third aspect of the invention, the record management table in which the entries including the semaphore of the record of the shared data are arranged is placed in the distributed shared memory, and the shared data is based on the semaphore state of this record management table. Since it is configured to access the record of the above, there is an effect that exclusive control of the record access of the shared data between the distributed nodes becomes possible.

According to the invention described in claim 4, each distributed node writes the operating status symbol of its own node in the node status monitoring table placed on the distributed shared memory, and each distributed node stores the node status monitoring table. Since the configuration is referred to, it is possible to monitor the status of each distributed node and detect anomalies on the node status monitoring table.

According to the invention described in claim 5, the application execution task of each distributed node is managed by the task execution management table placed in the distributed shared memory, and when a certain distributed node fails, the function of the failed node is replaced. Since the distributed node that is possible and has a high execution priority is configured to activate the function as an application task, a distributed processing system that operates so as to cooperatively complement the processing of the failed distributed node in the entire system is provided. There is an effect to be obtained.

According to the invention described in claim 6, according to the node state monitoring table placed on the distributed shared memory and the state transition sequence defined in advance, it is determined whether or not the own node is capable of mode transition, and the determination is made. Since I configured it to notify the application task of the result,
The state transition of each distributed node can be managed and controlled in cooperation with the distributed state, and there is an effect that it is possible to manage the node start / stop sequence at the time of starting and shutting down the entire system.

According to the invention described in claim 7, when the distributed node recovers from a failure or when it newly enters the system, the contents of the distributed shared memory are transferred from the healthy distributed node to the distributed node. Since it is configured, there is an effect that the contents of the distributed shared memory of all distributed nodes can be easily equalized.

According to the invention described in claim 8, the address or area of the distributed shared memory is designated, and the value of the address or area is periodically read and displayed on the display device. The changing contents of the distributed shared memory can be displayed at a predetermined cycle, and the system builder (user) can dynamically monitor the change of the value of the distributed shared memory by the application of its own node or another node. There is an effect that a possible distributed shared memory monitoring device can be obtained.

According to the invention described in claim 9, the display control of the display device is performed by passing the difference between the value read by the memory reading section in the previous cycle and the value read by the current cycle to the display control section. Since it is configured so that it is possible to change the display only for those that have changed in value from the previous cycle, there is an effect that printing and screen updating can be performed at high speed.

According to the invention described in claim 10, the reception of the packet on the distributed shared memory network is detected,
Since the received packet contents are displayed on the display device in the order of reception, it is possible to trace the update order of the distributed shared memory on the distributed shared memory network, and to verify the parallel access operation of the distributed shared memory. There is an effect that a distributed shared memory network tracking device capable of easily confirming the property is obtained.

According to the eleventh aspect of the present invention, the information not included in the packet but related to the memory update is also displayed on the display device together with the packet content read by the packet acquisition unit. In a shared memory network, when multiple distributed nodes are writing data to a specific area in the distributed shared memory, if there is no change in the memory update time, it means that the corresponding distributed node has failed. There is a detectable effect.

According to the twelfth aspect of the present invention, the packet condition designated by the user is acquired, and the reading and display control of the packet are performed according to the packet condition. Therefore, in the distributed shared memory network. When a plurality of distributed nodes are writing a specific area on the distributed shared memory, the user can know the memory update order of the specific distributed node.

According to the thirteenth aspect of the present invention, if the parameter specified by the user is the control function setting, setting information is created and set in the distributed shared memory network setting register, and if the setting is referred to, the distributed shared memory is set. Since the setting information read out from the network setting register is displayed on the display device, the user can easily perform various settings of the distributed shared memory network and refer to the setting information. There is an effect that can be obtained.

According to the fourteenth aspect of the present invention, since the display control unit is configured to be activated by the timer at a predetermined cycle, the setting information set in the distributed shared memory network setting register is periodically displayed. This makes it possible to obtain a distributed shared memory network setting support device that is easier to use.

According to the fifteenth aspect of the present invention, the distributed shared memory network setting register reference setting unit controls the display of the display device according to the difference between the value read in the previous cycle and the value read in the current cycle. Since it is configured, it is possible to change the display only for those that have changed in value from the previous cycle, and there is an effect that printing and screen updating can be performed at high speed.

[Brief description of drawings]

FIG. 1 is a block diagram showing a distributed processing system using a cooperative distributed processing method according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing a flow of processing in the above embodiment.

FIG. 3 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a second embodiment of the present invention.

FIG. 4 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a third embodiment of the present invention.

FIG. 5 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a fourth embodiment of the present invention.

FIG. 6 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a fifth embodiment of the present invention.

FIG. 7 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a sixth embodiment of the present invention.

FIG. 8 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to a seventh embodiment of the present invention.

FIG. 9 is a block diagram showing a main part of a distributed processing system using a cooperative distributed processing method according to an eighth embodiment of the present invention.

FIG. 10 is a block diagram showing a distributed shared memory monitoring device according to a ninth embodiment of the present invention.

FIG. 11 is a flowchart showing the flow of processing in the above embodiment.

FIG. 12 is an explanatory diagram showing a display example in the above embodiment.

FIG. 13 is a block diagram showing a distributed shared memory monitoring device according to Embodiment 10 of the present invention.

FIG. 14 is a block diagram showing a distributed shared memory network tracking device according to Embodiment 11 of the present invention.

FIG. 15 is a flowchart showing the flow of processing in the above embodiment.

FIG. 16 is an explanatory diagram showing a display example in the above embodiment.

FIG. 17 is a block diagram showing a distributed shared memory network tracking device according to Embodiment 12 of the present invention.

FIG. 18 is a block diagram showing a distributed shared memory network tracking device according to Embodiment 13 of the present invention.

FIG. 19 is a block diagram showing a distributed shared memory network setting support device according to a fourteenth embodiment of the present invention.

FIG. 20 is a flowchart showing the flow of processing in the above embodiment.

FIG. 21 is an explanatory diagram showing a display example in the above embodiment.

FIG. 22 is an explanatory diagram showing another display example according to the above-mentioned embodiment.

FIG. 23 is a block diagram showing a distributed shared memory network setting support device according to a fifteenth embodiment of the present invention.

FIG. 24 is a block diagram showing a distributed shared memory network setting support device according to a sixteenth embodiment of the present invention.

FIG. 25 is a block diagram showing a conventional cooperative distributed processing system.

FIG. 26 is a block diagram showing a storage method of a shared distributed memory in a conventional distributed processing system.

[Explanation of symbols]

10, 10a to 10d, 10n distributed nodes, 13 distributed shared memory, 14 shared data, 16 application tasks, 20 distributed shared memory network, 21
Distributed shared memory management table, 23 records, 24
Record management table, 26 node state monitoring table, 31 task execution management table, 33 state transition sequence, 40 display device, 41 partial area (specified address or area), 44 data processing section (memory reading section), 45, 52 , 61 display control unit, 46, 6
3 difference display control unit, 48 distributed shared memory update packet (packet), 50 packet reception detection unit, 51 packet acquisition unit, 55 packet related information reading unit, 5
6 parameter setting unit, 57 distributed shared memory network setting register, 60 distributed shared memory network setting register reference setting unit, 62 timer.

Claims (15)

[Claims] 1. A distributed shared memory of a plurality of distributed nodes is connected by a distributed shared memory network, and the content change of each distributed shared memory is communicated to the distributed shared memory of another distributed node independently of the operation of the distributed node. In the distributed processing method of the distributed processing system in which the memory contents are reflected,
On the distributed shared memory, shared data shared by each distributed node and a state management table in which distributed system management data are recorded are placed, and each distributed node mutually monitors by the distributed system management data of the state management table. When a failure of another distributed node is detected while performing normal processing, the loss of the failed distributed node between normal distributed nodes is performed according to the priority of the execution right of the application task executed by the failed distributed node. A cooperative distributed processing method characterized by substituting functions. 2. A distributed shared memory management table for managing a keyword is placed on the distributed shared memory, the distributed shared memory management table is referred to based on an access request by the keyword from an application task, and the distributed shared memory management table is distributed between the distributed nodes. The shared distributed processing method according to claim 1, wherein access management of shared data is performed via a keyword. 3. The address of the record in the shared data,
A record management table, which is an array of entries consisting of a size and a semaphore, is placed on a shared distributed memory, the record management table is referenced when accessing the record, and the same record information is distributed based on the state of the semaphore of the record. The cooperative distributed processing method according to claim 1, wherein access exclusive control between nodes is managed. 4. A node state monitoring table, which is an array having the node number of each distributed node and the operating state symbol of each distributed node as an entry, is placed on a shared distributed memory,
The operating state symbol of the self node is written in the node state monitoring table at a predetermined timing, the state of each distributed node is monitored and an abnormality is detected by referring to the node state monitoring table, and it is notified to a designated application task. The cooperative distributed processing method according to claim 1, characterized in that. 5. An entry having an identification name or identification number of an application task, a symbol indicating a task execution mode of the application task, and an execution priority indicating which distributed node can execute the application task. , A task execution management table for centrally managing application tasks of all distributed nodes is placed on a shared distributed memory, and when a failure of another distributed node is detected, the execution right of the application task lost due to the failure is placed. 2. The cooperative distributed processing method according to claim 1, wherein the task execution management table is updated in the distributed node having the second runner priority and the application task is executed. 6. A node state monitoring table, which is an array having the node number of each distributed node and the operation state symbol of each distributed node as an entry, is placed on a shared distributed memory,
If the operating state symbols of the other distributed nodes in the node state monitoring table match the conditions for causing the state transition of the own node according to the predefined state transition sequence of the own node, the state of the own node is transited. The cooperative distributed processing method according to claim 1, characterized in that. 7. When the failed distributed node is restored,
Or, when a distributed node newly enters the system, the contents of the distributed shared memory of each distributed node are transferred to the distributed node by transferring the contents of the distributed shared memory from a healthy distributed node that has already entered the system. The cooperative distributed processing method according to claim 1, wherein the cooperative distributed processing method is realized. 8. The distributed shared memory of a plurality of distributed nodes is connected by a distributed shared memory network, and the content change of each distributed shared memory is communicated to the distributed shared memory of another distributed node independently of the operation of the distributed node. In a distributed shared memory monitoring device of a distributed processing system in which a memory content is reflected, a memory reading unit that reads out a value of a specified address or area of the distributed shared memory at each distributed node at a predetermined cycle, and the memory. A distributed shared memory monitoring device, comprising: a display control unit that displays the value read by the reading unit on the display device at the predetermined cycle. 9. A difference display control unit for calculating a difference between a value read in the previous cycle and a value read in the current cycle by the memory reading unit and passing the difference to the display control unit. The distributed shared memory monitoring device according to claim 8. 10. The distributed shared memory of a plurality of distributed nodes is connected by a distributed shared memory network, and the content change of each distributed shared memory is communicated to the distributed shared memory of another distributed node independently of the operation of the distributed node. In a distributed shared memory network tracing device of a distributed processing system in which memory contents are reflected, a packet reception detection unit that detects a packet on the distributed shared memory network at each distributed node for each reception, and the packet reception detection unit A packet acquisition unit that acquires the packet contents received and detected by
A distributed shared memory network tracking device, comprising: a display control unit that displays the information acquired by the packet acquisition unit on a display device in the order of reception. 11. A packet-related information reading unit is provided, which acquires information related to memory update, which is not included in a packet detected at each reception by the packet reception detection unit, and transfers the information to the display control unit. The distributed shared memory network tracking device according to claim 10. 12. The distributed shared memory network tracking according to claim 10, further comprising a parameter setting unit that acquires a packet condition designated by a user and passes it to the packet acquisition unit and the display control unit. apparatus. 13. A distributed shared memory of a plurality of distributed nodes is connected by a distributed shared memory network, and the content change of each distributed shared memory is communicated to the distributed shared memory of another distributed node independently of the operation of the distributed node. In a distributed shared memory network setting support device of a distributed processing system in which the memory content is reflected, a distributed shared memory network setting register in which setting information of the control function of the distributed shared memory network is set, and a user instructed, If the control function setting of the distributed shared memory network setting register and the parameter for setting reference are acquired, if the acquired parameter is the control function setting, the setting information is created and set in the distributed shared memory network setting register. For example, set from the distributed shared memory network setting register A distributed shared memory network setting register reference setting unit that reads information, and a display control unit that displays on the display device the setting information read from the distributed shared memory network setting register by the distributed shared memory network setting register reference setting unit. A distributed shared memory network setting support device characterized by having. 14. A timer is provided for activating the display control unit at a constant time cycle.
3. The distributed shared memory network setting support device described in 3. 15. A difference for calculating a difference between a value read in the previous cycle and a value read in the current cycle by the distributed shared memory network setting register reference setting unit, and controlling the display of the display device accordingly. 15. The distributed shared memory network setting support device according to claim 14, further comprising a display control unit.