JP2009271857A - Computer system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To quickly detect that a client computer fails by reducing the load of a server computer or a network load near the server computer. <P>SOLUTION: In this computer system, a first callback processing part 161a transmits an alive notification packet including an address of the client computer 10a at a time interval T1. A client reception packet processing part 14b stores a record including the address included in the transmitted alive notification packet and a receiving time when the alive notification packet is received in a client connection state table 15b. A second callback processing part 162b acquires a record in which a difference between the current time and the receiving time is larger than a time interval T2 from the client connection state table 15b at the time interval T2. The second callback processing part 162b notifies the server computer 20 that the client computer 10a fails to which an address included in the acquired record is allocated. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a computer system that detects that a client computer connected to a server computer is down.

  In a client server system (computer system) including a plurality of computers (hereinafter referred to as client computers) and a server computer, it may be monitored that the client computer is down due to the occurrence of a failure in the client computer, for example.

  For example, it is assumed that when a client computer logs in to a client server system, information indicating that the client computer has logged in is temporarily stored in a database. In this case, when the client computer is terminated (down), the next login cannot be performed unless the information is immediately deleted from the database. In such a case, it is necessary to detect that the client computer is down by monitoring the client computer.

  In addition, even when the server computer has resources corresponding to the client computer, if the client computer goes down, the server computer must continue the service if the resources used by the client computer are not released. I can't. Therefore, when a client computer goes down, it is necessary to detect that the client computer is down and immediately release resources.

  So, for example, when each computer has a failure in its own computer, the failure information is broadcast to other computers at the same time, and the other computer that has received the failure information further passes the information to its own computer via a communication path. A technique (hereinafter referred to as Prior Art 1) that broadcasts simultaneously to other connected computers is disclosed (for example, see Patent Document 1). According to this prior art 1, since each computer is notified of the failure information by a detour notification from another computer in addition to the direct notification of the failure information from the computer in which the failure has occurred, a certain communication path becomes unable to communicate. Even in the case of failure, failure information can be received via another communication path.

  Here, in the above-described prior art 1, it is assumed that communication is performed using a protocol that does not transmit any packet (information) in an idle state such as TCP / IP (Transmission Control Protocol / Internet Protocol). . In such a case, for example, when the power of the client computer suddenly drops or the physical line in the client computer is cut off, failure information cannot be transmitted in the prior art, and the normal close Since this sequence does not operate, other client computers cannot detect that a failure has occurred in the client computer (that is, that the client computer has gone down).

Therefore, for example, each of the plurality of client computers periodically transmits a packet indicating that the client computer is alive (hereinafter referred to as a survival notification packet) to the server computer. As a result, a technique (hereinafter referred to as Prior Art 2) is known in which when a survival notification packet is not received from a client computer for a certain period of time, the server computer detects that the client computer is down.
JP-A-11-175589

  By the way, in the prior art 2 described above, in order to immediately detect that the client computer has gone down, it is necessary to shorten the transmission interval of the survival notification packet transmitted from the client computer to the server computer.

  However, if the transmission interval of the survival notification packet is shortened in this way, for example, when the number of client computers constituting the client server system is large, the load on the server computer (CPU load) or the load on the network in the vicinity of the server computer increases. .

  Therefore, an object of the present invention is to provide a computer system that can quickly detect that a client computer has gone down by reducing the load on the server computer or the network load in the vicinity of the server computer.

  According to one aspect of the present invention, a first computer, a second computer connected to the first computer, a third computer connected to the first computer and the second computer. And the first computer includes transmission means for transmitting a survival notification packet including an address assigned to the first computer at a predetermined first time interval, and the second computer Is a record including receiving means for receiving the transmitted survival notification packet, an address assigned to the first computer included in the received survival notification packet, and a reception time when the survival notification packet is received And a record including the reception time at which the difference between the current time and the reception time is equal to or greater than a predetermined second time interval is taken from the table at the second time interval. And a notifying means for notifying the third computer that the first computer to which the address included in the acquired record has been assigned is down is provided. Is done.

  According to the present invention, the load on the server computer or the network load in the vicinity of the server computer is reduced, and it is possible to quickly detect that the client computer is down.

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

  FIG. 1 is a block diagram showing a hardware configuration of a client server system (computer system) according to the present embodiment.

  The client server system shown in FIG. 1 mainly includes a plurality of client computers 10 (first and second computers), and a server computer 20 (third computer) accessed from the plurality of client computers 10. . The plurality of client computers 10 include, for example, client computers 10a and 10b. The plurality of client computers 10 are connected to the server computer 20 via a network 30 such as a local area network (LAN). Client computers 10 a and 10 b included in the plurality of client computers 10 are connected to each other via a network 30.

  In addition, a plurality of client computers 10 (client computers 10a and 10b included therein) form a group with the client computers, for example. The plurality of client computers 10 constituting this group have a function of notifying the server computer 20 that the other client computers are down by transmitting packets for confirming the existence of each other.

  FIG. 2 is a block diagram mainly showing a functional configuration of the client computers 10a and 10b included in the plurality of client computers 10 shown in FIG.

  The client computer 10a includes a client timer unit 11a, a client connection unit 12a, a client internal storage unit 13a, a client reception packet processing unit 14a, a client connection state table 15a, and a client callback processing unit 16a.

  The client timer unit 11a has a function of acquiring the current time. Further, the client timer unit 11a has a function of notifying that when a registered time has elapsed, for example. The client timer unit 11a is realized by, for example, a real-time clock function or an alarm function built in the client computer 10a.

  The client connection unit 12a transmits a multicast address request packet for requesting a multicast address determined in advance in the client server system to the server computer 20. This multicast address is held in the server computer 20. For example, when a packet is transmitted to this multicast address, it is received by, for example, a plurality of client computers 10 constituting the client server system.

  The client connection unit 12a receives a multicast address response packet including the multicast address from the server computer 20 as a response to the multicast address request packet. The client connection unit 12a acquires a multicast address from the received multicast address response packet. The client connection unit 12a stores (stores) the acquired multicast address in the client internal storage unit 13a.

  The client internal storage unit 13a stores, for example, a time when the client computer 10a starts connection with the server computer 20 (hereinafter referred to as a start time), a survival confirmation transmission skip flag, which will be described later, and the like.

  The client connection unit 12a stores, in the client internal storage unit 13a, the group participation request multipacket in which the client address assigned to the client computer 10a and the start time stored in the client internal storage unit 13a are stored. Sent to a multicast address. As a result, another client computer (for example, client computer 10b) constituting the client server system is requested to participate in the group.

  For example, when the client computer 10 a connects to the server computer 20, the client connection unit 12 a transmits a connection request packet to the server computer 20. The connection request packet stores a client address assigned to the client computer 10a and a start time saved in the client internal storage unit 13a. Further, in this connection request packet, for example, a group flag (“TRUE” or “FALSE”) indicating whether or not the client computer 10a forms a group (that is, grouped) is stored. Whether or not the client computers 10a are grouped is determined with reference to a client connection state table 15a described later.

  The client connection unit 12a receives a connection response packet from the server computer 20 as a response to the connection request packet. In this connection response packet, a client transmission interval (first time interval), a client timeout time (second time interval), and a server transmission interval (third time interval) are stored (set). The client transmission interval indicates, for example, an interval at which the client computer 10a notifies its existence to another client computer (for example, the client computer 10b) constituting the group. The client timeout time indicates, for example, a time until the client computer 10a considers that another client computer constituting the group is down. The server transmission interval indicates, for example, an interval at which the client computer 10a notifies its own existence to the server computer 20.

  The client connection unit 12a registers a callback process in the client timer unit 11a according to the received connection response packet. At this time, the client connection 12a stores the client transmission interval, the client timeout time, and the server transmission interval included in the received connection response packet in the client internal storage unit 13a. Details of the callback process registration will be described later.

  The client reception packet processing unit 14a receives, for example, a group participation response multicast packet from the client computer 10b as a response to the group participation request multicast packet transmitted by the client connection unit 12a. In this group participation response multicast packet, the client address assigned to the client computer 10b that is the transmission destination of the group participation request multicast packet and the time (start time) at which the client computer 10b starts connection with the server computer 20, for example. included.

  The client reception packet processing unit 14a stores (registers) a record (client connection state table record) including the client address and start time included in the received group participation response multicast packet in the client connection state table 15a.

  The client connection state table 15a holds a record including the client address and start time of another client computer (for example, the client computer 10b) that forms a group with the client computer 10a. The record held in the client connection state table 15a includes a client address (field), a start time (field), and a final confirmation time (field).

  In addition, when a group participation request multicast packet is transmitted from another client computer (for example, the client computer 10b), the client reception packet processing unit 14a receives the group participation request multicast packet. When the group participation request multicast packet is received, the client reception packet processing unit 14a stores the client address and start time included in the group participation request multicast packet as a new record in the client connection state table 15a. The client reception packet processing unit 14a transmits a client participation response multicast packet including the client address and start time of the client computer 10a to the multicast address as a response to the received group participation request multicast packet.

  The client reception packet processing unit 14a updates (that is, stores a record) the client connection state table 15a according to the received client participation request multipacket, and then stores (holds) the client connection state table 15a in the client connection state table 15a. When the number of records is 1, all the callback processes registered in the client timer unit 11a are canceled. In this case, the client reception packet processing unit 14a transmits to the server computer 20 a grouped packet in which a group flag indicating that the client computer 10a is grouped is stored.

  The client reception packet processing unit 14a receives a grouping response packet from the server computer 20 as a response to the grouping packet. This grouping response packet stores (sets) a client transmission interval, a client timeout time, and a server transmission interval in the same manner as the connection response packet described above. The client reception packet processing unit 14a registers a callback process in the client timer unit 11a according to the received grouping response packet. At this time, the client transmission interval, client timeout time, and server transmission interval included in the grouping response packet are stored in the client internal storage unit 13a.

  The client reception packet processing unit 14a receives a survival notification packet (hereinafter referred to as a client survival notification multicast packet) for notifying the existence of the client computer 10b from another client computer (for example, the client computer 10b). In this case, the client existence notification multicast packet is received. This client existence notification multicast packet includes the client address and start time of the client computer 10b that is the transmission source of the client existence notification multicast packet.

  When a client survival notification multicast packet is received, the client reception packet processing unit 14a acquires a reception time (current time) when the client survival notification multicast packet is received from the client timer unit 11a. The client reception packet processing unit 14a extracts a record including (matching) the client address and start time included in the received survival notification multicast packet from the client connection state table 15a.

  The client reception packet processing unit 14a stores (replaces) the current time (reception time) acquired from the client timer unit 11a in the final confirmation time field included in the extracted record.

  In other words, the last confirmation time field included in the record held in the client connection state table 15a receives the client survival notification multicast packet last transmitted from another client computer (for example, the client computer 10b) constituting the group. The received reception time is stored.

  The client reception packet processing unit 14a receives, for example, a client multicast packet that has been notified of survival, which will be described later, from the client computer 10b. The existence-notified client multicast packet stores a client address and a start time. The client reception packet processing unit 14a sets the survival confirmation transmission skip flag “TRUE” in the client internal storage unit 13a when the client address and the start time of the client computer 10a are stored in the received multicast notification client multicast packet. (save. The existence confirmation transmission skip flag includes “TRUE” and “FALSE”.

  The client callback processing unit 16a executes a callback process registered in the client timer unit 11a. In this callback process, for example, a packet for notifying the above-mentioned existence is transmitted to, for example, the client computer 10b and the server computer 20 constituting the group. In the callback process, for example, the server computer 20 is notified that the client computer 10b constituting the group has gone down.

  The client callback processing unit 16a includes a first callback processing unit 161a, a second callback processing unit 162a, and a third callback processing unit 163a.

  In response to the notification from the client timer unit 11a, the first callback processing unit 161a is a process for notifying the survival of the client computer 10a to another client computer (for example, the client computer 10b) constituting the group (a survival notification packet). Execute client send callback processing). The client timer unit 11a performs notification at a time interval registered in the survival notification packet client transmission callback process.

  When the first callback processing unit 161a receives the notification from the client timer unit 11a, the first callback processing unit 161a transmits the client existence notification multicast packet including the client address of the client computer 10a and the start time stored in the client internal storage unit 13a to the client It transmits to the multicast address stored in the internal storage unit 13a.

  The second callback processing unit 162a confirms the survival of another client computer (for example, the client computer 10b) constituting the group in response to the notification from the client timer unit 11a (client survival confirmation callback processing). Execute. The client timer unit 11a notifies at a registered time interval in the client survival confirmation callback process.

  Upon receiving the notification from the client timer unit 11a, the second callback processing unit 162a acquires the current time from the client timer unit 11a. The second callback processing unit 162a calculates the difference between the acquired current time and the value (time) of the last confirmation time field held in the client connection state table 15a. The second callback processing unit 162a extracts from the client connection state table 15a all records in which the calculated difference is greater than the client timeout time stored in the client internal storage unit 13a.

  The second callback processing unit 162a transmits to the server computer 20 a down notification packet in which the client address and the start time included in all the extracted records are stored. By transmitting this down notification packet, for example, the server computer 20 is notified that the client computer 10b is down. When the down notification packet is transmitted, the second callback processing unit 162a deletes all the extracted records from the client connection state table 15a.

  When the number of records stored in the client connection state table 15a after all the extracted records are deleted is 0, the second callback processing unit 162a groups the client computers 10a. A grouping packet in which a group flag indicating that there is not (that is, “FALSE”) is stored is transmitted to the server computer 20.

  The third callback processing unit 163a notifies the server computer 20 of the existence of a plurality of grouped client computers 10 (for example, client computers 10a and 10b) in response to the notification from the client timer unit 11a. Execute the process (survival notification packet server send callback process). The client timer unit 11a performs notification at time intervals registered in the survival notification packet server transmission callback process.

  Upon receiving the notification from the client timer unit 11a, the third callback processing unit 163a confirms the survival confirmation transmission skip flag stored in the client internal storage unit 13a. If the survival confirmation transmission skip flag is “TRUE”, the third callback processing unit 163a sets the survival confirmation transmission skip flag “FALSE” in the client internal storage unit 13a and ends the process.

  That is, when the survival confirmation transmission skip flag “TRUE” is set (saved) in the client internal storage unit 13a, the survival notification packet server transmission callback process is skipped once.

  On the other hand, when the survival confirmation skip flag is “FALSE”, the third callback processing unit 162a acquires the current time from the client timer unit 11a. The third callback processing unit 163a calculates the difference between the acquired current time and the value (time) of the final confirmation time field included in the record held in the client connection state table 15a.

  In this final confirmation time field, the last time the client existence notification multicast packet was received from the client computer (for example, the client computer 10b) to which the client address included in the record including the final confirmation time field is assigned (reception) Time) is saved.

  The third callback processing unit 163a extracts from the client connection state table 15a all the records whose calculated difference is not equal to or greater than the server transmission interval stored in the client internal storage unit 13a, that is, smaller than the server interval.

  The third callback processing unit 163a transmits to the server computer 20 a survival notification packet in which the client address and the start time included in all the extracted records are stored. In this survival notification packet, in addition to the client address and start time included in all the extracted records, the client address of the client computer 10a and the start time stored in the client internal storage unit 13a are stored.

  When the third callback processing unit 163a transmits the existence notification packet to the server computer 20, the existence notification notified client in which the client addresses and start times included in all the records extracted from the client connection state table 15a are stored. The multicast packet is transmitted (multicast) to the multicast address stored in the client internal storage unit 13a. As a result, the client computer 10b is notified that the server computer 20 has been notified of the existence of the client computer (for example, the client computer 10b) to which the client address included in the client multicast packet that has been notified of the existence is assigned. .

  The client computer 10b includes a client timer unit 11b, a client connection unit 12b, a client internal storage unit 13b, a client reception packet processing unit 14b, a client connection state table 15b, and a client callback processing unit 16b.

  Since the units 11b to 16b included in the client computer 10b are the same as the units 11a to 16a included in the client computer 10a, detailed description thereof is omitted.

  The client callback processing unit 16b of the client computer 10b includes a first client callback processing unit 161b, a second client callback processing unit 162b, and a third client callback processing unit 163b. Since these units 161b to 163b are the same as the units 161a to 163a included in the client callback processing unit 16a of the client computer 10a, detailed description thereof will be omitted.

  Although the configuration of the client computers 10a and 10b has been described with reference to FIG. 2, the same applies to client computers other than the client computers 10a and 10b.

  FIG. 3 is a block diagram mainly showing a functional configuration of the server computer 20. As shown in FIG. 3, the server computer 20 includes a server timer unit 21, a server connection standby unit 22, a server connection state table 23, a server reception packet processing unit 24, and a server callback processing unit 25.

  The server computer 20 includes T1 as a time interval (client transmission interval) for notifying the existence of each other among the plurality of client computers 10 constituting the group, and the plurality of client computers constituting the group are other clients. T2 as a time interval (client timeout time) until the computer is considered to be down, T3 as a time interval (server transmission interval) for each of the plurality of client computers 10 to notify the server computer 20 of its existence, and the server computer 20 It is assumed that T4 is determined in advance as a time interval (fourth time interval) until each of the plurality of client computers 10 is considered to be down.

  Here, it is assumed that the time interval T2 is longer than the time interval T1, and the time interval T4 is longer than the time interval T3. The time interval T3 is set longer than the time interval T1, and the time interval T4 is set longer than the time interval T2.

  The server timer unit 21 has a function of acquiring the current time. The server timer unit 21 has a function of notifying that when, for example, a registered time has elapsed. The server timer unit 21 is realized by, for example, a real-time clock function or an alarm function built in the server computer 20.

  The server connection standby unit 22 receives, for example, a multicast address request packet transmitted by the client computer 10a. When a multicast address request packet is received, the server connection standby unit 22 transmits a multicast address response packet including a multicast address determined in advance by the client server system to the client computer 10a.

  The server connection standby unit 22 receives, for example, a connection request packet transmitted by the client computer 10a.

  If the group flag included in the received connection request packet is “TRUE (that is, the client computers 10a are grouped)”, the server connection standby unit 22 sets the client transmission interval as a response to the connection request packet. A connection response packet in which T1, T2 is set as the client timeout time, and T3 is set as the server transmission interval is transmitted to the client computer 10a. At this time, the server connection standby unit 22 registers the callback processing with the server timer unit 21 at the time interval T4 described above.

  If the group flag included in the received connection request packet is “FALSE” (that is, the client computer 10a is not grouped), the server connection standby unit 22 sends a client transmission interval as a response to the connection request packet. The connection response packet in which 0 is set to 0, 0 is set to the client timeout period and T1 is set to the server transmission interval is transmitted to the client computer 10a. At this time, the server connection standby unit 22 registers the callback process with the server timer unit 21 at the time interval T2.

  The server connection state table 23 includes a record (server connection state table record) including client addresses, start times, and final confirmation times of a plurality of client computers 10 (for example, client computers 10a and 10b) connected to the server computer 20. Is retained.

  The server reception packet processing unit 24 receives, for example, a grouped packet transmitted by the client computer 10a.

  Here, it is assumed that a group flag “TRUE” indicating that the client computers 10a are grouped is stored in the grouped packet received by the server reception packet processing unit 24, for example. In this case, as a response to the received grouped packet, the server received packet processing unit 24 sends a grouped response packet in which T1 is set in the client transmission interval, T2 is set in the client timeout period, and T3 is set in the server transmission interval to the client computer 10a Send to.

  On the other hand, it is assumed that a group flag “FALSE” indicating that, for example, the client computer 10a is not grouped is stored in the grouped packet received by the server reception packet processing unit 24. In this case, the server reception packet processing unit 24 sends, as a response to the received grouping packet, a grouping response packet in which 0 is set in the client transmission interval, 0 is set in the client timeout period, and T1 is set in the server transmission interval. Send to.

  Various callback processes are registered in the client timer unit 11a according to the client transmission interval, client timeout time, and server transmission interval set in the grouping response packet transmitted by the server reception packet processing unit 24.

  As described above, the client transmission interval, the client timeout time, and the server transmission interval stored in the grouping response packet transmitted by the server reception packet processing unit 24 differ depending on the group flag stored in the grouping packet. That is, for example, the time interval registered in the callback process in the client timer unit 11a differs depending on whether or not the client computers 10a are grouped.

  In other words, if the client computer 10a is grouped, the callback processing when the client computer 10a is grouped is registered in the client timer unit 11a. On the other hand, if the client computer 10a is not grouped, a callback process when the client computer 10a is not grouped is registered in the client timer unit 11a.

  The server reception packet processing unit 24 receives, for example, a down notification packet transmitted by the client computer 10a. The server reception packet processing unit 24 deletes the record including the client address and the start time included in the received down notification packet from the server connection state table 23.

  Further, the server reception packet processing unit 24 receives a survival notification packet transmitted by the client computer 10a, for example. The server reception packet processing unit 24 replaces the value of the last confirmation time field included in all records including the client address and the start time stored in the received survival notification packet with the current time. The current time is acquired from the server timer unit 21.

  In response to the notification from the server timer unit 21, the server callback processing unit 25 confirms the existence of a plurality of client computers 10 (for example, client computers 10a and 10b) connected to the server computer 20 (server existence). Confirmation callback processing) is executed. In this case, the server timer unit 21 notifies the server callback processing unit 25 at the time interval (T2 or T4) registered in the server timer unit 21. Here, the time interval registered in the server timer unit 21 will be described as the time interval T4.

  Upon receiving a notification from the server timer unit 21, the server callback processing unit 25 acquires the current time from the server timer unit 21. The server callback processing unit 25 calculates the difference between the acquired current time and the value (time) of the last confirmation time field held in the server connection state table 23. The server callback processing unit 25 searches the server connection state table 23 for all records having the calculated difference larger than the time interval T4. Thereby, the server callback processing unit 25 detects that the client computer to which the client address included in all the records whose calculated difference is larger than the time interval T4 is assigned is down. The server callback processing unit 25 deletes all the retrieved records from the server connection state table 23.

  FIG. 4 shows an example of the data structure of the client internal storage unit 13a (and client internal storage unit 13b) shown in FIG. As shown in FIG. 4, the client internal storage unit 13a stores a start time, a survival confirmation transmission skip flag, a multicast address, a client transmission interval, a client timeout time, and a server transmission interval.

  The start time indicates the time when the client computer 10a starts connection with the server computer 20. The survival confirmation transmission skip flag indicates whether or not to skip the survival notification packet server transmission callback process (“TRUE” or “FALSE”) as described above. The multicast address is an address for which a plurality of client computers 10 are designated. The client transmission interval, client timeout time, and server transmission interval are time intervals registered in the various callback processes described above.

  FIG. 5 shows an example of the data structure of records held in the client connection state table 15a shown in FIG. As shown in FIG. 5, the record includes a client address field, a start time field, and a final confirmation time field.

  The client address field stores a client address assigned to a client computer (for example, the client computer 10b) grouped with the client computer 10a. In the start time field, the time when the client computer to which the client address stored in the client address field is assigned starts connection with the server computer 20 is stored. In the final confirmation time field, the client address field stores the time at which the survival of the client computer to which the client address is assigned was last confirmed (for example, the reception time when the survival notification packet was received). .

  The data structure of the record held in the server connection state table 23 shown in FIG. 4 is the same as the data structure shown in FIG.

  Next, with reference to the sequence chart of FIG. 6, for example, a processing procedure until the client computer 10a joins the group including the client computer 10b and connects to the server computer 20 will be described. It is assumed that the client computer 10b is already connected to the server computer 20.

  Here, as described above, for example, T1 is the time interval (client transmission interval) for notifying the existence of each other among the plurality of client computers 10 constituting the group, and each of the plurality of client computers 10 constituting the group is the other T2 as a time interval (client timeout time) until the client computer is considered to be down, T3 as a time interval (server transmission interval) for each of the plurality of client computers 10 to notify the server computer 20 of its existence, and the server computer 20 Assume that T4 is predetermined as a time interval until each of the plurality of client computers 10 is considered to be down.

  As described above, the time intervals T1 to T4 are registered in the various callback processes described above. However, the time intervals are registered depending on whether or not the client computers connected to the server computer 20 are grouped. Is different.

  First, when connecting to the server computer 20, the client computer 10a executes an initialization process for the client internal storage unit 13a. In this case, the client computer 10a stores the current time as the start time stored in the client internal storage unit 13a. This current time is acquired from the client timer unit 11a. In addition, the client computer 10a sets (saves) a survival confirmation transmission skip flag “FALSE” in the client internal storage unit 13a.

  The client connection unit 12a of the client computer 10a transmits a multicast address request packet for requesting a multicast address to the server connection standby unit 22 of the server computer 20 (step S1). This multicast address is determined in advance in the client server system.

  The server connection standby unit 22 of the server computer 22 receives the multicast address request packet transmitted by the client connection unit 12a. When the multicast address request packet is received, the server connection standby unit 22 transmits a multicast address response packet to the client connection unit 12a as a response to the multicast address request packet.

  Here, FIG. 7 shows an example of the data structure of the multicast address response packet. As shown in FIG. 7, a multicast address is stored in the multicast address response packet.

  Returning again to FIG. 6, the client connection unit 12 a of the client computer 10 a receives the multicast address response packet transmitted by the server connection standby unit 22. The client connection unit 12a acquires the multicast address stored in the received multicast address response packet. The client connection unit 12a stores the acquired multicast address in the client internal storage unit 13a (step S3).

  Next, the client connection unit 12a transmits a group participation request multicast packet for requesting participation in the group to the multicast address stored in the client internal storage unit 13a (step S4). Thereby, the group participation request multicast packet is transmitted to a plurality of client computers 10 including the client computer 10b. After transmitting the group participation request multicast packet, the client connection unit 12a waits for a predetermined time (Tc).

  Here, FIG. 8 shows an example of the data structure of the group participation request multicast packet. As shown in FIG. 8, the client address and the start time are stored in the group participation request multicast packet. This client address is an address assigned to the client computer 10a that is the transmission source of the group participation request multicast packet. The start time is a start time stored in the client internal storage unit 13a.

  Returning to FIG. 6 again, the client reception packet processing unit 14b of the client computer 10b that has already completed connection processing with the server computer 20 receives the group participation request multicast packet transmitted by the client connection unit 12a. The client reception packet processing unit 14b updates the client connection state table 15b according to the received group participation request multicast packet (step S5). In this case, the client reception packet processing unit 14b stores the client address and start time of the client computer 10a stored in the received group join request multicast packet as a new record in the client connection state table 15b.

  Next, the client reception packet processing unit 14b transmits a group participation response multicast packet to the multicast address as a response to the received group participation request multicast packet (step S6).

  Here, since the client computer 10b is already connected to the server computer 20, processing corresponding to the above-described steps S1 to S3 is executed. Therefore, in the process of step S6, a multicast packet already stored in the client internal storage unit 13b is used. Thereby, the group participation response multicast packet is transmitted to a plurality of client computers 10 including the client computer 10a.

  The group participation response multicast packet stores the client address and the start time in the same manner as the group participation request multicast packet shown in FIG. This client address is the client address of the client computer 10b. The start time is a start time stored in the client internal storage unit 13b.

  The client reception packet processing unit 14a of the client computer 10a receives the group participation response multicast packet transmitted by the client reception packet processing unit 14b. The client reception packet processing unit 14a updates the client connection state table 15a according to the received group participation response multicast packet (step S7). In this case, the client reception packet processing unit 14a stores the client address and start time of the client computer 10b stored in the received group join response multicast packet as a new record in the client connection state table 15a.

  Note that the processing in steps S5 to S7 described above is executed, for example, during a predetermined time Tc in which the client connection unit 12a is on standby.

  On the other hand, the client reception packet processing unit 14b of the client computer 10b confirms the number of records held in the client connection state table 15b. At this time, if the number of records held in the client connection state table 15b is 1, all callback processes registered in the client timer unit 11b are canceled.

  The number of records held in the client connection state table 15b represents the number of client computers that form a group with the client computer 10b. That is, when the number of records held in the client connection state table 15b is 1 or more, it indicates that the client computer 10b is grouped.

  At this time, if the number of records held in the client connection status table 15b is 1, the records held in the client connection status table 15b are stored in the client address of the client computer 10a stored in step S5 described above. This record contains the start time.

  Therefore, since the client computer 10b is not grouped until the client computer 10a is connected (that is, the number of records in the client connection state table 15b is 0), the client timer unit 11b includes the client computer 10b described above. Callback processing when 10b is not grouped is registered.

  Here, since the number of records in the client connection state table 15b becomes 1 when the client computer 10a is connected, the client computers 10b are grouped. Therefore, in order to register the callback processing when the client computer 10b is grouped in the client timer unit 11b, the callback registered in the client timer unit 11b (when not grouped) as described above. Cancel all processing.

  If the number of records held in the client connection state table 15b is not 1 (that is, 2 or more), the callback processing when already grouped is registered in the client timer unit 11b. Therefore, the processing of step S8 to step S10 shown below is not executed.

  When all the callback processes are canceled, the client reception packet processing unit 14b transmits the grouped packet to the server reception packet processing unit 24 of the server computer 20 (step S8).

  Here, FIG. 9 shows an example of the data structure of the grouped packet. As shown in FIG. 9, a group flag is stored in the grouped packet. This group flag includes “TRUE” indicating grouping and “FALSE” indicating not grouping.

  In step S8 described above, a grouping packet storing “TRUE” indicating that the client computers 10b are grouped is transmitted.

  Returning to FIG. 6 again, the server reception packet processing unit 24 of the server computer 20 receives the grouped packet transmitted by the client reception packet processing unit 14b. The server reception packet processing unit 24 checks the group flag stored in the received grouped packet. Here, the group flag stored in the received grouped packet is “TRUE”. In this case, the server reception packet processing unit 24 transmits a grouping response packet to the client reception packet processing unit 14b of the client computer 10b as a response to the received grouping packet (step S9).

  Here, FIG. 10 shows an example of the data structure of the grouping response packet. As shown in FIG. 10, the grouping response packet stores a server transmission interval, a client transmission interval, and a client timeout time.

  In step S9 described above, a grouping response packet storing (setting) T1 as the client transmission interval, T2 as the client timeout time, and T3 as the server transmission interval is transmitted. That is, a grouping response packet in which a value (time interval) for registering callback processing when grouped is stored is transmitted.

  Returning again to FIG. 6, the client reception packet processing unit 14 b of the client computer 10 b receives the grouping response packet transmitted by the server reception packet processing unit 24. The client reception packet processing unit 14b stores the client transmission interval (T1), client timeout time (T2), and server transmission interval (T3) stored in the received grouping response packet in the client internal storage unit 13b.

  The client reception packet processing unit 14b performs various callback processes on the basis of the client transmission interval (T1), client timeout time (T2), and server transmission interval (T3) stored in the received grouping response packet. Register in the timer unit 11b (step S10).

  Specifically, the client reception packet processing unit 14b registers the above-described survival notification packet client transmission callback processing in the client timer unit 11b with the processing time interval as the client transmission interval (T1). Thereby, the survival notification packet client transmission callback process is executed at the time interval T1.

  The client reception packet processing unit 14b registers the above-described client survival confirmation callback processing in the client timer unit 11b with the processing time interval as the client timeout period (T2). Thereby, the client existence confirmation callback process is executed at the time interval T2.

  Further, the client reception packet processing unit 14b registers the above-described survival notification packet server transmission callback processing in the client timer unit 11b with the processing time interval as the server transmission interval (T3). Thereby, the survival notification packet server transmission callback process is executed at the time interval T3.

  On the other hand, the client connection unit 12a of the client computer 10a waits for the predetermined time (Tc), and then transmits a connection request packet to the server connection standby unit 22 of the server computer 20 with reference to the client connection state table 15a (step S11).

  Here, FIG. 11 shows an example of the data structure of the connection request packet. As shown in FIG. 11, the connection request packet stores a client address, a start time, and a group flag.

  Here, in the client connection state table 15a, a record including the client address and the start time of the client computer 10b is stored in step S7 described above. Therefore, since the number of records held in the client connection state table 15a is 1 or more, the client computers 10a are grouped. As a result, in the above-described step S11, the client address of the client computer 10a, the start time stored in the client internal storage unit 13a, and “TRUE” indicating that the client computers 10a are grouped are stored. A request packet is sent.

  Returning to FIG. 6 again, the server connection standby unit 22 of the server computer 20 receives the connection request packet transmitted by the client connection unit 12a. The server connection standby unit 22 confirms the group flag stored in the received connection request packet.

  The server connection standby unit 22 transmits a connection response packet to the client connection unit 12a of the client computer 10a as a response to the connection request packet (step S12). In this connection response packet, the client transmission interval, the client timeout time, and the server transmission interval are stored as in the grouping response packet shown in FIG.

  Here, the group flag is “TRUE”. In this case, the connection response packet stores T1 as the client transmission interval, T2 as the client timeout time, and T3 as the server transmission interval. That is, a connection response packet in which a value (time interval) for registering callback processing when grouped is stored is transmitted.

  The server connection standby unit 22 registers the server survival confirmation callback process in the server timer unit 21 with the transmission of the connection response packet as the time interval T4 set in advance (step S13).

  The client connection unit 12a of the client computer 10a receives the connection response packet transmitted by the server connection standby unit 22. The client connection unit 12a stores the client transmission interval, client timeout time, and server transmission interval included in the received connection response packet in the client internal storage unit 13a.

  The client connection unit 12a performs various callback processes based on the client transmission interval (T1), the client timeout period (T2), and the server transmission interval (T3) stored in the received connection response packet. (Step S14).

  Specifically, the client connection unit 12a registers the above-described survival notification packet client transmission callback processing in the client timer unit 11a with the processing time interval as the client transmission interval (T1).

  The client connection unit 12a registers the above-described client survival confirmation callback processing in the client timer unit 11a with the time interval to be processed as the client timeout time (T2).

  Further, the client connection unit 12a registers the above-described survival notification packet server transmission callback processing in the client timer unit 11a with the processing time interval as the server transmission interval (T3).

  If the group flag stored in the connection request packet transmitted in step S11 described above is, for example, “FALSE”, the client computers 10a are not grouped. In this case, the server connection standby unit 22 transmits a connection response packet storing 0 as the client transmission interval, 0 as the client timeout time, and T1 as the server transmission interval to the client computer 10a. That is, a connection response packet storing a value (time interval) for registering callback processing when not grouped is transmitted.

  At this time, the server connection standby unit 22 registers the server survival confirmation callback process in the server timer unit 21 with the time interval to be processed as a predetermined time interval T2.

  The client connection unit 12a of the client computer 10a performs the above-described survival notification packet server transmission callback process with the processing time interval as the server transmission interval (T1) included in the connection response packet transmitted by the server connection standby unit 22. Register in the client timer unit 11a.

  Since the client transmission interval and the client timeout time included in the connection response packet are 0, they are not registered in the client timer unit 11a. That is, when the client computer 10a is not grouped, the above-described survival notification packet client transmission callback processing and client survival confirmation callback processing are not executed in the client computer 10a. In other words, when the client computer 10a is not grouped, the client computer 10a does not notify other client computers of its own existence, but only notifies the server computer 20.

  Next, various callback processes when the client computer 10a is grouped by the process shown in FIG. 6 (steady state) will be described.

  With reference to the sequence chart of FIG. 12, the processing procedure of the survival notification packet client transmission callback processing by the first callback processing unit 161a included in the client callback processing unit 16a of the client computer 10a will be described.

  For example, when the client computer 10a is grouped, the survival notification packet client transmission callback process is registered in the client timer unit 11a with the processing time interval as T1. Thereby, the survival notification packet client transmission callback process is executed at the time interval T1.

  As described above, when the client computers 10a are not grouped, the survival notification packet client transmission callback process is not executed because it is not registered in the client timer unit 11a.

  First, the client timer unit 11a, for example, has passed the time interval T1 from the start time stored in the client internal storage unit 13a (or the survival notification packet client transmission callback process by the previous first callback processing unit 161a). In such a case, a notification is sent to the first callback processing unit 161a (step S21). That is, the client timer unit 11a notifies at every registered time interval T1 (every time interval T1).

  Upon receiving the notification from the client timer unit 11a, the first callback processing unit 161a obtains the client address of the client computer 10a and the start time stored in the client internal storage unit 13a (step S22).

  The first callback processing unit 161a stores the acquired client address and start time in the client existence notification multicast packet. The first callback processing unit 161a transmits the client existence notification multicast packet in which the client address and the start time are stored to the multicast address stored in the client internal storage unit 13a (step S23).

  The client reception packet processing unit 14b of the client computer 10b receives the client existence notification multicast packet transmitted by the first callback processing unit 161a. The client reception packet processing unit 14b extracts a record including the client address and the start time stored in the received client existence notification multicast packet from the client connection state table 15b.

  The client reception packet processing unit 14b replaces the last confirmation time (field value) included in the extracted record with the current time. Thereby, the client reception packet processing unit 14b updates the client connection state table 15b (step S24).

  As described above, for example, by transmitting a client existence notification multicast packet from the client computer 10a to another client computer (for example, the client computer 10b) at every time interval T1 registered in the client timer unit 11a, for example, The other client computers are notified of the existence of the client computer 10a.

  Here, the survival notification packet client transmission callback processing by the first callback processing unit 161a included in the client callback processing unit 16a of the client computer 10a has been described. For example, the client callback processing unit 16b of the client computer 10b includes The same applies to the survival notification packet client transmission callback processing by the included first callback processing unit 161b.

  Next, with reference to a sequence chart of FIG. 13, a processing procedure of client survival confirmation callback processing by the second callback processing unit 162a included in the client callback processing unit 16a of the client computer 10a will be described.

  For example, when the client computer 10a is grouped, this client existence confirmation callback process is registered in the client timer unit 11a with the processing time interval as T2. Thereby, the client existence confirmation callback process is executed at the time interval T2. The time interval T2 is stored in the client internal storage unit 13a as a client timeout time.

  As described above, when the client computers 10a are not grouped, the client existence confirmation callback process is not executed because it is not registered in the client timer unit 11a.

  For example, when the time interval T2 has elapsed since the start time stored in the client internal storage unit 13a (or the client survival confirmation callback processing by the second callback processing unit 162a of the previous time), the client timer unit 11a 2 is notified to the callback processing unit 162a (step S31). That is, the client timer unit 11a notifies at each registered time interval T2.

  Upon receiving the notification from the client timer unit 11a, the second callback processing unit 162a acquires the current time from the client timer unit 11a. The second callback processing unit 162a calculates the difference between the acquired current time and the value (time) of the final confirmation time field included in the record held in the client connection state table 15a. The second callback processing unit 162a acquires, from the client connection state table 15a, all records in which the calculated difference is stored in the client internal storage unit 13a, that is, the client time-out period, that is, a record that is larger than the time interval T2. S32).

  Next, the second callback processing unit 162a transmits a down notification packet to the server reception packet processing unit 24 of the server computer 20 (step S33).

  Here, FIG. 14 shows an example of the data structure of the down notification packet. As shown in FIG. 14, the down notification packet stores, for example, a set of a plurality of client addresses and start times. In the example shown in FIG. 14, N (N is an integer of 2 or more) sets of client addresses and start times from the set of client address 1 and start time 1 to the set of client address N and start time N are stored. Yes.

  Here, the down notification packet stores a set of the client address and the start time included in each of all the records acquired in step S22. In FIG. 14, the down notification packet has been described as storing a plurality of sets of client addresses and start times. However, if there is one record acquired in step S22, for example, The notification packet stores a set of client addresses and start time.

  When the second callback processing unit 162a transmits the down notification packet, the second callback processing unit 162a transmits a record including the client address and the start time (a set) included in the down notification packet (that is, the record acquired in step S22) to the client. Delete from the connection state table 15a (step S34).

  The server reception packet processing unit 24 of the server computer 20 receives the down notification packet transmitted by the second callback processing unit 162a. The server reception packet processing unit 24 deletes the record including the set of the client address and the start time stored in the received down notification packet from the server connection state table 23 (Step S35).

  On the other hand, the second callback processing unit 162a refers to the client connection state table 15a after the processing in step S24 is executed. At this time, if the number of records held in the client connection state table 15a is 0, the second callback processing unit 162a has a group flag “FALSE” indicating that it is not grouped. The packet is transmitted to the server reception packet processing unit 24 of the server computer 20 (step S36). If the number of records held in the client connection state table 15a is not 0, the process is terminated.

  The server reception packet processing unit 24 of the server computer 20 receives the grouped packet transmitted by the second callback processing unit 162a. When the group flag stored in the received grouping packet is “FALSE”, the server reception packet processing unit 24 transmits a grouping response packet to the client reception packet processing unit 14a of the client computer 10a (step S37). ). Here, since the client computers 10a are not grouped, a grouping response packet storing 0 as the client transmission interval, 0 as the client timeout time, and T1 as the server transmission interval is transmitted.

  The client reception packet processing unit 14a of the client computer 10a receives the grouping response packet transmitted by the server reception packet processing unit 24. The client reception packet processing unit 14a stores the client transmission interval (0), client timeout time (0), and server transmission interval (T1) stored in the received grouping response packet in the client internal storage unit 13a.

  The client reception packet processing unit 14a performs various callback processing based on the client transmission interval (0), the client timeout time (0), and the server transmission interval (T1) stored in the received grouping response packet. Is registered in the client timer unit 11a.

  In this case, the survival notification packet server transmission callback process is registered in the client timer unit 11a with the processing time interval as the server transmission interval (T1). Since both the client transmission interval and the client timeout time are 0, they are not registered.

  As described above, when the record is deleted from the client connection state table 15a, for example, when the client computer 10a is not grouped, the server computer 20 is notified of this, and the client computer 10a is notified. A callback process when not grouped is registered in the timer unit 11a.

  Here, the client survival confirmation callback processing by the second callback processing unit 162a included in the client callback processing unit 16a of the client computer 10a has been described. For example, the client callback processing unit 16b of the client computer 10b includes The same applies to the client existence confirmation callback process by the second callback processing unit 162b.

  Next, with reference to the sequence chart of FIG. 15, the processing procedure of the survival notification packet server transmission callback processing by the third callback processing unit 163a included in the client callback processing unit 16a of the client computer 10a will be described.

  For example, when the client computer 10a is grouped, the survival notification packet server transmission callback process is registered in the client timer unit 11a with the processing time interval as T3. In this case, the survival notification packet server transmission callback process is executed at the time interval T3.

  On the other hand, for example, when the client computer 10a is not grouped, the survival notification packet server transmission callback process is registered in the client timer unit 11a with the processing time interval as T1. In this case, the survival notification packet server transmission callback process is executed at the time interval T1.

  Here, the survival notification packet server transmission callback process is described as being registered in the client timer unit 11a with the time interval to be processed as T3.

  In this case, the time interval T3 is stored in the client internal storage unit 13a as a server transmission interval.

  First, the client timer unit 11a, for example, the start time stored in the client internal storage unit 13a (or when the time interval T3 has elapsed from the survival notification packet server transmission callback processing by the previous third callback processing unit 163a) In addition, a notification is sent to the third callback processing unit 163a (step S41), that is, the client timer unit 11a notifies every registered time interval T3.

  When the third callback processing unit 163a receives a notification from the client timer unit 11a, the third callback processing unit 163a executes a survival notification packet server transmission callback process. The third callback processing unit 163a confirms the survival confirmation skip flag stored in the client internal storage unit 13a.

  If the survival confirmation skip flag is “TRUE”, the third callback processing unit 163a sets the survival confirmation skip flag to “FALSE” and ends the survival notification packet server transmission callback process. That is, when the survival confirmation skip flag is “TRUE”, the survival notification packet server transmission callback process executed every time interval T3 is skipped once.

  On the other hand, if the survival confirmation skip flag is “FALSE”, the third callback processing unit 163a acquires the current time from the client timer unit 11a. The third callback processing unit 163a calculates a difference between the acquired current time and the value (time) of the last confirmation time field held in the client connection state table 15a. The third callback processing unit 163a acquires, from the client connection state table 15a, all records whose calculated differences are stored in the client internal storage unit 13a and that are smaller than the server transmission interval, that is, the time interval T3 (Step S1). S42).

  Next, the third callback processing unit 163a transmits a survival notification packet to the server reception packet processing unit 24 of the server computer 20 (step S43).

  Here, like the down notification packet shown in FIG. 14 described above, a plurality of sets of client addresses and start times are stored in the survival notification packet. Here, the existence notification packet includes the client address of the client computer 10a and the set of the start time stored in the client internal storage unit 13a, and the client address and start included in each of all the records acquired in step S42. A set of times is stored.

  The server reception packet processing unit 24 of the server computer 20 receives the survival notification packet transmitted by the third callback processing unit 163a. In the server connection state table 23, the server reception packet processing unit 24 sets the final confirmation time (field value) of all records including the set of the client address and the start time stored in the received survival notification packet as the current time. Replace with Thereby, the server reception packet processing unit 24 updates the server connection state table 23 (step S44). The current time is acquired from the server timer unit 21.

  In addition, the third callback processing unit 163a transmits the existence-sold client multicast packet to the multicast address stored in the client internal storage unit 13a (step S45). This survival notification-completed client multicast packet stores a plurality of sets of client addresses and start times, as in the down notification packet shown in FIG. In the existence-notified client multicast packet, a set of the client address and the start time included in each of all the records acquired in step S42 is stored.

  The client reception packet processing unit 14b of the client computer 10b receives the existence-notified client multicast packet transmitted by the third callback processing unit 163a. The client reception packet processing unit 14b confirms whether or not the combination of the client address of the client computer 10b and the start time stored in the client internal storage unit 13b is stored in the received client multicast packet that has been notified of existence.

  When the set of the client address of the client computer 10b and the start time stored in the client internal storage unit 13b is stored, the client reception packet processing unit 14b skips the existence confirmation transmission stored in the client internal storage unit 13b. The flag is rewritten to “TRUE” (step S46). On the other hand, when the set of the client address of the client computer 10b and the start time stored in the client internal storage unit 13b is not stored, the process is terminated.

  As described above, in the survival notification packet server transmission callback processing by the third callback processing unit 163a, the client computer 10a and another client computer (for example, the client computer 10b) grouped with the client computer 10a. Is notified to the server computer 20.

  Further, the third callback processing unit 163a notifies the client computer 10b that the client computer 10b has been notified of the existence of the client computer 10b, for example, so that the client computer 10b rewrites the existence confirmation transmission skip flag to “TRUE”. Thereby, in the client computer 10b, the process of notifying the server computer 20 of survival (survival notification packet server transmission callback process) is skipped, so the amount of communication with the server computer 20 can be reduced.

  Here, the survival notification packet server transmission callback processing by the third callback processing unit 163a included in the client callback processing unit 16a of the client computer 10a has been described. For example, the client callback processing unit 16b of the client computer 10b includes The same applies to the survival notification packet server transmission callback processing by the included third callback processing unit 163b.

  Next, with reference to a sequence chart of FIG. 16, a processing procedure of server survival confirmation callback processing by the server callback processing unit 25 of the server computer 20 will be described.

  For example, when the client computer 10a connected to the server computer 20 is grouped, the server survival confirmation callback process is registered in the client timer unit 21 with the processing time interval as T4. In this case, the server survival confirmation callback process is executed at the time interval T4.

  On the other hand, for example, when the client computer 10a connected to the server computer 20 is not grouped, the server existence confirmation callback processing is registered in the server timer unit 21 with the processing time interval being T2. In this case, the server existence confirmation callback process is executed at the time interval T2.

  The server timer unit 21 notifies the server callback processing unit 25 at every registered time interval T4 (step S51).

  Upon receiving the notification from the server timer unit 21, the server callback processing unit 25 executes server survival confirmation callback processing. The server callback processing unit 25 acquires the current time from the server timer unit 21. The server timer unit 21 calculates the difference between the acquired current time and the value of the last confirmation time field held in the server connection state table 23. The server callback processing unit 25 identifies all records in the server connection state table 23 where the calculated difference is greater than the time interval T4 (step S52).

  In this case, the server callback processing unit 25 considers that the client computer to which the client address included in the specified record is assigned is down, and deletes the record from the server connection state table 23 (step S53).

  Next, with reference to FIG. 17 and FIG. 18, the flow of packets in the client server system according to the present embodiment will be described.

  As shown in FIG. 17, the client server system includes client computers 10 a, 10 b, 10 c and a server computer 20. The client computers 10a, 10b, 10c and the server computer 20 are connected to each other via a network. The client computers 10a, 10b, and 10c form a group (grouped).

  In this case, the client computer 10a transmits a client existence notification multicast packet to the multicast address every time interval T1. That is, the client computer 10a notifies its existence to the client computers 10b and 10c at the time interval T1.

  Further, the client computer 10a transmits a survival notification packet to the server computer 20 at the time interval T3. That is, the client computer 10a notifies its own existence to the server computer 20 at the time interval T3.

  Similarly for the client computers 10b and 10c, the client existence notification multicast packet is transmitted to the other client computers at the time interval T1, and the existence notification packet is transmitted to the server computer 20 at the time interval T3.

  Here, as described above, the time interval T3 is sufficiently longer than the time interval T1. As a result, the number of survival notification packets received by the server computer 20 can be reduced as compared with client survival notification multicast packets transmitted and received between the client computers 10a, 10b, and 10c. Therefore, the server computer 20 and the network load in the vicinity of the server computer 20 can be suppressed.

  Along with this, for example, the amount of communication between the client computers 10a, 10b and 10c increases. Therefore, when configuring a group, it is preferable to group client computers in the vicinity of the network. Thereby, cost can be reduced.

  Further, in the client computers 10a, 10b, and 10c, for example, when the client existence notification multicast packet is not transmitted from another client computer during the time interval T2, it is considered that the other client computer is down.

  As shown in FIG. 19, for example, when the client computer 10b considers the client computer 10a to be down, the client computer 10b notifies the server computer 20 of this by a down notification packet. As a result, even if the time interval (that is, the time interval T3) of the survival notification packet transmitted from each of the client computers 10a, 10b, and 10c to the server computer 20 is long, for example, the client computer 10a is down. At that time, the client computer 10b notifies that fact, so the server computer 20 can quickly detect that the client computer 10a is down.

  As described above, in the present embodiment, a plurality of client computers 10 constituting a group transmit client survival notification multicast packets periodically (at a time interval T1) to notify each other of the survival, for example, client If the client existence notification multicast packet from the computer 10a is not received within a predetermined time, the client computer 10a is considered to be down and the server computer 20 is notified accordingly.

  Therefore, in this embodiment, the load due to the life notification packet for notifying the existence from the plurality of client computers 10 to the server computer 20 is distributed to the plurality of client computers 10, and the existence notification packet to the server computer 20 is transmitted. By reducing the load, the load on the server computer 20 and the network load in the vicinity of the server computer 20 can be reduced, and it is possible to quickly detect that each of the plurality of client computers 10 is down.

  In this embodiment, a survival notification packet is transmitted from the plurality of client computers 10 to the server computer 20 at the time interval T3. Thus, for example, even when all of the plurality of client computers 10 constituting the group are down, the server computer 20 can detect that the plurality of client computers 10 are down. Since this time interval T3 is set longer than the time interval T1, even in this case, the load on the server computer 20 and the network load in the vicinity of the server computer 20 can be reduced.

  In the present embodiment, the larger the difference between the time intervals T1 and T3 described above, the fewer the survival confirmation packets between the plurality of client computers 10 and the server computers 20, so the load on the server computer 20 and the server The effect of reducing the network load in the vicinity of the computer 20 is great. However, from the relationship between the time intervals T3 and T4 (T3 <T4), if the time interval T4 becomes too large by increasing the time interval T3, as described above, all of the plurality of client computers 10 constituting the group are down. It takes longer time to detect down. For this reason, it is preferable to determine an appropriate time interval in consideration of the above circumstances.

  In the present embodiment, the group is composed of two or more client computers (for example, the client computers 10a and 10b). However, for example, the group is composed of three or more client computers. It doesn't matter. In this case, until the group is configured by three or more client computers, the callback processing (that is, the existence check between the client computer and the server computer 20) is performed as described above when the group is not grouped. The

  Further, the present invention is not limited to the above-described embodiments as they are, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. Moreover, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment.

The block diagram which shows the hardware constitutions of the client server system which concerns on embodiment of this invention. FIG. 2 is a block diagram mainly showing a functional configuration of client computers 10a and 10b included in the plurality of client computers 10 shown in FIG. The block diagram which mainly shows a function structure of the server computer 20. FIG. The figure which shows an example of the data structure of the client internal memory | storage part 13a shown in FIG. The figure which shows an example of the data structure of the record hold | maintained in the client connection state table 15a shown in FIG. The sequence chart which shows the process sequence until the client computer 10a joins a group and connects to the server computer 20. FIG. The figure which shows an example of the data structure of a multicast address response packet. The figure which shows an example of the data structure of a group participation request | requirement multicast packet. The figure which shows an example of the data structure of a grouping packet. The figure which shows an example of the data structure of a grouping response packet. The figure which shows an example of the data structure of a connection request packet. The sequence chart which shows the process sequence of the survival notification packet client transmission callback process by the 1st callback process part 161a. The sequence chart which shows the process sequence of the client survival confirmation callback process by the 2nd callback process part 162a. The figure which shows an example of the data structure of a down notification packet. The sequence chart which shows the process sequence of the survival notification server transmission callback process by the 3rd callback process part 163a. The sequence chart which shows the process sequence of the server survival confirmation callback process by the server callback process part 25 of the server computer. The figure for demonstrating the flow of a packet when the client computer 10a notifies other client computer 10b and 10c and the server computer 20 of own survival. The figure for demonstrating the flow of a packet when the client computer 10b considers the client computer 10a to have gone down.

Explanation of symbols

  10a, 10b ... client computer, 11a, 11b ... client timer unit, 12a, 12b ... client connection unit, 13a, 13b ... client internal storage unit, 14a, 14b ... client received packet processing unit, 15a, 15b ... client connection state table , 16a, 16b ... client callback processing unit, 20 ... server computer, 21 ... server timer unit, 22 ... server connection standby unit, 23 ... server connection status table, 24 ... server received packet processing unit, 25 ... server callback processing 30, network, 161 a, 161 b, first callback processing unit, 162 a, 162 b, second callback processing unit, 163 a, 163 b, third callback processing unit.

Claims (4)

A first computer; a second computer connected to the first computer; a third computer connected to the first computer and the second computer;
The first computer includes transmission means for transmitting a survival notification packet including an address assigned to the first computer at a predetermined first time interval,
The second calculator is
Receiving means for receiving the transmitted survival notification packet;
A table holding a record including an address assigned to the first computer included in the received life notification packet and a reception time when the life notification packet is received;
An acquisition means for acquiring, from the table, a record including the reception time at which a difference between a current time and the reception time is greater than a predetermined second time interval, at the second time interval;
A computer system comprising: notification means for notifying the third computer that the first computer to which the address included in the acquired record has been assigned is down. The transmission means of the first computer transmits a survival notification packet including an address assigned to the first computer to the third computer at a predetermined third time interval.
The third computer is
Receiving means for receiving a survival notification packet transmitted at a third time interval by the transmitting means of the first computer;
A table holding a record including an address assigned to the first computer included in the life notification packet received by the reception unit of the third computer and a reception time when the life notification packet is received;
A record having a difference between a current time and a reception time included in a record held in the table of the third computer that is greater than a predetermined fourth time interval is set to the third time interval at the fourth time interval. Acquisition means for acquiring from the table of the computer,
The computer system according to claim 1, further comprising: a detecting unit that detects that the first computer to which an address included in the record acquired by the acquiring unit of the third computer is assigned is down.   3. The computer system according to claim 2, wherein the third time interval is longer than the first time interval. The acquisition means of the second computer obtains the record in which the difference between the current time and the reception time included in the record held in the table of the second computer is greater than the third time interval. From the table of the second computer at time intervals of
The second calculator is
First transmission means for transmitting a survival notification packet including an address included in the record acquired at the third time interval by the acquisition means of the second computer to the third computer;
A second sending means for sending a survival-notified packet including an address included in the record obtained at the third time interval by the obtaining means of the second computer to the first computer;
The receiving means of the third computer receives the survival notification packet transmitted by the first transmitting means of the second computer,
The third computer uses the reception time included in the record held in the table of the third computer including the address included in the survival notification packet received by the receiving unit of the third computer as the survival time. Update means for updating at a reception time when the notification packet is received;
The first computer further includes receiving means for receiving the survival-notified packet transmitted by the second transmitting means of the second computer,
When the transmission means of the first computer receives the alive notification-completed packet including the address assigned to the first computer by the reception means of the first computer, the third time interval The computer system according to claim 2, wherein transmission of a survival notification packet transmitted in step (1) is skipped.

Images