JP5287440B2 - Non-stop communication recovery system and method in case of failure - Google Patents

Description

  The present invention relates to a non-disruptive communication recovery system and method that can be used for a file server in a cluster environment.

  In a cluster environment, data is exchanged on a cluster basis, and it is required that the service be continued. In the cluster environment, a communication failure may occur when a service is being performed. Therefore, a method for dealing with the failure has been developed.

The occurrence of the communication failure is not limited to the field of the cluster environment, but is assumed in a field where communication is applied. A failure recovery method in a cluster system is disclosed in Patent Document 1. Patent Document 2 discloses a relay system that shortens a system stop time based on a cold standby. Patent Document 3 discloses a multiple interconnection network with improved durability against a reconfigurable failure.
Republished patent WO2005-060187 JP 2007-81933 A JP-A-5-207011 JP 11-85556 A

  However, in Patent Document 1, when a service node is switched due to a communication failure in a cluster environment, it is necessary to share all session state information on the standby system side in order to reconnect the connection. In this case, there is a problem that the resources on the standby side may increase.

  Also in Patent Document 2, when a communication failure occurs in a relay device that relays between two terminals, all the journals necessary for restart processing are used to switch this relay device to another relay device. It is necessary to secure this information, and the same problem as in Patent Document 1 remains.

  Patent Document 3 dynamically configures a network of switches, and relates to a configuration in which ports used for input / output are selectively connected by a plurality of interconnected active logical switches. It cannot be applied as it is, leaving room for improvement.

  In Patent Document 4, network information is stored in a shared memory in association with an execution system control process and a standby system control process, and communication is continued by taking over this information. In order to do this, it was necessary to share all session state information.

  In a cluster environment, in order to continue a service by exchanging data on a cluster basis, it is necessary to restore the communication state in addition to resuming the session. However, all of the methods disclosed in the above-mentioned patent documents need to secure all the session state information in order to reconnect the connection, and only resume the session using the information. Therefore, the situation is that the recovery of the communication state has not been realized.

  An object of the present invention is to provide a non-stop communication recovery system and method in the event of a failure in a cluster environment, in which a communication state is recovered from packet information and a service is continued without reconnection.

In order to achieve the above object, a non-stop communication recovery system in the event of a failure according to the present invention uses one cluster node in a cluster environment as a service system and another cluster node as a standby system, and the service system cluster node and client through a switch. A non-stop communication recovery system at the time of failure to restart communication when a failure occurs in the service cluster node when operating communication with
The switch includes a packet duplicating unit that duplicates a packet communicated between the client and the service client node through a communication port, and transmits the duplicated packet to the standby cluster node;
The standby cluster node extracts a packet related to a connection from the duplicated packet, and analyzes a connection analysis based on the extracted packet, and corresponds to an operation monitoring status of the service cluster node. And a connection restoration module for restoring a communication state based on the packet related to the connection and the connection information.

  In the above description, the present invention is constructed as a non-stop communication recovery system in the event of a failure due to a hardware configuration, but is not limited to this. The present invention may be constructed as a switch or standby cluster node used in the non-stop communication recovery system at the time of failure. Furthermore, the present invention may be constructed as a non-stop communication recovery method at the time of failure or as a non-stop communication recovery control program.

When the present invention is constructed as a switch for use in the non-disruptive communication recovery system in the event of a failure, the non-disruptive communication recovery system switch according to the present invention waits for one cluster node in the cluster environment and the other cluster node. A switch used for a non-stop communication recovery system that resumes communication when a failure occurs in the service cluster node when operating communication as a system,
The switch communicates a packet through a communication port between the service cluster node and a client, and duplicates a packet communicated between the client and the service client node through the communication port. The duplicated packet is constructed as a configuration having packet duplicating means for transmitting the duplicated packet to the standby cluster node.

When the present invention is constructed as a switch for use in the non-stop communication recovery system at the time of failure, the standby cluster node of the non-stop communication recovery system at the time of failure according to the present invention is configured such that one cluster node of the cluster environment is a service system, A standby cluster node used for a non-stop communication recovery system that resumes communication when a failure occurs in the service cluster node when operating communication with another cluster node as a standby system,
A packet analysis module that extracts a packet related to a connection from the duplicated packet, analyzes connection information based on the extracted packet, and a packet related to the connection and an operation monitoring status of the service cluster node Based on the connection information, it is constructed as a configuration having a connection restoration module for restoring the communication state.

When the present invention is constructed as a non-stop communication recovery method at the time of failure, the non-stop communication recovery method at the time of failure according to the present invention is to use one cluster node in a cluster environment as a service system and another cluster node as a standby system. When operating communication between the service-system cluster node and the client through a switch, a non-stop communication recovery method at the time of failure to restart communication when a failure occurs in the service-system cluster node,
A packet communicated between the client and the service client node through the communication port is duplicated, the duplicated packet is transmitted to the standby cluster node, and the duplicated packet is transmitted on the standby cluster node side. A packet related to the connection is extracted from the packet, connection information is analyzed based on the extracted packet, and communication is performed based on the packet related to the connection and the connection information corresponding to the operation monitoring status of the service cluster node. Build as a configuration to restore the state.

When the present invention is constructed as a non-stop communication recovery control program in the event of a failure, the non-stop communication recovery control program according to the present invention communicates with one cluster node as a service system and another cluster node as a standby system. A control program for controlling the operation of non-stop communication recovery for resuming communication when a failure occurs in the service system cluster node when operating,
In the computer constituting the standby cluster node,
A function for extracting a packet related to a connection from the duplicated packet and analyzing connection information based on the extracted packet, and a packet related to the connection and the connection information corresponding to the operation monitoring status of the service cluster node Based on the above, it is constructed as a configuration for executing the function of restoring the communication state.

  According to the present invention, when a failure occurs in a service cluster node in a cluster node of a cluster environment and the standby cluster node takes over the previous service, the standby cluster node flows in advance through the communication port of the service cluster node. The packet can be buffered (mirrored), the communication status can be restored from the packet information while tracing the communication status (for example, connection establishment / disconnection), and the service can be continued without reconnecting the connection. .

It is a block diagram which shows the non-stop communication recovery system at the time of the failure which concerns on embodiment of this invention. It is a figure which shows the sequence in the case of performing the non-stop communication recovery method at the time of failure using the non-stop communication recovery system at the time of failure which concerns on embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  As shown in FIG. 1, a non-stop communication recovery system in the event of a failure according to an embodiment of the present invention uses one cluster node in a cluster environment as a service system and the other cluster node as a standby system. When operating communication between the node (2) and the client 6, the system is intended for a non-disruptive communication recovery system at the time of failure to restart communication when a failure occurs in the service cluster node (2) It is.

  As shown in FIG. 1, the non-stop communication recovery system at the time of failure according to the embodiment of the present invention has a basic configuration in which the switch 1 is connected to the client 6 and the service system client node (2) through a communication port. And a packet duplicating means 10 for duplicating the packet communicated with the standby cluster node and transmitting the duplicated packet to the standby cluster node. Further, the standby cluster node (3) extracts a packet related to a connection from the duplicated packet, analyzes the connection information based on the extracted packet, and the service cluster node (2) And a connection restoration module 30 for restoring the communication state based on the packet related to the connection and the connection information.

  In the embodiment of the present invention, the packet duplication unit 10 of the switch 1 is used to duplicate a packet communicated between the client 6 and the service system client node (2) through the communication port, and the duplicated packet To the standby cluster node (3), and on the standby cluster node (3) side, the packet analysis module 50 is used to extract a packet related to the connection from the duplicated packet, and the extracted packet And analyzing the connection information based on the packet related to the connection and the connection information corresponding to the operation monitoring status of the service cluster node (2) using the connection restoration module 30. Restore.

  Next, a more detailed description will be given based on an example in which the non-stop communication recovery system at the time of failure according to the embodiment of the present invention is applied to a file server in a cluster environment.

  When the non-disruptive communication recovery system at the time of failure according to the embodiment of the present invention is applied to a file server in a cluster environment, the service cluster node corresponds to the service file server 2 shown in FIG. Corresponds to the standby file server 3 shown in FIG.

  As shown in FIG. 1, a non-disruptive recovery system for a failure according to an embodiment of the present invention applied to a file server in a cluster environment includes a switch 1, a service file server 2, a standby file server 3, and a client 6. And have.

  The switch 1 exchanges packets between the service file server 2 and the client 6 through a communication port, and has a packet duplicating means 10. The packet duplicating means 10 duplicates a packet exchanged through the communication port, and flows the duplicated packet to a communication port different from the communication port. As will be described later, the standby file server 3 is connected to the different communication ports, and the packet duplicating means 10 transmits the duplicated packet to the standby file server 3.

  The standby file server 3 provides services to the client 6 in the same manner as the service server 2. When the service server 2 is used as a working server, a failure occurs in the service server 2. In this case, the service is provided to the client 6 instead of the service server 2. The standby file server 3 includes a connection restoration module 30, an alive monitoring unit 40, and a packet analysis module 50. The life and death monitoring means 40 is provided in the standby file server 3, but is not limited thereto, and may be provided independently of the standby file server 3.

  The packet analysis module 50 extracts a packet related to a connection from the duplicated packet, and analyzes connection information based on the extracted packet. The packet receiving module 51, the packet extracting module 52, and a packet analyzing module Means 53, a packet buffer 54, and a connection information storage unit 55 are provided.

  The packet receiving means 51 is connected to the communication port selected by the packet duplicating means 10, receives the packet duplicated by the packet duplicating means 10, and delivers the received packet to the packet extracting means 52. is there.

  The packet extraction means 52 extracts a packet related to a connection from the received packet, specifically, a packet related to connection establishment / disconnection, NFS (Network File System) / CIFS (Common Internet File System) request / reply, The extracted packet is transferred to the packet analysis means 53.

  The packet analysis unit 53 receives the packet extracted by the packet extraction unit 52, analyzes connection information from the extracted packet, and outputs the analysis result to the packet buffer 54 and the connection information storage unit 55. Is.

  The packet buffer 54 obtains an analysis result from the packet analysis means 53 and stores a packet related to a connection exchanged for each connection, specifically a packet related to an NFS / CIFS request / reply. The packet buffer 54 stores the packet in a FIFO (First-In First-Out) system based on a registration request from the packet analysis unit 53. Here, the FIFO method is a method in which stored data is taken out in the order in which they were stored, and the most recently stored data is taken out last.

  The connection information storage unit 55 obtains an analysis result from the packet analysis unit 53 and analyzes the connection information (for example, TCP (Transmission Control Protocol) / IP (Internet Protocol) header, NFS) analyzed by the packet analysis unit 53. A file handler and an SMB (Server Message Block) header) are stored in a table format. The alive monitoring means 40 monitors the operating status of the service file server 2 and monitors whether a failure has occurred in the service file server 2.

  The connection restoration module 30 restores the communication state based on the packet related to the connection and the connection information corresponding to the operation monitoring status of the service cluster node 2. Specifically, the connection restoration module 30 restores connection information that has been communicated in the past from the header information and data of the packet related to the connection, and restores the communication state based on the restored connection information. . Then, when the connection information cannot be restored, the connection restoration module 30 determines the communication port of the switch from the packet relating to the connection, and transmits a reset to the switch to prompt the restoration of the communication state. In addition, the connection restoration module 30 analyzes the packet related to the connection extracted from the duplicated packet and the connection information analyzed based on the extracted packet, and determines the communication state when the connection information is available. When the connection information is not prepared, a command to disconnect the communication state is transmitted to the switch 1.

  More specifically, the connection restoration module 30 includes a connection analysis unit 31, a connection restoration unit 32, and a connection disconnection unit 33.

  The connection analysis unit 31 obtains a monitoring result by the life and death monitoring unit 40, and based on the information stored in the connection storage unit 55 and the information stored in the packet buffer 54 (connection state) ).

  The connection restoring means 32 receives the result analyzed by the connection analyzing means 31, restores the connection information that has been communicated in the past from the header information and data of the packet related to the connection, and converts the connection information into the restored connection information. Based on this, the communication state is restored. Then, the connection restoration means 32 determines the communication port of the switch 1 from the packet related to the connection and resets it to the switch 1 when the connection information cannot be restored as a result of the analysis by the connection analysis means 31 ( RST) is transmitted to prompt the restoration of the communication state. The connection restoration unit 32 receives the result of the analysis by the connection analysis unit 31 and analyzes the packet related to the connection extracted from the duplicated packet and the connection information analyzed based on the extracted packet. The communication state is restored when the connection information is available.

  The connection disconnecting unit 33 receives a result analyzed by the connection analyzing unit 31 and transmits a reset (RST) to the switch 1 in response to an instruction from the connection analyzing unit 31 when the connection information is not complete. is there. The switch 1 receives the reset from the connection disconnecting means 33 and disconnects the connection.

  In the above description, the embodiment of the present invention is constructed as a non-stop communication recovery system in the event of a failure by a hardware configuration, but is not limited to this. The connection restoration module 30 and the packet analysis module 50 that construct the standby file server 3 in the embodiment of the present invention may be realized on software by causing a CPU to execute a program.

  Next, a case where the non-stop communication recovery method at the time of failure in the cluster environment is executed using the non-stop communication recovery system according to the embodiment of the present invention will be described with reference to FIGS.

  Hereinafter, a non-stop communication recovery method when the client 6 accesses a file using the service file server 2 will be described.

  When a connection establishment request / response packet is transmitted / received between the client 6 and the service file server 2 through the communication port of the switch 1 (steps S1, S2, S3, and S4 in FIG. 2), the packet duplication of the switch 1 is performed. The means 10 receives the packet transmitted / received through the communication port, duplicates the packet, and transmits the duplicated packet to the packet receiving means 51 of the standby file server 3 through a communication port different from the communication port. (Step S5 in FIG. 2).

  When the packet receiving means 51 receives the packet duplicated by the packet duplicating means 10, it transmits the packet as it is to the packet extracting means 52. When the packet extraction unit 52 receives the packet from the packet reception unit 51, the packet extraction unit 52 determines whether the request is a connection establishment request or a response to the connection establishment request based on a flag of the TCP header of the packet, and connects the connection information to the connection information. The information is output to the information storage unit 55. The connection information storage unit 55 records a packet related to the connection by the FIFO method based on the output from the packet extraction unit 53.

  Next, when a file access request / response packet is returned from the client 6 to the service file server 2 through the communication port of the switch 1, the packet duplicating means 10 is also returned through the communication port. The packet is duplicated, and the duplicated packet is transmitted to the packet receiving means 51 of the standby file server 3 through a communication port different from the communication port.

  When the packet receiving means 51 receives the packet duplicated by the packet duplicating means, the packet receiving means 51 transmits the packet as it is to the packet extracting means 52.

  When the packet extraction unit 52 receives the packet from the packet reception unit 51, the packet extraction unit 52 extracts a packet related to the connection from the packet and outputs the packet to the packet analysis unit 53.

  When the packet analysis unit 53 receives the packet from the packet extraction unit 52, the packet analysis unit 53 reads the connection information stored in the connection information storage unit 55, and stores it in the TCP header in the packet received from the packet extraction unit 52. On the basis of the communication information from the client 6 to the service file server 2 is determined based on the connection information stored in the connection information storage unit 55. Based on the NFS / CIFS header in the packet, it is determined whether it is a file access request or a response to the file access request, and the packet received from the packet extraction means 52 is output to the packet buffer 54. The packet buffer 54 stores the packet received from the packet analysis unit 53 in a FIFO manner.

  Here, when a connection disconnection request packet is returned from the client 6 to the service-related file server 2 through the communication port of the switch 1, the packet of the switch 1 is transmitted in the same manner as the processing at the time of the connection opening request. The duplication unit 10 duplicates the packet returned through the communication port, and transmits the duplicated packet to the packet extraction unit 52 of the standby file server 3.

  When receiving the connection disconnection request packet extracted by the packet extraction unit 52, the packet analysis unit 53 receives the connection disconnection request packet extracted by the packet extraction unit 52, based on the TCP header of the received packet. When the connection information of the connection disconnection request exists in the connection information stored in the server, a command to delete the connection information is output to the connection information storage unit 55. The connection information storage unit 55 receives a command from the packet analysis unit 52 and deletes connection information corresponding to the connection information of the connection disconnection request from the stored connection information.

  Next, a case where a failure occurs in the service file server 2 will be described.

  The alive monitoring means 40 of the standby file server 3 monitors the operating state of the service file server 2. When a failure of the service file server 3 occurs while being monitored, the alive monitoring means 40 immediately informs the connection restoration module 30 of the standby file server 3 that a failure has occurred in the service file server 2. Notice.

  When the connection restoration module 30 receives a notification that a failure has occurred in the service file server 2 from the alive monitoring unit 40, the connection analysis unit 31 of the connection restoration module 30 includes the connection information storage unit 55 and the packet buffer. The information stored in 54 is read, the communication state (connection state) is analyzed based on the read information, and the analysis result is output to the connection restoration means 32 and the connection disconnection means 33.

  The connection restoring means 32 receives the result analyzed by the connection analyzing means 31, restores the connection information that has been communicated in the past from the header information and data of the packet related to the connection, and converts the connection information into the restored connection information. Based on this, the communication state is restored. Then, the connection restoration means 32 determines the communication port of the switch 1 from the packet related to the connection and resets it to the switch 1 when the connection information cannot be restored as a result of the analysis by the connection analysis means 31 ( RST) is transmitted to prompt the restoration of the communication state. The connection restoration unit 32 receives the result of the analysis by the connection analysis unit 31 and analyzes the packet related to the connection extracted from the duplicated packet and the connection information analyzed based on the extracted packet. The communication state is restored when the connection information is available.

  Further, when the connection analyzing unit 31 analyzes that the request is received last based on the last request / reply information remaining in the packet buffer 54, the connection restoring unit 32 performs processing for the request. A command to restart the service is output to the standby file server 3.

  The standby file server 3 obtains the result of restoration by the connection restoration means 32 described above, and resumes the service for the client 6 on behalf of the service file server 2. When the standby file server 3 receives an instruction to restart the service from the process for the request from the connection restoration unit 32, the standby file server 3 restarts the service from the process for the request (step S6 in FIG. 2). S7).

  The connection analyzing unit 31 reads the connection information stored in the connection information storage unit 55 and the packet related to the connection stored in the packet buffer 54, and the connection status is based on the read data. When a request / reply packet associated with the analyzed connection information does not exist in the packet buffer 54, a disconnection command is output to the connection disconnecting means 33.

  The connection disconnecting means 33 receives a result analyzed by the connection analyzing means 31 and transmits a reset (RST) to the switch 1 in accordance with an instruction from the connection analyzing means 31 when the connection information is not complete. The switch 1 receives the reset from the connection disconnecting means 33 and disconnects the connection.

  As described above, according to the embodiment of the present invention, when a failure occurs in the service cluster node in the cluster node of the cluster environment and the standby cluster node takes over the previous service, the service is previously performed in the standby cluster node. The packet flowing through the communication port of the host cluster node is buffered (mirrored), the communication state is restored from the packet information while tracing the communication state (for example, connection establishment / disconnection), and the connection is not reconnected. Can continue service.

  Further, according to the embodiment of the present invention, the file access connection that was in service can be maintained even if a service file server fails over, and communication can be continued without reconnection processing. The reason is that the packet communicated with the service file server can be mirrored with the standby file server, the connection state can be traced, and the connection can be reconstructed.

  In addition, the processing load for restoring the connection in the standby file server can be kept low. The reason is that only the packet flowing as a mirror is processed during standby, not the connection itself, but the connection is restored based on the connection information and the packet stored when a failure occurs.

  In addition, a connection that is frequently used at the time of connection recovery can be recovered with priority. The reason is that recording is performed in the packet buffer based on the FIFO method, packet information communicated in the past in the past is prioritized and the connection is restored using this information.

  As another embodiment of the present invention, when extended to an N-node cluster environment, for example, if the cluster node is 10 nodes, a switch 1 having 1 Gbps port × 10 and 10 Gbps port × 1 is prepared. Each node equipped with the connection restoration module 30 is connected to a 1 Gbps port, and the packet analysis module 50 is connected to a 10 Gbps port to perform packet mirroring. It is also possible to restore the connection by transferring the stored connection information to the connection restoration module 30 of any node.

  In the embodiment described above, the example in which the non-stop communication recovery system at the time of failure according to the embodiment of the present invention is applied to the file server in the cluster environment has been described. However, the present invention is not limited to this. As long as one cluster node in a cluster environment is a service system and the other cluster node is a standby system, and the system operates communication between the service system cluster node and a client through a switch, there is no need for the embodiment of the present invention. By applying the stop communication recovery system, communication can be resumed when a failure occurs in the service cluster node, and the non-stop communication at the time of failure can be recovered.

  According to the present invention, for example, it can be used for a file server in a cluster environment, and can contribute to realizing non-stop communication recovery in the event of a failure.

1 Switch 2 Service File Server 3 Standby File Server 30 Connection Restoration Module 50 Packet Analysis Module

Claims (19)

When one cluster node in a cluster environment is a service system and the other cluster node is a standby system, and a communication is operated between the service system cluster node and a client through a switch, a failure has occurred in the service system cluster node Non-stop communication recovery system at the time of failure to resume communication at the time,
The switch
A packet duplicating means for duplicating a packet communicated between the client and the service cluster node through a communication port, and transmitting the duplicated packet to the standby cluster node;
The standby cluster node is:
A packet analysis module that extracts a packet related to a connection from the duplicated packet and analyzes connection information based on the extracted packet;
Non-disruptive communication recovery at the time of failure characterized by having a connection restoration module for restoring the communication state based on the packet related to the connection and the connection information corresponding to the operation monitoring status of the service cluster node system.   2. The connection restoration module restores connection information that has been communicated in the past recently from header information and data of a packet related to the connection, and restores the communication state based on the restored connection information. Non-stop communication recovery system at the time of failure described in 1.   The connection restoration module is configured to determine a communication port of the switch from a packet related to the connection when the connection information cannot be restored, and to transmit a reset to the switch to promote restoration of a communication state. The non-stop communication recovery system at the time of failure described in 1.   The connection restoration module analyzes a packet related to the connection extracted from the duplicated packet and connection information analyzed based on the extracted packet, and restores a communication state when connection information is prepared, The non-stop communication recovery system at the time of failure according to claim 1, wherein a command for disconnecting a communication state is transmitted to the switch when connection information is not prepared.   The non-stop communication recovery system at the time of failure according to claim 1, wherein the packet analysis module stores a packet related to the connection for each connection by a FIFO method. When one cluster node in a cluster environment is a service system and the other cluster node is a standby system, and a communication is operated between the service system cluster node and a client through a switch , a failure has occurred in the service system cluster node A standby cluster node used in a non-stop communication recovery system in the event of a failure to resume communication
A packet analysis module that extracts a packet relating to a connection from a packet that is duplicated by the switch and communicated between the client and the service cluster node through a communication port, and that analyzes connection information based on the extracted packet;
Non-disruptive communication recovery at the time of failure characterized by having a connection restoration module for restoring the communication state based on the packet related to the connection and the connection information corresponding to the operation monitoring status of the service cluster node The standby cluster node of the system. The connection restoration module, claim from the header information and data of the packet relating to the connection to restore the connection information which has been communicated to the recent past, is to restore the communication state based on the restored connection information 6 Non-stop communication recovery system standby cluster node described in 1. The connection restoration module is configured to determine a communication port of the switch from a packet related to the connection when the connection information cannot be restored, and to transmit a reset to the switch to promote restoration of a communication state. nonstop communication recovery system standby cluster node according to 6. The connection restoration module analyzes a packet related to the connection extracted from the duplicated packet and connection information analyzed based on the extracted packet, and restores a communication state when connection information is prepared, The non-stop communication recovery system standby cluster node according to claim 6 , wherein a command for disconnecting a communication state is transmitted to the switch when connection information is not available. The non-stop communication recovery system standby cluster node according to claim 6 , wherein the packet analysis module stores a packet related to the connection for each connection by a FIFO method. When one cluster node in a cluster environment is a service system and the other cluster node is a standby system, and a communication is operated between the service system cluster node and a client through a switch, a failure has occurred in the service system cluster node Non-stop communication recovery method at the time of failure to restart communication when
Duplicating a packet communicated between the client and the service cluster node through a communication port, and sending the duplicated packet to the standby cluster node;
On the standby cluster node side, extract a packet related to the connection from the duplicated packet, analyze connection information based on the extracted packet,
A non-disruptive communication recovery method at the time of failure, wherein the communication state is restored based on the packet related to the connection and the connection information in response to the operation monitoring status of the service cluster node. In the standby cluster node side, to restore the connection information which has been communicated to the recent past and a header information and data of the packet relating to the connection, to claim 11 for restoring the communication state based on the restored connection information Non-stop communication recovery method at the time of failure described. In the standby cluster node side, the when the connection information is not restored, indexing the communication ports of the switch from the packet relating to the connection, according to claim 11 to promote recovery of the communication state by sending a reset to the switch Non-stop communication recovery method at the time of failure described in 1. On the standby cluster node side, the packet related to the connection extracted from the duplicated packet and the connection information analyzed based on the extracted packet are analyzed, and the communication state is restored when the connection information is available. 12. The non-stop communication recovery method at the time of failure according to claim 11 , wherein a command for disconnecting a communication state is transmitted to the switch when connection information is not prepared. 12. The non-disruptive communication recovery method at the time of failure according to claim 11 , wherein a packet related to the connection is stored and buffered for each connection by a FIFO method on the standby cluster node side. When one cluster node in a cluster environment is a service system and the other cluster node is a standby system, and a communication is operated between the service system cluster node and a client through a switch , a failure has occurred in the service system cluster node Control program for controlling the operation of non-stop communication recovery to resume communication when
In the computer constituting the standby cluster node,
A function of extracting a packet related to a connection from a packet communicated between the client and the service-related cluster node through a communication port copied by the switch, and analyzing connection information based on the extracted packet;
A non-stop communication recovery control program for executing a function of restoring a communication state based on a packet related to the connection and the connection information in response to an operation monitoring status of the service cluster node. In the computer,
The non-stop according to claim 16 , wherein connection information that has been communicated in the past in the past is restored from packet header information and data related to the connection, and the communication state is restored based on the restored connection information. Communication recovery control program. In the computer,
17. The function according to claim 16 , wherein when the connection information cannot be restored, a function for prompting restoration of a communication state is performed by determining a communication port of the switch from a packet related to the connection and transmitting a reset to the switch. Stop communication recovery control program. In the computer,
Analyzing the connection-related packet extracted from the duplicated packet and the connection information analyzed based on the extracted packet, restores the communication state when the connection information is complete, and the connection information is not complete The non-stop communication restoration control program according to claim 16 , wherein a function for outputting a command to disconnect the communication state to the switch is executed.

Images