WO2016101409A1 - Data switching method, device and system - Google Patents

Abstract

Disclosed are a data switching method, device and system. The method comprises: under the condition that a first device receives a transmission control protocol (TCP) switching request, transmitting first information to a second device, wherein the first information is used for the second device to copy a socket connection of the first device; and transmitting second information and third information to the second device, wherein the second information is used for the second device to generate a socket ID of the socket connection, and the third information is used for assigning relevant information about a network file system (NFS) of the first device into the socket connection corresponding to the socket ID.

Description

Data switching method, device and system Technical field

The present invention relates to the field of network data storage, and in particular, to a data switching method, device and system.

Background technique

File lock is a means of keeping files synchronized. When multiple users operate the same file at the same time, the file lock can ensure that data does not conflict. Many database softwares need file lock support in the process of reading and writing data. Network File System (NFS) is a powerful network file system. File locks play a vital role in maintaining file synchronization. Therefore, the maintenance of file locks is also crucial for NFS. In cluster mode, if a node in the data access is abnormal, it may face the risk of file lock information loss. In most cases, the client checks the lock status through the state protocol to recover the lock, but through the lock state. The monitoring is to recover the lock for a long time, and in the special case, some locks cannot be recovered, and a technique for quickly recovering or saving the lock state is urgently needed to solve the problem.

Summary of the invention

The present invention provides a data switching method, device and system, the main purpose of which is to solve the technical problem of how to quickly recover file lock information in the event of a node failure.

A method of data switching, including:

When the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket connection of the first device;

Transmitting, by the first device, the second information and the third information to the second device, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used for The network file system NFS related information of the first device is assigned to the socket connection corresponding to the socket ID.

Optionally, in the case that the first device receives the TCP switching request, the sending the first information to the second device includes:

When the first device receives the TCP switching request, the TCP module of the first device acquires the first information, where the first information includes at least a keyword of a control block structure in the first device TCP module. And a keyword of the management structure in the first device socket module;

The TCP module of the first device sends the first information to a TCP module of the second device.

Optionally, the sending the second information to the second device includes:

The TCP module of the first device acquires the second information, where the second information includes at least the packet in the TCP module sending buffer and the packet in the socket receiving buffer.

The TCP module of the first device sends the second information to the TCP module of the second device through a cluster channel, so that the TCP module of the second device assigns the second information to the copied socket connecting.

Optionally, the sending the third information to the second device includes:

The NFS module of the first device acquires the third information, where the third information includes at least information about a control plane management structure of the NFS module and an input and output IO request that is not completed by the NFS module, where the NFS module The information of the control plane management structure includes NFS lock information;

The NFS module of the first device sends the third information to the NFS module of the second device by using a cluster channel.

Optionally, the NFS module of the first device acquires the third information, including:

The NFS module of the first device acquires NFS lock information;

The NFS module of the first device sends the third information to the NFS module of the second device by using a cluster channel, including:

The NFS module of the first device encapsulates the NFS lock information according to a packet format, where the packet format includes at least the information of the socket connection, the packet fragment number, the number of the NFS lock information, and the The end identifier of the message;

The NFS module of the first device sends the encapsulated NFS lock information to the second NFS module.

Optionally, after the sending the first information to the second device, the method further includes:

The first device receives a zero window message sent by the second device.

Optionally, after the sending the second information and the third information to the second device, the method further includes:

After the TCP module of the first device sends the second information to the second device, the The NFS module of the first device sends a message that the switching is completed.

If the NFS module of the first device receives the message that the NFS module of the first device sends a switching completion, and the NFS module of the first device has sent the third information to the second device, The NFS module of the first device closes the TCP switching request;

If the NFS module of the first device receives the message that the first NFS module sends a switching completion, and the NFS module of the first device does not send the third information to the second device, The NFS module of the first device closes the TCP switching request after sending the third information to the second device.

A method of data switching, including:

Receiving, by the second device, the first information that is sent to the second device when the first device receives the transmission control protocol TCP switching request, and copies the socket connection of the first device according to the first information;

Receiving, by the second device, the second information and the third information that are sent by the first device, generating a socket ID of the socket connection according to the second information, and using the third information, the network file system of the first device The NFS related information is assigned to the socket connection corresponding to the socket ID.

Optionally, the second device receives the second information and the third information that are sent by the first device, and generates the socket ID of the socket connection according to the second information, including:

The TCP module of the second device receives the second information;

The TCP module of the second device assigns the second information to the copied socket connection, and generates a socket ID of the socket connection;

The NFS module of the second device receives the third information;

The NFS module of the second device matches the socket connection corresponding to the socket ID according to the quintuple information in the third information, and if yes, assigns the third information to the socket ID Corresponding socket connection.

Optionally, the NFS module of the second device matches the socket connection corresponding to the socket ID according to the quintuple information in the third information, and if yes, assigns the third information to the After the socket connection corresponding to the socket ID, the method further includes:

The NFS module of the second device sends a message that the switching is completed to the second TCP module.

The protocol stack IP layer of the second device opens and receives the packet of the NET, and the TCP module of the second device sends a window recovery packet to the first device, and the NFS module of the second device sends and receives the packet.

Optionally, after the receiving, by the second device, the first information sent by the first device, the method further includes:

Sending a zero window message to the first device.

A first device includes a first transmission control protocol TCP module and a first network file system NFS module;

The first TCP module is configured to: when the first device receives the transmission control protocol TCP switching request, send the first information to the second device, where the first information is used by the second device to be copied. a socket connection of the first device;

The first TCP module is further configured to: send second information to the second device, where the second information is used by the second device to generate a socket ID of the socket connection;

The first NFS module is configured to: send third information to the second device, where the third information is used to assign NFS related information of the first device to a socket connection corresponding to the socket ID. .

Optionally, the first TCP module includes:

a first acquiring unit, configured to: when the first device receives the TCP switching request, acquire first information, where the first information includes at least a keyword of a control block structure in the first device TCP module And a keyword of the management structure in the first device socket module;

The first sending unit is configured to: send the first information to the second TCP module.

Optionally, the first TCP module includes:

a second acquiring unit, configured to: acquire the second information, where the second information includes at least a packet in the TCP module sending buffer and a packet in the socket receiving buffer;

The second sending unit is configured to: send the second information to the second TCP module by using a cluster channel, so that the second TCP module assigns the second information to the copied socket connection.

Optionally, the first NFS module includes:

The third obtaining unit is configured to: acquire the third information, where the third information includes at least information about a control plane management structure of the NFS module and an input/output IO request that is not completed by the NFS module, where the NFS module The information of the control plane management structure includes NFS lock information;

The third sending unit is configured to: send the third information to the second NFS module by using a cluster channel.

Optionally, the first NFS module includes:

The fourth obtaining unit is configured to: obtain NFS lock information;

The encapsulating unit is configured to: encapsulate the NFS lock information according to a packet format, where the packet format includes at least the socket connection information, a packet fragment number, a number of the NFS lock information, and the packet The end of the text; and

The fourth sending unit is configured to: send the encapsulated NFS lock information to the second NFS module.

Optionally, the first TCP module further includes:

The first receiving unit is configured to: receive a zero window message sent by the second device.

Optionally, the first TCP module further includes:

The fifth sending unit is configured to: after sending the second information to the second device, send a message that the switching is completed to the first NFS module;

The first NFS module further includes:

And the closing unit is configured to: if the first NFS module receives the message that the first NFS module sends a switching completion, and the first NFS module has sent the third information to the second device, Close the TCP switching request;

The shutting down unit is further configured to: if the first NFS module receives the message that the first NFS module sends a switch completion, and the first NFS module does not send the third message to the second device The information is closed after the third information is sent to the second device.

A second device includes a second transmission control protocol TCP module and a second network file system NFS module;

The second TCP module is configured to: when the first device receives the TCP switching request, receive the first information sent by the first device, and copy the socket of the first device according to the first information connection;

The second TCP module is configured to: receive second information sent by the first device, and generate a socket ID of the socket connection according to the second information;

The second NFS module is configured to receive the third information sent by the first device, and assign the NFS related information of the first device to the socket connection corresponding to the socket ID according to the third information.

Optionally, the second TCP module includes:

a second receiving unit, configured to: receive the second information;

a generating unit, configured to: assign the second information to the copied socket connection, and generate a socket ID of the socket connection;

The second NFS module includes:

a third receiving unit, configured to: receive the third information;

a matching unit, configured to: match a socket connection corresponding to the socket ID according to the quintuple information in the third information;

The assignment unit is configured to: if matched, assign the third information to the socket connection corresponding to the socket ID.

Optionally, the second NFS module further includes:

a sixth sending unit, configured to: send a message that the switching is completed to the second TCP module;

The second TCP module further includes:

The seventh sending unit is configured to: send a window recovery message to the first device.

Optionally, the second TCP module further includes:

The eighth sending unit is configured to: send a zero window message to the first device.

A system comprising the first device described above and the second device.

A computer readable storage medium storing program instructions that, when executed, implement the methods described above.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

BRIEF abstract

1 is a schematic flow chart of a first embodiment of a method for data switching;

2 is a schematic flow chart of a second embodiment of a method for data switching;

3 is a schematic flow chart of a third embodiment of a method for data switching;

4 is a schematic flow chart of a fourth embodiment of a method for data switching;

FIG. 5 is a schematic flowchart of a fifth embodiment of a method for data switching;

6 is a schematic flow chart of a sixth embodiment of a method for data switching;

7 is a schematic diagram of an interaction process of a seventh embodiment of a method for data switching;

8 is a schematic diagram of functional modules of a first embodiment of a first device of the present invention;

9 is a schematic diagram of functional modules of a second embodiment of a first device of the present invention;

10 is a schematic diagram of functional modules of a third embodiment of a first device of the present invention;

11 is a schematic diagram of functional modules of a first embodiment of a second device of the present invention;

12 is a schematic diagram of functional modules of a second embodiment of a second device of the present invention;

13 is a schematic diagram of functional modules of a third embodiment of a second device of the present invention;

Figure 14 is a schematic diagram of functional modules of a first embodiment of the system of the present invention.

The implementation, functional features, and advantages of the present invention will be further described in conjunction with the embodiments.

Preferred embodiment of the invention

It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The invention provides a method of data switching.

Referring to FIG. 1, FIG. 1 is a schematic flowchart of a first embodiment of a method for data switching.

In the first embodiment, the data switching method includes:

Step 101: If the first device receives the TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket connection of the first device.

Optionally, in the case that the first device receives the Transmission Control Protocol (TCP) switching request, the sending the first information to the second device includes:

When the first device receives the TCP switching request, the TCP module of the first device acquires first information, where the first information includes at least a key of a control block structure in the TCP module of the first device. The word and the keyword of the management structure in the first device socket module;

The TCP module of the first device sends the first information to a TCP module of the second device.

The first device may be a faulty node, and the second device may be a takeover node.

If the faulty node is faulty or the faulty node is upgraded, the user can manually migrate the IP to the takeover node, and then power off the faulty node. After the IP address of the faulty node is migrated, the TCP module of the faulty end obtains the connection on the current switched IP. The faulty TCP module acquires the control block structure of the TCP module and the key field in the management structure of the socket module, and sends the obtained control block structure of the TCP module and the key field in the management structure of the socket module to the cluster channel through the cluster channel. Take over the node's TCP module so that the TCP module that takes over the node can clone a new socket connection in the first time.

Step 102: The first device sends the second information and the third information to the second device, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to The NFS related information of the first device is assigned to a socket connection corresponding to the socket ID.

Optionally, the sending the second information to the second device includes:

The TCP module of the first device acquires the second information, where the second information includes at least the packet in the TCP module sending buffer and the packet in the socket receiving buffer.

The TCP module of the first device sends the second information to the second TCP module through a cluster channel, so that the TCP module of the second device assigns the second information to the copied socket connection. .

Optionally, the sending the third information to the second device includes:

The network file system (NFS) module of the first device acquires the third information, where the third information includes at least information about a control plane management structure of the NFS module and an unfinished NFS module. An input/output (IO) request, the information of the control plane management structure of the NFS module includes NFS lock information;

The first NFS module sends the third information to the second NFS module through a cluster channel.

The TCP layer of the faulty end notifies the NFS service layer that the NFS service and the TCP layer start the switching of the connection at the same time. NFS receives the switch message, NFS stops the IO operation of the back-end disk and the packet of the front-end disk; NFS collects the information of the switch connection, including the NFS control plane management structure and the unfinished IO request sent to the pair through the cluster channel. On the end node, the NFS control plane management structure includes NFS lock information. At the same time, the TCP layer also collects the data packets in the TCP send buffer and the packets in the socket receiving ring, and sends them to the TCP layer of the takeover node through the cluster channel, and takes over the TCP layer of the node. To assign this information to a new socket connection, complete the migration of the TCP connection. Both NFS and TCP are performed simultaneously here.

The TCP module of the takeover node restores the received switch data to the new socket connection. The TCP module of the takeover node processes all the data and sends the new socket ID to the NFS module of the takeover node. The NFS module that takes over the node needs to temporarily store the received switching data and then restore the NFS lock information. The recovery process of the NFS lock is to match the temporarily stored switching data with the new socket ID, and the matching condition is the quintuple information (protocol, source-destination IP, source-destination port) of the TCP layer and the new socket ID. The corresponding socket connections match.

Optionally, the sending the third information to the second device includes:

The NFS module of the first device acquires NFS lock information;

The NFS module of the first device encapsulates the NFS lock information according to a packet format, where the packet format includes at least the information of the socket connection, the packet fragment number, the number of the NFS lock information, and the The end identifier of the message;

The NFS module of the first device sends the encapsulated NFS lock information to the NFS module of the second device.

In order to implement the fast recovery of the NFS lock information, when the system detects that the faulty node access is abnormal, the port of the faulty node is triggered to be switched. After the target node of the switchover is determined, the NFS lock information migration is started.

After receiving the NFS lock information migration message, the faulty node first finds the lock of the corresponding socket information according to the socket information, and encapsulates the lock information according to the packet format shown in Table 1. When the lock information is too much, the message is fragmented. After receiving the message, the takeover node performs group packet processing according to the sequence number of the message fragment. After the NFS lock information is successfully sent, the NFS lock information of the corresponding connection of the node is cleared until all the connections that need to be switched on the faulty node are sent.

Message type Socket link information Packet fragment number Lock 1 Lock 2 ... Lock n End mark

Table 1

After the receiving node receives the switching packet of the faulty node, it will perform corresponding processing according to the packet type. When the packet information of the lock transition is received, the packet end information is marked according to the end of the packet and the file lock information needs to be grouped, and it is determined that if the currently connected port switching is successful, the migrated file is parsed on the node. Lock the information and actively initiate a lock recovery operation to lock the file on the failed node Resume on the takeover node; if the currently connected port switchover is not completed, it will wait. When the maximum delay time of the client lock operation is exceeded (the NFSv4 protocol defaults to 90s as the lease time), the migrated lock information is released because this Even if the recovery, the lock has expired on the client, losing the value of recovery, but the port switching speed is often much faster than the lock migration speed, only in rare cases, the lock migration is ahead of the port switching and the delay exceeds the customer. The default maximum delay case.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

The invention further provides a method of data switching.

Referring to FIG. 2, FIG. 2 is a schematic flowchart of a second embodiment of a method for data switching.

In the second embodiment, after step 101, the method further includes:

Step 103: Receive a zero window message sent by the second device.

Specifically, after the takeover node establishes a new socket connection, the takeover node sends a zero window message to the faulty node, and temporarily disables the corresponding NET to receive the packet at the network layer IP layer.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

Referring to FIG. 3, FIG. 3 is a schematic flowchart of a third embodiment of a method for data switching.

In the first embodiment, after step 102, the method further includes:

Step 104: After the TCP module of the first device sends the second information to the second device, the message is sent to the NFS module of the first device, and the NFS module of the first device closes the TCP. Switching request;

If the NFS module of the first device receives the message that the first NFS module sends the switching completion, and the NFS module of the first device has sent the third information to the second device, The NFS module of the first device closes the TCP switching request;

If the NFS module of the first device receives the message that the first NFS module sends a switching completion, and the NFS module of the first device does not send the third information to the second device, The NFS module of the first device closes the TCP switching request after sending the third information to the second device.

After the TCP switchover of the faulty end is completed, the NFS service is sent to the NFS service. The NFS receives the message. If the NFS is also switched, the NFS actively closes the request. If the NFS has not completed the data switchover, the NFS switchover is performed. Close the connection when done.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

Referring to FIG. 4, FIG. 4 is a schematic flowchart of a fourth embodiment of a method for data switching.

In the fourth embodiment, the data switching method includes:

Step 401: The second device receives the first information that is sent by the first device when receiving the TCP switching request, and copies the socket connection of the first device according to the first information.

The first device may be a faulty node, and the second device may be a takeover node.

If the faulty node is faulty or the faulty node is upgraded, the user can manually migrate the IP to the takeover node, and then power off the faulty node. After the IP address of the faulty node is migrated, the TCP module of the faulty end obtains the connection on the current switched IP. The faulty TCP module acquires the control block structure of the TCP module and the key field in the management structure of the socket module, and sends the obtained control block structure of the TCP module and the key field in the management structure of the socket module to the cluster channel through the cluster channel. Take over the node's TCP module so that the TCP module that takes over the node can clone a new socket connection in the first time.

Step 402: Receive second information and third information that are sent by the first device, generate a socket ID of the socket connection according to the second information, and use the third information according to the third information. The NFS related information is assigned to the socket connection corresponding to the socket ID.

Optionally, the receiving the second information and the third information that are sent by the first device, and generating the socket ID of the socket connection according to the second information, including:

The TCP module of the second device receives the second information;

The TCP module of the second device assigns the second information to the copied socket connection, and generates a socket ID of the socket connection;

The NFS module of the second device receives the third information;

The NFS module of the second device matches the socket connection corresponding to the socket ID according to the quintuple information in the third information, and if yes, assigns the third information to the socket ID Corresponding socket connection.

The TCP layer of the faulty end notifies the NFS service layer that the NFS service and the TCP layer start the switching of the connection at the same time. NFS receives the switch message, NFS stops the IO operation of the back-end disk and the packet of the front-end disk; NFS collects the information of the switch connection, including the NFS control plane management structure and the unfinished IO request sent to the pair through the cluster channel. On the end node, the NFS control plane management structure includes NFS lock information. At the same time, the TCP layer also collects the data packets in the TCP transmission buffer and the packets in the socket receiving ring, and sends them to the TCP layer of the takeover node through the cluster channel. The TCP layer of the takeover node needs to assign the information to the new socket connection. Complete the migration of the TCP connection. Both NFS and TCP are performed simultaneously here.

The TCP module of the takeover node restores the received switch data to the new socket connection. The TCP module of the takeover node processes all the data and sends the new socket ID to the NFS module of the takeover node. The NFS module that takes over the node needs to temporarily store the received switching data and then restore the NFS lock information. The recovery process of the NFS lock is to match the temporarily stored switching data with the new socket ID, and the matching condition is the quintuple information (protocol, source-destination IP, source-destination port) of the TCP layer and the new socket ID. The corresponding socket connections match.

In order to implement the fast recovery of the NFS lock information, the system detects the faulty node access abnormality and triggers the port switching of the faulty node. After the target node of the switchover is determined, the NFS lock information migration begins.

After the faulty node receives the NFS lock information migration message, it first finds the lock of the corresponding sockett information according to the socket information, and encapsulates the lock information according to the packet format shown in Table 1. When the lock information is too much, the message is fragmented. After receiving the message, the takeover node performs grouping according to the sequence number of the message fragment. After the NFS lock information is successfully sent, the NFS lock information of the corresponding connection of the node is cleared until all the connections that need to be switched on the faulty node are sent.

Message type Socket link information Packet fragment number Lock 1 Lock 2 ... Lock n End mark

Table 1

After the receiving node receives the switching packet of the faulty node, it will perform corresponding processing according to the packet type. When the packet information of the lock transition is received, the packet end information is marked according to the end of the packet and the file lock information needs to be grouped, and it is determined that if the currently connected port switching is successful, the migrated file is parsed on the node. Lock the information, and initiate the recovery operation of the lock, and restore the file lock on the faulty node to the takeover node; if the currently connected port switchover is not completed, it will wait, when the maximum delay time of the client lock operation is exceeded (NFSv4 protocol) When the default 90s is the lease time, the migrated lock information is released. Because the recovery is lost at the client and the recovery value is lost, the port switching speed is often much faster than the lock migration speed. Only in rare cases occurs when lock migration is ahead of port switching and the delay exceeds the client's default maximum latency.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

Referring to FIG. 5, FIG. 5 is a schematic flowchart of a fifth embodiment of a method for data switching.

In the fourth embodiment, after step 402, the method further includes:

Step 403: The NFS module of the second device sends a message that the switching is completed to the TCP module of the second device.

The protocol stack IP layer of the second device opens and receives the packet of the NET, and the TCP module of the second device sends a window recovery packet to the first device, and the second NFS module sends and receives the packet.

After the TCP switchover of the faulty end is completed, the NFS service is sent to the NFS service. The NFS receives the message. If the NFS is also switched, the NFS actively closes the request. If the NFS has not completed the data switchover, the NFS switchover is performed. Close the connection when done.

The embodiment of the present invention receives the transmission control protocol TCP switching request by the first device. Transmitting, to the second device, the first information, where the first information is used by the second device to copy the socket connection of the first device, and the second device is configured to send the second information and the third information, where The second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to assign NFS related information of the first device to a socket connection corresponding to the socket ID, thereby When the problem occurs in the faulty node, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring fast recovery of the file lock.

Referring to FIG. 6, FIG. 6 is a schematic flowchart of a sixth embodiment of a method for data switching.

In the fourth or fifth embodiment, after step 401, the method further includes:

Step 404: The second device sends a zero window message to the first device.

After the takeover node establishes a new socket connection, the takeover node sends a zero window message to the faulty node, and temporarily disables the corresponding NET to receive the message at the network layer IP layer.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

Referring to FIG. 7, FIG. 7 is a schematic diagram of an interaction process of a seventh embodiment of a method for data switching.

In the seventh embodiment, the data switching method includes:

Step 701: After the TCP layer of the faulty node's protocol stack obtains the takeover node, notify the NFS to start switching.

Step 702: The TCP module sends a key information (key) such as a TCP layer protocol control block to the takeover node by using a cross-node cluster channel, and establishes a socket connection at the takeover node;

Step 703, the TCP module and the NFS module of the faulty node simultaneously send respective data to the takeover node through the cluster channel;

Step 704, the data transmission of the TCP module of the faulty node is completed, the faulty node notifies the NFS to close the socket, and the takeover node notifies the NFS new connection (socket) to arrive;

In step 705, the NFS module of the takeover end acquires a new socket connection, and notifies that the TCP starts to work normally.

The embodiment of the invention provides a technology for quickly completing the lock migration in the NFS switching and switching process in the cluster mode. The principle is that the NFS connection of the server is copied to the takeover end by copying the NFS connection, and the client does not perceive. The copied information is divided into two parts according to the category: control information and data information. The information of the control plane mainly copies the key field, and the data plane information needs to copy all the uncompleted data request messages to the takeover end; The key information of the control plane includes the NFS lock information, and the processing of the NFS lock can avoid conflicting access to the file. Collect NFS lock information on the fault side and recover the information on the takeover side.

The following technical solutions are adopted in the embodiment of the present invention: when the front-end IP migration is completed, the faulty end notifies the local end of the TCP layer switching, and the TCP notifies the application layer to switch, which is equivalent to migrating the TCP connection of the server to the take-over end; There is a cluster channel as a switched data channel, which can make various forms of physical channels, and each node controller can access the shared storage pool of the back end. After receiving the switching message, the application layer transfers the NFS file lock and NFS request data and other related information to the switched node through the internal communication of the cluster, and quickly recovers the lock information on the takeover node. The whole process is divided into TCP layer and NFS data migration. In order to improve the speed of data migration, TCP and NFS data migration are performed in parallel, and the cluster channel is a high-speed and reliable channel, so the data migration speed is fast, and the whole migration process is at the millisecond level.

The embodiments of the present invention are applicable to the high availability of the distributed network file system NFS in the cluster mode, and the applied technologies include reliable access of application services and reliable access of the distributed network file system in the cluster mode. It mainly solves the problem of ensuring the reliability and stability of the NFS service in the case of a single-point link failure in a big data storage cluster environment. The specific solution is: when a link fault occurs between the client and the access node, the node sends the file lock and related information to another takeover node through the network port-based NFS switch, and clones and copies a new one on the takeover node. Connection, copy the old NFS connection data to the new connection, the new connection has all the attributes and status of the old connection, including the information of the NFS lock, so that from the client's point of view, the connection of the server is changed, the client The file lock migration is completed without being aware of the server-side NFS service switch. Through this technology, a master multi-standby can be implemented. That is, by adding a port of each node in the cluster as an alternate port of the access node port, the lock information on the access node can be migrated to any node in the cluster.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting to send the second information and the third information to the second device, the second The information is used by the second device to generate a socket ID of the socket connection, where the third information is used to assign NFS related information of the first device to a socket connection corresponding to the socket ID, thereby implementing a fault. When there is a problem with the node, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thus ensuring fast recovery of the file lock.

The present invention provides an apparatus.

Referring to FIG. 8, FIG. 8 is a schematic diagram of functional modules of a first embodiment of a first device of the present invention.

In the first embodiment, the first device includes:

a first TCP module 801 and a first NFS module 802;

The first TCP module 801 is configured to: when the first device receives the transmission control protocol TCP switching request, send the first information to the second device, where the first information is used by the second device The socket connection of the first device;

Optionally, the first TCP module 801 includes:

The first obtaining unit 8011 is configured to: when the first device receives the TCP switching request, acquire first information, where the first information includes at least a key of a control block structure in the first device TCP module. The word and the keyword of the management structure in the first device socket module;

The first sending unit 8012 is configured to: send the first information to the second TCP module.

The first device may be a faulty node, and the second device may be a takeover node.

Specifically, if the faulty node is faulty or the faulty node is upgraded, the user can manually migrate the IP to the takeover node, and then power off the faulty node. After the IP address of the faulty node is migrated, the TCP module of the faulty end acquires the current handover IP address. The TCP module of the fault end acquires the control block structure of the TCP module and the key field in the management structure of the socket module, and sends the control block structure of the obtained TCP module and the key field in the management structure of the socket module through the cluster channel. Go to the TCP module of the takeover node so that the TCP module of the takeover node can clone a new socket connection in the first time.

The first TCP module 801 is configured to: send second information to the second device, where the second information is used by the second device to generate a socket ID of the socket connection;

The first NFS module 802 is configured to: send third information to the second device, where the third information is used to assign NFS related information of the first device to a socket connection corresponding to the socket ID. in.

Optionally, the first TCP module 801 includes:

The second obtaining unit 8013 is configured to: acquire the second information, where the second information includes at least the packet in the TCP module sending buffer and the packet in the socket receiving buffer;

The second sending unit 8014 is configured to: send the second information to the second TCP module by using a cluster channel, so that the second TCP module assigns the second information to the copied socket connection.

Optionally, the first NFS module 802 includes:

The third obtaining unit 8021 is configured to: acquire the third information, where the third information includes at least information about a control plane management structure of the NFS module and an IO request that is not completed by the NFS module, where the NFS module The information of the control plane management structure includes NFS lock information;

The third sending unit 8022 is configured to: send the third information to the second NFS module by using a cluster channel.

The TCP layer of the faulty end notifies the NFS service layer that the NFS service and the TCP layer start the switching of the connection at the same time. NFS receives the switch message, NFS stops the IO operation of the back-end disk and the packet of the front-end disk; NFS collects the information of the switch connection, including the NFS control plane management structure and the unfinished IO request sent to the pair through the cluster channel. On the end node, the NFS control plane management structure includes NFS lock information. At the same time, the TCP layer also collects the data packets in the TCP transmission buffer and the packets in the socket receiving ring, and sends them to the TCP layer of the takeover node through the cluster channel. The TCP layer of the takeover node needs to assign the information to the new socket connection. Complete the migration of the TCP connection. Both NFS and TCP are performed simultaneously here.

The TCP module of the takeover node restores the received switch data to the new socket connection. The TCP module of the takeover node processes all the data and sends the new socket ID to the NFS module of the takeover node. The NFS module that takes over the node needs to temporarily store the received switching data and then restore the NFS lock information. The recovery process of the NFS lock is to match the temporarily stored switching data with the new socket ID, and the matching condition is the quintuple information (protocol, source-destination IP, source-destination port) of the TCP layer and the new socket ID. The corresponding socket connections match.

Optionally, the first NFS module 802 includes:

The fourth obtaining unit 8023 is configured to: obtain NFS lock information;

The encapsulating unit 8024 is configured to: encapsulate the NFS lock information according to a packet format, where the packet format includes at least the socket connection information, a packet fragment number, a number of the NFS lock information, and the End of the message;

The fourth sending unit 8025 is configured to: send the encapsulated NFS lock information to the second NFS module.

In order to implement the fast recovery of the NFS lock information, when the system detects that the faulty node access is abnormal, the port of the faulty node is triggered to be switched. After the target node of the switchover is determined, the NFS lock information migration is started.

After the faulty node receives the NFS lock information migration message, it first finds the lock of the corresponding sockett information according to the socket information, and encapsulates the lock information according to the packet format shown in Table 1. When the lock information is too much, the message is fragmented. After receiving the message, the takeover node performs group packet processing according to the sequence number of the message fragment. After the NFS lock information is successfully sent, the NFS lock information of the corresponding connection of the node is cleared until all the connections that need to be switched on the faulty node are sent.

Message type Socket link information Packet fragment number Lock 1 Lock 2 ... Lock n End mark

Table 1

After the receiving node receives the switching packet of the faulty node, it will perform corresponding processing according to the packet type. When the packet information of the lock transition is received, the packet end information is marked according to the end of the packet and the file lock information needs to be grouped, and it is determined that if the currently connected port switching is successful, the migrated file is parsed on the node. Lock the information, and initiate the recovery operation of the lock, and restore the file lock on the faulty node to the takeover node; if the currently connected port switchover is not completed, it will wait, when the maximum delay time of the client lock operation is exceeded (NFSv4 protocol) When the default 90s is the lease time, the migrated lock information is released. Because the recovery is lost at the client and the recovery value is lost, the port switching speed is often much faster than the lock migration speed. Only in rare cases occurs when lock migration is ahead of port switching and the delay exceeds the client's default maximum latency.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

The invention further provides a first device.

Referring to FIG. 9, FIG. 9 is a schematic diagram of a functional module of a second embodiment of a first device of the present invention.

On the basis of the first embodiment, the first TCP module 801 further includes:

The first receiving unit 8015 is configured to: receive a zero window message sent by the second device.

After the takeover node establishes a new socket connection, the takeover node sends a zero window message to the faulty node, and temporarily disables the corresponding NET to receive the message at the network layer IP layer.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

Referring to FIG. 10, FIG. 10 is a schematic diagram of a functional module of a third embodiment of a first device of the present invention.

On the basis of the first embodiment, the first TCP module 801 further includes:

The fifth sending unit 8016 is configured to: after sending the second information to the second device, send a message that the switching is completed to the first NFS module;

The first NFS module 802 further includes:

The closing unit 8026 is configured to: if the first NFS module receives the message that the first NFS module sends a switching completion, and the first NFS module has sent the third information to the second device, Turning off the TCP switching request;

The closing unit 8026 is further configured to: if the first NFS module receives the message that the first NFS module sends a switching completion, and the first NFS module does not send the third information to the second device And closing the TCP switching request after sending the third information to the second device.

After the TCP switchover of the faulty end is completed, the NFS service is sent to the NFS service. The NFS receives the message. If the NFS is also switched, the NFS actively closes the request. If the NFS has not completed the data switchover, the NFS switchover is performed. Close the connection when done.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting to send the second information and the third information to the second device, the second The information is used by the second device to generate a socket ID of the socket connection, where the third information is used to assign NFS related information of the first device to a socket connection corresponding to the socket ID, thereby implementing a fault. When there is a problem with the node, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thus ensuring fast recovery of the file lock.

Referring to FIG. 11, FIG. 11 is a schematic diagram of functional modules of a first embodiment of a second device of the present invention.

In the first embodiment, the second device includes a second TCP module 1101 and a second NFS module 1102;

The second TCP module 1101 is configured to: when the first device receives the TCP switching request, receive the first information sent by the first device, and copy the first device according to the first information. Socket connection

The first device may be a faulty node, and the second device may be a takeover node.

If the faulty node is faulty or the faulty node is upgraded, the user can manually migrate the IP to the takeover node, and then power off the faulty node. After the IP address of the faulty node is migrated, the TCP module of the faulty end obtains the connection on the current switched IP. The faulty TCP module acquires the control block structure of the TCP module and the key field in the management structure of the socket module, and sends the obtained control block structure of the TCP module and the key field in the management structure of the socket module to the cluster channel through the cluster channel. Take over the node's TCP module so that the TCP module that takes over the node can clone a new socket connection in the first time.

The second TCP module 1101 is configured to: receive second information sent by the first device, and generate a socket ID of the socket connection according to the second information;

Optionally, the second TCP module 1101 includes:

The second receiving unit 11011 is configured to: receive the second information;

The generating unit 11012 is configured to: assign the second information to the copied socket connection, and generate a socket ID of the socket connection;

The second NFS module 1102 is configured to: receive the third information sent by the first device, and assign the NFS related information of the first device to the socket connection corresponding to the socket ID according to the third information.

Optionally, the second NFS module 1102 includes:

The third receiving unit 11021 is configured to: receive the third information;

The matching unit 11022 is configured to: match the socket connection corresponding to the socket ID according to the quintuple information in the third information;

The assignment unit 11023 is configured to: if matched, assign the third information to the socket connection corresponding to the socket ID.

The TCP layer of the faulty end notifies the NFS service layer that the NFS service and the TCP layer start the switching of the connection at the same time. NFS receives the switch message, NFS stops the IO operation of the back-end disk and the packet of the front-end disk; NFS collects the information of the switch connection, including the NFS control plane management structure and the unfinished IO request sent to the pair through the cluster channel. On the end node, the NFS control plane management structure includes NFS lock information. At the same time, the TCP layer also collects the data packets in the TCP transmission buffer and the packets in the socket receiving ring, and sends them to the TCP layer of the takeover node through the cluster channel. The TCP layer of the takeover node needs to assign the information to the new socket connection. Complete the migration of the TCP connection. Both NFS and TCP are performed simultaneously here.

The TCP module of the takeover node restores the received switch data to the new socket connection. The TCP module of the takeover node processes all the data and sends the new socket ID to the NFS module of the takeover node. The NFS module that takes over the node needs to temporarily store the received switching data and then restore the NFS lock information. The recovery process of the NFS lock is to match the temporarily stored switching data with the new socket ID, and the matching condition is the quintuple information (protocol, source-destination IP, source-destination port) of the TCP layer and the new socket ID. The corresponding socket connections match.

In order to implement the fast recovery of the NFS lock information, the system detects the faulty node access abnormality and triggers the port switching of the faulty node. After the target node of the switchover is determined, the NFS lock information migration begins.

After the faulty node receives the NFS lock information migration message, it first finds the lock of the corresponding sockett information according to the socket information, and encapsulates the lock information according to the packet format shown in Table 1. When the lock information is too much, the message is fragmented. After receiving the message, the takeover node performs group packet processing according to the sequence number of the message fragment. After the NFS lock information is successfully sent, the NFS lock information of the corresponding connection of the node is cleared until all the connections that need to be switched on the faulty node are sent.

Message type Socket link information Packet fragment number Lock 1 Lock 2 ... Lock n End mark

Table 1

After the receiving node receives the switching packet of the faulty node, it will perform corresponding processing according to the packet type. When the packet information of the lock transition is received, the packet end information is marked according to the end of the packet and the file lock information needs to be grouped, and it is determined that if the currently connected port switching is successful, the migrated file is parsed on the node. Lock the information and actively initiate a lock recovery operation to lock the file on the failed node Resume on the takeover node; if the currently connected port switchover is not completed, it will wait. When the maximum delay time of the client lock operation is exceeded (the NFSv4 protocol defaults to 90s as the lease time), the migrated lock information is released because this Even if the recovery, the lock has expired on the client, losing the value of recovery, but the port switching speed is often much faster than the lock migration speed, only in rare cases, the lock migration is ahead of the port switching and the delay exceeds the customer. The default maximum delay case.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

Referring to FIG. 12, FIG. 12 is a schematic diagram of functional modules of a second embodiment of a second device of the present invention.

In the first embodiment, the second NFS module 1102 further includes:

The sixth sending unit 11024 is configured to: send a message that the switching is completed to the second TCP module;

The second TCP module 1101 further includes:

The seventh sending unit 11013 is configured to: send a window recovery message to the first device.

After the TCP switchover of the faulty end is completed, the NFS service is sent to the NFS service. The NFS receives the message. If the NFS is also switched, the NFS actively closes the request. If the NFS has not completed the data switchover, the NFS switchover is performed. Close the connection when done.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

Referring to FIG. 13, FIG. 13 is a schematic diagram of functional modules of a third embodiment of a second device of the present invention.

In the first or second embodiment, the second TCP module 1101 further includes:

The eighth sending unit 11014 is configured to: send a zero window message to the first device.

After the takeover node establishes a new socket connection, the takeover node sends a zero window message to the faulty node, and temporarily disables the corresponding NET to receive the message at the network layer IP layer.

In addition, in the actual device, the module functions of the first device 800 as shown in FIGS. 8 to 10 and the module functions of the second device 110 as shown in FIGS. 11 to 13 can be simultaneously collected.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

The invention further provides a system.

Referring to FIG. 14, FIG. 14 is a schematic structural diagram of a system according to a first embodiment of the system of the present invention. The system includes a first device 800 as shown in Figures 8-10 and a second device 110 as shown in Figures 11-13.

In the embodiment of the present invention, when the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket of the first device. Connecting, sending, to the second device, second information and third information, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used to be the first The NFS-related information of the device is assigned to the socket connection corresponding to the socket ID, so that when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring the file lock. Fast recovery.

The above is only an alternative embodiment of the present invention, and thus does not limit the scope of the invention, and the equivalent structure or equivalent process transformation made by using the specification and the drawings of the present invention, or directly or indirectly applied to other related technologies. The fields are all included in the scope of patent protection of the present invention.

One of ordinary skill in the art will appreciate that all or a portion of the steps of the above-described embodiments can be implemented using a computer program flow, which can be stored in a computer readable storage medium, such as on a corresponding hardware platform (eg, The system, device, device, device, etc. are executed, and when executed, include one or a combination of the steps of the method embodiments.

Alternatively, all or part of the steps of the above embodiments may also be implemented by using an integrated circuit. These steps may be separately fabricated into individual integrated circuit modules, or multiple modules or steps may be fabricated into a single integrated circuit module. achieve. Thus, the invention is not limited to any specific combination of hardware and software.

The devices/function modules/functional units in the above embodiments may be implemented by a general-purpose computing device, which may be centralized on a single computing device or distributed over a network of multiple computing devices.

When each device/function module/functional unit in the above embodiment is implemented in the form of a software function module and sold or used as a stand-alone product, it can be stored in a computer readable storage medium. The above mentioned computer readable storage medium may be a read only memory, a magnetic disk or an optical disk or the like.

Industrial applicability

In the embodiment of the present invention, when the faulty node has a problem, the lock information of the faulty node and its corresponding data connection can be migrated to another normal node, thereby ensuring fast recovery of the file lock.

Claims (24)

A method of data switching, including: When the first device receives the transmission control protocol TCP switching request, the first information is sent to the second device, where the first information is used by the second device to copy the socket connection of the first device; Transmitting, by the first device, the second information and the third information to the second device, where the second information is used by the second device to generate a socket ID of the socket connection, where the third information is used for The network file system NFS related information of the first device is assigned to the socket connection corresponding to the socket ID. The method of claim 1, wherein the transmitting the first information to the second device, in the case that the first device receives the TCP switching request, includes: When the first device receives the TCP switching request, the TCP module of the first device acquires the first information, where the first information includes at least a keyword of a control block structure in the first device TCP module. And a keyword of the management structure in the first device socket module; The TCP module of the first device sends the first information to a TCP module of the second device. The method of claim 1, wherein the transmitting the second information to the second device comprises: The TCP module of the first device acquires the second information, where the second information includes at least the packet in the TCP module sending buffer and the packet in the socket receiving buffer. The TCP module of the first device sends the second information to the TCP module of the second device through a cluster channel, so that the TCP module of the second device assigns the second information to the copied socket connecting. The method of claim 1, wherein the transmitting the third information to the second device comprises: The NFS module of the first device acquires the third information, where the third information includes at least information about a control plane management structure of the NFS module and an input and output IO that is not completed by the NFS module. Requesting, the information of the control plane management structure of the NFS module includes NFS lock information; The NFS module of the first device sends the third information to the NFS module of the second device by using a cluster channel. The method of claim 4, wherein the NFS module of the first device acquires the third information, including: The NFS module of the first device acquires NFS lock information; The NFS module of the first device sends the third information to the NFS module of the second device by using a cluster channel, including: The NFS module of the first device encapsulates the NFS lock information according to a packet format, where the packet format includes at least the information of the socket connection, the packet fragment number, the number of the NFS lock information, and the The end identifier of the message; The NFS module of the first device sends the encapsulated NFS lock information to the second NFS module. The method according to any one of claims 1 to 5, wherein after the sending the first information to the second device, the method further includes: The first device receives a zero window message sent by the second device. The method according to any one of claims 1 to 5, wherein after the sending the second information and the third information to the second device, the method further includes: After the TCP module of the first device sends the second information to the second device, the TCP module sends a message that the switching is completed to the NFS module of the first device. If the NFS module of the first device receives the message that the NFS module of the first device sends a switching completion, and the NFS module of the first device has sent the third information to the second device, The NFS module of the first device closes the TCP switching request; If the NFS module of the first device receives the message that the first NFS module sends a switching completion, and the NFS module of the first device does not send the third information to the second device, The NFS module of the first device closes the TCP switching request after sending the third information to the second device. A method of data switching, including: Receiving, by the second device, the first information that is sent to the second device when the first device receives the transmission control protocol TCP switching request, and copies the socket connection of the first device according to the first information; Receiving, by the second device, the second information and the third information that are sent by the first device, generating a socket ID of the socket connection according to the second information, and using the third information, the network file system of the first device The NFS related information is assigned to the socket connection corresponding to the socket ID. The method of claim 8, wherein the second device receives the second information and the third information sent by the first device, and generates the socket ID of the socket connection according to the second information, including: The TCP module of the second device receives the second information; The TCP module of the second device assigns the second information to the copied socket connection, and generates a socket ID of the socket connection; The NFS module of the second device receives the third information; The NFS module of the second device matches the socket connection corresponding to the socket ID according to the quintuple information in the third information, and if yes, assigns the third information to the socket ID Corresponding socket connection. The method of claim 9, wherein the NFS module of the second device matches the socket connection corresponding to the socket ID according to the quintuple information in the third information, and if yes, the method After the three information is assigned to the socket connection corresponding to the socket ID, the information includes: The NFS module of the second device sends a message that the switching is completed to the second TCP module. The protocol stack IP layer of the second device opens and receives the packet of the NET, and the TCP module of the second device sends a window recovery packet to the first device, and the NFS module of the second device sends and receives the packet. The method according to any one of claims 8 to 10, wherein after the receiving, by the second device, the first information sent by the first device, the method further includes: Sending a zero window message to the first device. A first device includes a first transmission control protocol TCP module and a first network file system NFS module; The first TCP module is configured to: when the first device receives the transmission control protocol TCP switching request, send the first information to the second device, where the first information is used by the second device to be copied. a socket connection of the first device; The first TCP module is further configured to: send second information to the second device, where the second information is used by the second device to generate a socket ID of the socket connection; The first NFS module is configured to: send third information to the second device, where the third information is used to assign NFS related information of the first device to a socket connection corresponding to the socket ID. . The first device of claim 12, wherein the first TCP module comprises: a first acquiring unit, configured to: when the first device receives the TCP switching request, acquire first information, where the first information includes at least a keyword of a control block structure in the first device TCP module And a keyword of the management structure in the first device socket module; The first sending unit is configured to: send the first information to the second TCP module. The first device of claim 12, wherein the first TCP module comprises: a second acquiring unit, configured to: acquire the second information, where the second information includes at least a packet in the TCP module sending buffer and a packet in the socket receiving buffer; The second sending unit is configured to: send the second information to the second TCP module by using a cluster channel, so that the second TCP module assigns the second information to the copied socket connection. The first device of claim 12, wherein the first NFS module comprises: The third obtaining unit is configured to: acquire the third information, where the third information includes at least information about a control plane management structure of the NFS module and an input/output IO request that is not completed by the NFS module, where the NFS module The information of the control plane management structure includes NFS lock information; The third sending unit is configured to: send the third information to the second NFS module by using a cluster channel. The first device of claim 15, wherein the first NFS module comprises: The fourth obtaining unit is configured to: obtain NFS lock information; The encapsulating unit is configured to: encapsulate the NFS lock information according to a packet format, where the packet format includes at least the socket connection information, a packet fragment number, a number of the NFS lock information, and the packet The end of the text; and The fourth sending unit is configured to: send the encapsulated NFS lock information to the second NFS module. The first device according to any one of claims 12 to 16, wherein the first TCP module further comprises: The first receiving unit is configured to: receive a zero window message sent by the second device. The first device according to any one of claims 12 to 16, wherein the first TCP module further comprises: The fifth sending unit is configured to: after sending the second information to the second device, send a message that the switching is completed to the first NFS module; The first NFS module further includes: And the closing unit is configured to: if the first NFS module receives the message that the first NFS module sends a switching completion, and the first NFS module has sent the third information to the second device, Close the TCP switching request; The shutting down unit is further configured to: if the first NFS module receives the message that the first NFS module sends a switch completion, and the first NFS module does not send the third message to the second device The information is closed after the third information is sent to the second device. begging. A second device includes a second transmission control protocol TCP module and a second network file system NFS module; The second TCP module is configured to: when the first device receives the TCP switching request, receive the first information sent by the first device, and copy the socket of the first device according to the first information connection; The second TCP module is configured to: receive second information sent by the first device, and generate a socket ID of the socket connection according to the second information; The second NFS module is configured to receive the third information sent by the first device, and assign the NFS related information of the first device to the socket connection corresponding to the socket ID according to the third information. A second device according to claim 19, wherein The second TCP module includes: a second receiving unit, configured to: receive the second information; a generating unit, configured to: assign the second information to the copied socket connection, and generate a socket ID of the socket connection; The second NFS module includes: a third receiving unit, configured to: receive the third information; a matching unit, configured to: match a socket connection corresponding to the socket ID according to the quintuple information in the third information; The assignment unit is configured to: if matched, assign the third information to the socket connection corresponding to the socket ID. The second device of claim 20, wherein the second NFS module further comprises: a sixth sending unit, configured to: send a message that the switching is completed to the second TCP module; The second TCP module further includes: The seventh sending unit is configured to: send a window recovery message to the first device. The second device according to claim 19, wherein the second TCP module further comprises: The eighth sending unit is configured to: send a zero window message to the first device. A system comprising the first device of claims 12 to 18 and the second device of claims 19 to 22. A computer readable storage medium storing program instructions that, when executed, implement the method of any of claims 1-11.

Images