Dual-machine high-availability system

Abstract

The present invention discloses a dual-machine high-availability system comprising a gateway, a master server, and a slave server. The master server and the slave server are both connected to the gateway via a service line. The master server and the slave server also sense the status of the peer server via a heartbeat line. When either server senses a heartbeat line failure with the peer server, it configures a floating IP address on the corresponding server according to pre-set configuration information and preset configuration rules when the peer server's network status is normal, as detected by the service line. This system can resolve the split-brain problem in master-slave dual-machine applications and achieve high availability of the master-slave dual-machine architecture.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a dual-machine high-availability system. Background Art

[0002] In the field of power dispatching, dispatching equipment has very high reliability requirements. No single device failure point is allowed at the software and hardware levels, so important servers must all be in dual-machine or multi-machine mode. When one device fails, the other device must be able to take over the business.

[0003] The existing technology deploys two machines, each with a complete set of server software. A heartbeat line is used between the two devices to determine the other's survival. If one machine fails, the other binds to the service IP address to handle subsequent requests. However, if the heartbeat line between the two devices is disconnected, they lose contact and become two independent entities. Each machine assumes the other has failed and binds to the service IP address, resulting in a split-brain phenomenon. Summary of the Invention

[0004] To solve the above problems existing in the prior art, the present invention provides a dual-machine high-availability system that can solve the brain split problem in master-slave dual-machine applications and achieve high availability of master-slave dual-machines.

[0005] The present invention provides a dual-machine high-availability system, the system comprising a gateway, a master server and a slave server;

[0006] The master server and the slave server are both connected to the gateway via a service line;

[0007] The master server and the slave server also sense the status of the peer server through a heartbeat line;

[0008] When any server detects a heartbeat line failure with the peer server, based on pre-set configuration information, when the business line detects that the network status of the peer server is normal, the floating IP is configured on the corresponding server according to the preset configuration rules.

[0009] Preferably, the configuration rule is specifically to stop the slave server and bind the floating IP to the master server.

[0010] As a preferred solution, the configuration rules are specifically as follows:

[0011] When it is determined that the floating IP does not exist, binding the floating IP to the primary server;

[0012] If a floating IP address exists, the floating IP address configuration is retained.

[0013] Preferably, the configuration information includes a gateway IP, a peer service port IP, and a peer heartbeat port IP.

[0014] Preferably, any of the servers continuously obtains the network status of the opposite server by pinging at preset intervals.

[0015] As a preferred solution, any of the servers monitors the designated ports of the gateway and the opposite server through TCP SYN scanning technology to sense the network status of the opposite server.

[0016] Preferably, when any of the servers senses that the server, service line and heartbeat line of the system are normal, the KA program is started, and the KA program negotiates and binds the floating IP.

[0017] Preferably, when any of the servers senses that the opposite server is abnormal, it starts a KA program, and the KA program binds the floating IP to the server.

[0018] As a preferred solution, the manner in which any server performs network status perception is specifically as follows:

[0019] Send a data packet to the gateway through the business line of the server; and judge whether the business line of the server is normal based on the feedback data packet of the gateway; after determining that the business line of the server is normal, send a data packet to the opposite server through the business line of the server, the gateway, and the business line of the opposite server, and judge whether the business line of the opposite server is normal based on the feedback data packet of the opposite server.

[0020] An embodiment of the present invention further provides a dual-machine high-availability system, the system comprising: a gateway, a master server, and a slave server;

[0021] The master server and the slave server are both connected to the gateway via a service line;

[0022] The master server and the slave server both perform heartbeat sensing through the service line between the gateways to determine the status of the peer server;

[0023] When any server cannot determine the peer server through heartbeat perception, it uses the business line to perceive the network status according to the pre-set configuration information; when it is found that the business line of the peer server has failed, the floating IP is configured on this server.

[0024] Preferably, the configuration information includes a gateway IP, a peer service port IP, and a peer heartbeat port IP.

[0025] Preferably, any of the servers continuously obtains the network status of the opposite server by pinging at preset intervals.

[0026] As a preferred solution, any of the servers monitors the designated ports of the gateway and the opposite server through TCP SYN scanning technology to sense the network status of the opposite server.

[0027] Preferably, when any of the servers senses that the server, service line and heartbeat line of the system are normal, the KA program is started, and the KA program negotiates and binds the floating IP.

[0028] Preferably, when any of the servers senses that the opposite server is abnormal, it starts a KA program, and the KA program binds the floating IP to the server.

[0029] As a preferred solution, the manner in which any server performs network status perception is specifically as follows:

[0030] Send a data packet to the gateway through the business line of the server; and judge whether the business line of the server is normal based on the feedback data packet of the gateway; after determining that the business line of the server is normal, send a data packet to the opposite server through the business line of the server, the gateway, and the business line of the opposite server, and judge whether the business line of the opposite server is normal based on the feedback data packet of the opposite server.

[0031] The present invention provides a dual-machine high-availability system comprising a gateway, a master server, and a slave server. The master server and the slave server are both connected to the gateway via a service line. The master server and the slave server also sense the status of the peer server via a heartbeat line. When either server senses a heartbeat line failure with the peer server, it configures a floating IP address on the corresponding server according to pre-set configuration information and preset configuration rules when it senses the peer server's network status is normal via the service line. This system can resolve the split-brain problem in master-slave dual-machine applications and achieve high availability of the master-slave dual-machine architecture. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a schematic diagram of the structure of a dual-machine high-availability system provided by an embodiment of the present invention;

[0033] Figure 2 This is a structural diagram of a dual-machine high-availability system provided by another embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0035] An embodiment of the present invention provides a dual-machine high-availability system, the system comprising a gateway, a master server and a slave server;

[0036] The master server and the slave server are both connected to the gateway via a service line;

[0037] The master server and the slave server also sense the status of the peer server through a heartbeat line;

[0038] When any server detects a heartbeat line failure with the peer server, based on pre-set configuration information, when the business line detects that the network status of the peer server is normal, the floating IP is configured on the corresponding server according to the preset configuration rules.

[0039] When implementing this embodiment, see Figure 1 , is a schematic diagram of the structure of a dual-machine high-availability system provided by an embodiment of the present invention. The system includes a master server A, a slave server B, and a gateway to form a service side. When the service side communicates with a terminal, it needs to be configured with an IP address for processing the terminal's communication request;

[0040] Redundancy is achieved by designing a master server and a slave server. When a network failure occurs on one of the servers, the corresponding communication business can be handled by the other server.

[0041] In the system, the master server A is connected to the gateway via a first service line B1, and the slave server B is connected to the gateway via a second service line B2; the master server A and the slave server B are also connected via a heartbeat line H, and both the master server A and the slave server B sense the network status of the peer server through the heartbeat line;

[0042] The master server A, slave server B, and gateway together form a dual-machine system. A floating IP (business IP) is exposed to the outside world for business association. The floating IP is only bound to the network port connected to the gateway on the master server. The master server A can be configured as the nominal master device and the slave server B as the slave device. In actual use, the real master-slave switching will be implemented according to the policy. The heartbeat line H is used to provide heartbeat for the KA (keepalived) program. The KA program provides the function of binding the floating IP to a specified network card. The master server runs the KA1 program and the slave server runs the KA2 program. In addition, when both machine networks are normal, KA is relied on to achieve automatic master-slave switching. The two servers will have their own non-business IP. This is different from the floating IP. It does not provide business services to the outside world and is only used to provide background management connection functions.

[0043] From the perspective of operation and maintenance, the system will have 8 network states and corresponding descriptions, see Table 1;

[0044] Table 1 Network status and corresponding description

[0045]

[0046] In the existing technology, when the heartbeat line is normal, the allocation of floating IP addresses in master-slave dual-machine devices when network changes occur can be solved by controlling whether the KA program in the master and slave servers is running. In other words, the master-slave switch can be solved.

[0047] However, when both servers' business lines functioned normally and the heartbeat line connecting the two servers was disconnected, the previously unified, coordinated system split into two independent entities. Losing communication, each server assumed the other had failed. The KA software on both servers, like a "split-brain patient," bound to the business IP address, competed for "application services" simultaneously, resulting in a split-brain phenomenon.

[0048] To address the split-brain phenomenon, when any server in this system senses that the heartbeat line between it and the peer server has failed, it uses the business line to sense the network status based on the pre-set configuration information. This network sensing through the business line can determine the network status of the peer server.

[0049] When it is detected that the network status of the peer server is normal, that is, only the heartbeat line in the system is faulty and all business lines are normal, the floating IP is configured on the corresponding server according to the preset configuration rules;

[0050] It should be noted that any server in this embodiment can be both the master and slave servers in the system. For each server, if the heartbeat line fails, it loses awareness of the peer server's status. The business line then uses peer server status awareness to control whether the KA program is running, indirectly controlling which server the floating IP is bound to. This simply resolves the split-brain issue caused by using only the KA program.

[0051] In another embodiment provided by the present invention, the configuration rule is specifically to stop the slave server and bind the floating IP to the master server.

[0052] In the specific implementation of this embodiment, when only the heartbeat line H among the first service line B1, the second service line B2 and the heartbeat line H is disconnected, the specific method of configuring the floating IP using the preset configuration rules is as follows:

[0053] Stop the slave server according to the configuration and ensure that the floating IP is configured on the master server.

[0054] By forcibly configuring the floating IP on the master server and stopping the KA program on the slave server, a split-brain phenomenon between the two servers can be avoided.

[0055] In another embodiment provided by the present invention, the configuration rule is specifically:

[0056] When it is determined that the floating IP does not exist, binding the floating IP to the primary server;

[0057] If a floating IP address exists, the floating IP address configuration is retained.

[0058] In the specific implementation of this embodiment, it is determined whether the floating IP already exists;

[0059] If the floating IP already exists, it may exist on the master server or the slave server. In this case, the floating IP switching will not be performed.

[0060] However, if the floating IP does not exist at this time, it means that the floating IP has not been configured yet. In this case, the floating IP will be bound to the configuration master server first according to the configuration.

[0061] By forcibly assigning the floating IP to the primary server, you can avoid split-brain issues. However, this also has the disadvantage of potentially switching the floating IP. This means that if the floating IP is originally on a slave server, it will be forcibly transferred to the primary server. By determining whether the floating IP exists, you can reduce the switching of floating IPs and ensure stable communication.

[0062] In another embodiment of the present invention, the configuration information includes a gateway IP, a peer service port IP, and a peer heartbeat port IP.

[0063] During the specific implementation of this embodiment, when the server perceives the network status of the peer server through the service line, it needs to first obtain configuration information, namely the gateway IP, the peer service port IP and the peer heartbeat port IP in the system.

[0064] Each server can have two network ports connected to the gateway, and the servers are pre-configured with their own assigned IPs and gateway IPs when they are deployed. By obtaining the configuration information, the current status of the two servers can be perceived through the program on any server; that is, by obtaining the configuration information, the network status of the opposite server can be perceived through the business line.

[0065] In another embodiment of the present invention, any server continuously obtains the network status of the opposite server by pinging at preset time intervals.

[0066] In the specific implementation of this embodiment, the process of any server perceiving the network status of the opposite server through the service line according to the preset configuration information specifically includes:

[0067] The server uses the ping method to obtain the status of the peer server at specified intervals. The status of the peer server is determined by whether the peer server can be pinged successfully.

[0068] Master server A senses that the gateways can ping each other, but master server A cannot ping slave server B. Regardless of whether the heartbeat line is normal, slave server B is disconnected from the gateway network. Then the second business line B2 network failure occurs. At this time, the master server A network is normal, while the slave server B network is abnormal. At this time, you need to stop KA on slave server B to release the floating IP, and start KA on master server A to bind the floating IP.

[0069] Use ping to determine the network status of the peer server, determine the floating IP binding, and avoid split-brain phenomenon.

[0070] In another embodiment of the present invention, any of the servers monitors the designated ports of the gateway and the peer server through TCP SYN scanning technology to sense the network status of the peer server.

[0071] In the specific implementation of this embodiment, any server obtains the gateway IP, the opposite end service port IP and the opposite end heartbeat port IP according to the configuration, monitors the designated port of the opposite end server, and uses TCP SYN scanning technology to detect whether the other server network is normal;

[0072] In addition, the server can also determine the network status of the server and gateway by sensing the 80 port of the gateway's management page.

[0073] In another embodiment of the present invention, when any server senses that the server, service line and heartbeat line of the system are normal, the KA program is started, and the KA program negotiates and binds the floating IP.

[0074] Any server in the system senses that the first service line B1, the second service line B2, and the heartbeat line H are all normal: both servers can start the KA program, and the floating IP performs the negotiation according to the KA program. In this state, the split-brain phenomenon will not occur.

[0075] In another embodiment of the present invention, when any server senses that the opposite server is abnormal, it starts a KA program, and the KA program binds the floating IP to the server.

[0076] In the specific implementation of this embodiment, when the master server A detects that the second service line B2 of the slave server B is abnormal, the master server A starts the KA program to bind the floating IP to the master server A. When the slave server B detects that the second service line B2 of the server itself has a network failure, the slave server B stops the KA2 program to release the floating IP.

[0077] When slave server B detects an abnormality in the primary service line B1 of master server A, it starts the KA program to bind the floating IP address to slave server B. When master server A detects a network failure in the primary service line B1 of its own server, it stops the KA1 program to release the floating IP address.

[0078] Each server automatically decides to start or stop the KA program by sensing the network status of the business line, avoiding the brain split caused by two independent servers starting the KA program at the same time.

[0079] In another embodiment of the present invention, the manner in which any server performs network status perception is specifically as follows:

[0080] Send a data packet to the gateway through the business line of the server; and judge whether the business line of the server is normal based on the feedback data packet of the gateway; after determining that the business line of the server is normal, send a data packet to the opposite server through the business line of the server, the gateway, and the business line of the opposite server, and judge whether the business line of the opposite server is normal based on the feedback data packet of the opposite server.

[0081] In the specific implementation of this embodiment, the specific method of network perception is described by taking the main server A as an example:

[0082] Send a data packet to the gateway through the first business line B1 of the main server A, and determine whether the first business line B1 is normal based on the feedback data packet from the gateway;

[0083] When no feedback is received from the gateway, it is determined that the first business line B1 is abnormal;

[0084] When receiving the feedback data packet corresponding to the gateway feedback, it is determined that the first business line B1 is normal, and a data packet is sent to the slave server B through the first business line B1, the gateway, and the second business line B2;

[0085] When no feedback is received from server B, the second business line B2 is determined to be abnormal;

[0086] After receiving the corresponding feedback data packet from server B, it is determined that the second business line B2 is normal.

[0087] An embodiment of the present invention further provides a dual-machine high-availability system, the system comprising: a gateway, a master server, and a slave server;

[0088] The master server and the slave server are both connected to the gateway via a service line;

[0089] The master server and the slave server both perform heartbeat sensing through the service line between the gateways to determine the status of the peer server;

[0090] When any server cannot determine the peer server through heartbeat perception, it uses the business line to perceive the network status according to the pre-set configuration information; when it is found that the business line of the peer server has failed, the floating IP is configured on this server.

[0091] When implementing this embodiment, see Figure 2 , is a dual-machine high-availability system provided by another embodiment of the present invention. The system includes a master server A, a slave server B, and a gateway to form a service side. When the service side communicates with a terminal, it needs to configure an IP address for processing the terminal's communication request;

[0092] Redundancy is achieved by designing a master server and a slave server. When a network failure occurs on one of the servers, the corresponding communication business can be handled by the other server.

[0093] In the system, the master server A is connected to the gateway via the first service line B1, and the slave server B is connected to the gateway via the second service line B2; the master server A and the slave server B also perform heartbeat sensing via the service line, and both the master server A and the slave server B perceive the network status of the peer server through the heartbeat line;

[0094] The master server A, slave server B, and gateway together form a dual-machine system. A floating IP (business IP) is exposed to the outside world for business association. The floating IP is only bound to the network port connected to the gateway on the master server. The master server A can be configured as the nominal master device and the slave server B as the slave device. In actual use, the real master-slave switching will be implemented according to the policy. The heartbeat line H is used to provide heartbeat for the KA (keepalived) program. The KA program provides the function of binding the floating IP to a specified network card. The master server runs the KA1 program and the slave server runs the KA2 program. In addition, when both machine networks are normal, KA is relied on to achieve automatic master-slave switching. The two servers will have their own non-business IP. This is different from the floating IP. It does not provide business services to the outside world and is only used to provide background management connection functions.

[0095] From the perspective of primary server A, we can see the following three situations in the entire dual-host network:

[0096] The first service line B1, the second service line B2, and the heartbeat line H are all normal: both servers can start the KA program, and the floating IP performs the negotiation based on the KA program. In this state, the split-brain phenomenon will not occur.

[0097] The first business line B1 is normal, and the second business line B2 is abnormal: In this state, regardless of whether the heartbeat line is normal, if the business line of the slave server B is abnormal, the master server A starts the KA program to bind the floating IP.

[0098] The first service line B1 is abnormal: In this state, the main server A does not need to know the network status of other devices. The first service line B1 of the main server A is abnormal, and the main server A stops the KA program to release the floating IP.

[0099] In this system structure, the master server A and the slave server B also perform heartbeat perception through the business line. When the heartbeat line fails, at least one of the first business line B1 and the second business line B2 is abnormal. According to the pre-set configuration information, the business line is used to perceive the network status; when it is obtained that the business line of the opposite server has failed, the floating IP is configured on this server.

[0100] When the service line of this server is detected to be faulty, stop the KA program to release the floating IP.

[0101] This application considers the dual-machine network environment as the starting point for analysis, and can fundamentally solve the brain split problem caused by abnormal heartbeat lines in master-slave dual machines. It does not require intrusion into the specific business of the server and can be applied to any master-slave dual-machine environment with strong versatility.

[0102] In another embodiment of the present invention, the configuration information includes a gateway IP, a peer service port IP, and a peer heartbeat port IP.

[0103] During the specific implementation of this embodiment, when the server perceives the network status of the peer server through the service line, it needs to first obtain configuration information, namely the gateway IP, the peer service port IP and the peer heartbeat port IP in the system.

[0104] Each server can have two network ports connected to the gateway, and the servers are pre-configured with their own assigned IPs and gateway IPs when they are deployed. By obtaining the configuration information, the current status of the two servers can be perceived through the program on any server; that is, by obtaining the configuration information, the network status of the opposite server can be perceived through the business line.

[0105] In another embodiment of the present invention, any server continuously obtains the network status of the opposite server by pinging at preset time intervals.

[0106] In the specific implementation of this embodiment, the process of any server perceiving the network status of the opposite server through the service line according to the preset configuration information specifically includes:

[0107] The server uses the ping method to obtain the status of the peer server at specified intervals. The status of the peer server is determined by whether the peer server can be pinged successfully.

[0108] Master server A senses that the gateways can ping each other, but master server A cannot ping slave server B. Regardless of whether the heartbeat line is normal, slave server B is disconnected from the gateway network. Then the second business line B2 network failure occurs. At this time, the master server A network is normal, while the slave server B network is abnormal. At this time, you need to stop KA on slave server B to release the floating IP, and start KA on master server A to bind the floating IP.

[0109] Use ping to determine the network status of the peer server, determine the floating IP binding, and avoid split-brain phenomenon.

[0110] In another embodiment of the present invention, any of the servers monitors the designated ports of the gateway and the peer server through TCP SYN scanning technology to sense the network status of the peer server.

[0111] In the specific implementation of this embodiment, any server obtains the gateway IP, the opposite end service port IP and the opposite end heartbeat port IP according to the configuration, monitors the designated port of the opposite end server, and uses TCP SYN scanning technology to detect whether the other server network is normal;

[0112] In addition, the server can also determine the network status of the server and gateway by sensing the 80 port of the gateway's management page.

[0113] In another embodiment of the present invention, when any server senses that the server, service line and heartbeat line of the system are normal, the KA program is started, and the KA program negotiates and binds the floating IP.

[0114] Any server in the system senses that the first service line B1, the second service line B2, and the heartbeat line H are all normal: both servers can start the KA program, and the floating IP performs the negotiation according to the KA program. In this state, the split-brain phenomenon will not occur.

[0115] In another embodiment of the present invention, the manner in which any server performs network status perception is specifically as follows:

[0116] Send a data packet to the gateway through the business line of the server; and judge whether the business line of the server is normal based on the feedback data packet of the gateway; after determining that the business line of the server is normal, send a data packet to the opposite server through the business line of the server, the gateway, and the business line of the opposite server, and judge whether the business line of the opposite server is normal based on the feedback data packet of the opposite server.

[0117] In the specific implementation of this embodiment, the specific method of network perception is described by taking the main server A as an example:

[0118] Send a data packet to the gateway through the first business line B1 of the main server A, and determine whether the first business line B1 is normal based on the feedback data packet from the gateway;

[0119] When no feedback is received from the gateway, it is determined that the first business line B1 is abnormal;

[0120] When receiving the feedback data packet corresponding to the gateway feedback, it is determined that the first business line B1 is normal, and a data packet is sent to the slave server B through the first business line B1, the gateway, and the second business line B2;

[0121] When no feedback is received from server B, it is determined that the second business line B2 is abnormal;

[0122] After receiving the corresponding feedback data packet from server B, it is determined that the second business line B2 is normal.

[0123] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A dual-machine high-availability system, characterized in that: The system includes a gateway, a master server and a slave server; The master server and the slave server are both connected to the gateway via a service line; The master server and the slave server also sense the status of the peer server through a heartbeat line; When any server detects a heartbeat failure with the peer server, it will reconfigure the floating IP address on the corresponding server according to the pre-set configuration information and the preset configuration rules when the peer server's network status is normal. The configuration rules are specifically as follows: When it is determined that the floating IP does not exist, binding the floating IP to the primary server; If a floating IP address exists, the floating IP address configuration is retained.

2. A dual-machine high-availability system according to claim 1, characterized in that: The configuration information includes the gateway IP, the peer service port IP, and the peer heartbeat port IP.

3. A dual-machine high-availability system according to claim 1, characterized in that: Any of the servers continuously obtains the network status of the opposite server by pinging at preset intervals.

4. A dual-machine high-availability system according to claim 1, characterized in that: Any of the servers monitors the designated ports of the gateway and the opposite server through the TCPSYN scanning technology to perceive the network status of the opposite server.

5. A dual-machine high-availability system according to claim 1, characterized in that: When any of the servers senses that the server, business line and heartbeat line of the system are normal, the KA program is started, and the KA program negotiates and binds the floating IP.

6. A dual-machine high-availability system according to claim 1, characterized in that: When any of the servers senses that the opposite server is abnormal, it starts the KA program, and the KA program binds the floating IP to the server.

7. A dual-machine high-availability system according to claim 1, characterized in that: The specific method for any server to perform network status perception is: Send a data packet to the gateway through the business line of the server; and judge whether the business line of the server is normal based on the feedback data packet of the gateway; after determining that the business line of the server is normal, send a data packet to the opposite server through the business line of the server, the gateway, and the business line of the opposite server, and judge whether the business line of the opposite server is normal based on the feedback data packet of the opposite server.

Images