WO2013044828A1

WO2013044828A1 - Virtual cluster system, processing method and device thereof

Info

Publication number: WO2013044828A1
Application number: PCT/CN2012/082196
Authority: WO
Inventors: 江滢
Original assignee: 华为技术有限公司
Priority date: 2011-09-27
Filing date: 2012-09-27
Publication date: 2013-04-04
Also published as: CN102355369B; CN102355369A

Abstract

Disclosed are a virtual cluster system, a processing method and a device thereof. The system comprises at least two partitions, wherein each partition comprises one master node and at least one slave node; each master node and each slave node are respectively provided with at least one virtual machine; a peer-to-peer architecture is used between the master nodes in different partitions; a star architecture is used between the master node and the slave node in each partition; the master nodes comprise one management master node and at least one normal master node, wherein the management master node is used for reselecting a new normal master node or slave node in the partition of the ineffective normal node or slave node when the normal node or the slave node is ineffective, or, rebooting the virtual machine when the virtual machine on the normal master node or slave node is failed. According to the embodiment of the invention, the expandability and availability of the system can be improved.

Description

Virtualized cluster system and processing method and device thereof

This application claims the priority of the Chinese patent application filed on September 27, 2011, the Chinese Patent Office, Application No. 201110301796.0, entitled "Virtualized Cluster System and Processing Method and Equipment", the entire contents of which are incorporated by reference. In this application. TECHNICAL FIELD The present invention relates to network communication technologies, and in particular, to a virtualization cluster system and a processing method and apparatus therefor. BACKGROUND OF THE INVENTION A cluster system has a powerful overall computing performance, storage performance, and management performance, as well as a single system image service form, and transparent availability guarantee and fault tolerance to the user, becoming the main infrastructure structure of the data center. The application of virtualization technology provides a better and more promising solution for cluster development. Virtualization technology allows a single platform to run multiple operating systems simultaneously, and applications can run in separate spaces without affecting each other, significantly increasing the productivity of the computer. Running multiple virtual machines takes full advantage of the computing power of physical servers to provide a fast response capability for the data center.

With the introduction of virtualization technology, scalability and high availability are the biggest challenges facing cluster systems. SUMMARY OF THE INVENTION Embodiments of the present invention provide a virtualized cluster system and a processing method and device thereof, which improve scalability and usability of a virtual machine cluster system.

An embodiment of the present invention provides a method for processing a virtualized cluster system, including:

The node determines whether at least one of the following items occurs: a normal primary node that has failed, a failed standby node, or a virtual machine that is faulty;

After determining that there is a failed normal primary node, the node re-enacts the new normal primary node; After the failed standby node, the new standby node is re-enabled; or, after determining the faulty virtual machine, restart the virtual machine;

The common primary node and the standby node are divided into at least two partitions, each partition includes a primary node and at least one standby node; each primary node and each standby node are respectively provided with at least one a virtual machine; a peer-to-peer architecture is adopted between the master nodes in different partitions; a star schema is adopted between the master node and the standby node in each partition; the master node includes a management master node and at least one common master node .

An embodiment of the present invention provides a processing device for a virtualized cluster system, including:

a determining unit, configured to determine whether at least one of the following items occurs: a normal primary node that has failed, a standby node that has failed, or a virtual machine that is faulty;

The processing unit is configured to re-enable the new normal primary node after determining that the normal primary node that fails, and to re-enable the new standby node after determining that the failed standby node exists; or, after determining the faulty virtual machine, Restart the virtual machine;

An embodiment of the present invention provides a virtualized cluster system, including:

At least two partitions, each partition includes one primary node and at least one standby node; each primary node and each standby node respectively set at least one virtual machine;

A peer-to-peer architecture is used between the master nodes in different partitions;

A star schema is used between the primary node and the standby node in each partition;

The master node includes a management master node and at least one common master node, and the management master node is used to reselect one of the failed primary master nodes or the standby node in the partition where the normal master node or the standby node fails. A new normal primary or standby node, or restart the virtual machine when the virtual machine on the normal primary or standby node fails.

It can be seen from the above technical solution that the virtualized cluster system in the embodiment of the present invention can implement system expansion by adding partitions by partitioning; the primary nodes of the partitions adopt a peer-to-peer structure, which can be eliminated. In addition to bottlenecks, reliability can be improved; reliability can be improved by reselecting new master nodes, standby nodes, or restarting virtual machines. BRIEF DESCRIPTION OF THE DRAWINGS In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly described below. Obviously, the drawings in the following description are some of the present invention. For the embodiments, those skilled in the art can obtain other drawings according to the drawings without any creative labor.

1 is a schematic structural diagram of a system according to a first embodiment of the present invention;

2 is a schematic flow chart of a method according to a first embodiment of the present invention;

3 is a schematic structural diagram of a device according to a first embodiment of the present invention;

4 is a schematic flow chart of a method according to a second embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a system according to a second embodiment of the present invention; FIG.

6 is a schematic flow chart of a method according to a third embodiment of the present invention;

7 is a schematic structural diagram of a system according to a third embodiment of the present invention;

8 is a schematic flow chart of a method according to a fourth embodiment of the present invention;

FIG. 9 is a schematic structural diagram of a system according to a fourth embodiment of the present invention. The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. The embodiments are a part of the embodiments of the invention, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

1 is a schematic structural diagram of a system according to a first embodiment of the present invention. Referring to FIG. 1, the system includes at least two partitions 1, each partition including a master node 11 and at least one slave node (slave) 12 Each of the master node 11 and each of the standby nodes 12 is provided with at least one virtual machine (VM) 13 respectively. For example, referring to FIG. 1, the master node includes a master node A, a master node B, and a master node C. The standby node of the partition where the master node A is located includes the standby node a1 and the standby node a2, and the standby node of the partition where the master node B is located includes the standby node. The node bl, the standby node b2, and the like, and the standby node of the partition where the primary node C is located includes the standby node cl, the standby node c2, and the like.

The master node 11 in different partitions adopts a peer-to-peer architecture, that is, one master node can send resource state information to any other master node, and can also receive resource state information sent by any other master node. A star schema is adopted between the master node 11 and the standby node 12 in each partition, and the standby node sends resource state information to the master node, and the master node does not send resource state information to the standby node. The resource status information can indicate whether the corresponding node is normal or invalid.

The master node includes a management master node and at least one common master node, and the management master node is used to partition the failed normal primary node or standby node after the normal primary node or the standby node fails. Re-select a new normal primary or standby node, or restart the virtual machine when the virtual machine on the normal primary or standby node fails.

Wherein, one of the primary nodes may be pre-configured as a management primary node, and the remaining primary nodes are ordinary primary nodes, and the management primary node stores information of each of the primary node and the standby node and the virtual machine on the node, All nodes are managed in a unified manner, and the faults are handled uniformly after a failure. For example, referring to Figure 1, the master node C can be set as the management master node, and the master node and the master node B are ordinary master nodes.

Corresponding to the above system, the flow between the devices can be as follows.

2 is a schematic flow chart of a method according to a first embodiment of the present invention, including:

Step 21: The node determines whether at least one of the following items occurs: a normal primary node that has failed, a standby node that has failed, or a virtual machine that is faulty;

Step 22: After determining that there is a failed normal primary node, the node re-enacts the new normal primary node; after determining that the standby standby node exists, re-active the new standby node; or, after determining the faulty virtual machine, Restart the virtual machine;

The common primary node and the standby node are divided into at least two partitions, each partition includes a primary node and at least one standby node; each primary node and each standby node are respectively provided with at least one a virtual machine; a peer-to-peer architecture between the master nodes in different partitions; a star schema is adopted between the master node and the standby node in each partition; the master node includes a management master node and at least one Ordinary master node.

The above-mentioned nodes may be specifically a common master node, a management master node, and a standby node. When the nodes are different and the scenarios are different, the foregoing processes may have different specific implementation manners. See the subsequent examples for details.

Correspondingly, the device corresponding to the method can be as follows.

FIG. 3 is a schematic structural diagram of a device according to a first embodiment of the present invention, including a determining unit 31 and a processing unit 32. The determining unit 31 is configured to determine whether at least one of the following items occurs: a normal primary node that has failed, and a failed primary device a node, or a virtual machine that is faulty; the processing unit 32 is configured to re-enable the new normal primary node after determining that the normal primary node is invalid; and re-effective the new standby node after determining that the standby node is invalid; or After determining the faulty virtual machine, restarting the virtual machine; wherein the common primary node and the standby node are divided into at least two partitions, each partition includes one primary node and at least one standby node; At least one virtual machine is set on each of the primary node and each standby node; a peer-to-peer architecture is adopted between the primary nodes in different partitions; a star schema is adopted between the primary node and the standby node in each partition; The master node includes a management master node and at least one common master node.

Certainly, corresponding to the foregoing method flow, the foregoing device may be a common primary node, a management primary node, and a standby node, and the specific functions of the foregoing units are different in different nodes and scenarios. See the examples below for details.

The virtualized cluster system of the embodiment of the present invention can realize system expansion by dividing partitions by partitioning; the peer nodes of the partitions adopt a peer-to-peer structure, which can eliminate bottleneck problems and improve reliability; The primary node, the standby node, or the restart virtual machine can further improve reliability.

4 is a schematic flowchart of a method according to a second embodiment of the present invention. FIG. 5 is a schematic structural diagram of a system according to a second embodiment of the present invention.

Referring to FIG. 4, this embodiment includes:

Step 41: When the cluster is working normally, the normal master node of each partition detects the heartbeat with each other through the heartbeat detection module (heartbeatsync).

For example, the heartbeat detection module of the common primary node A sends the heartbeat information to the heartbeat detection module of the ordinary primary node B.

Step 42: If the heartbeat detection module of the ordinary primary node B detects the heartbeat of the ordinary primary node A If the information is stopped, the multicast fault message carries the identifier information of the common master node A to indicate that the normal master node A is invalid.

After the normal primary node B does not receive the heartbeat information of the ordinary primary node A within a certain period of time, it is determined that the heartbeat stop of the ordinary primary node A is detected.

The identification information can be used to distinguish each node, for example, the ID or address of the ordinary primary node A. The remaining common master node and the management master node will receive the fault message.

Step 43: After receiving the fault message, the heartbeat detection module of the management master node reports the master node fault message to the high availability (HA) module of the management master node, where the master node fault message carries the identifier of the common master node A. information.

Step 44: Manage the HA module of the master node in the partition where the common master node A is located, and reselect a standby node as the new common master node of the partition.

For example, according to the ID priority of each standby node and the dynamic load status of the standby node, the standby node al in the zone where A is located is selected as the new common primary node.

Step 45: The HA module of the management master node sends a migration virtual machine request to the resource management module (ResourceMgmt) of the management master node, where the migration virtual machine request carries the identifier information of the new common master node a1 and the identifier of the common master node A. information.

Step 46: The resource management module of the management master node migrates the virtual machine on the common primary node A to the new normal primary node al.

For example, the configuration information of the virtual machine on the normal primary node A is sent to the new normal primary node al, and the new normal primary node al is instructed to re-run the configuration information to restart the corresponding virtual machine. The configuration information of the virtual machine is information capable of starting the virtual machine, for example, virtual machine software, and the virtual machine can be started after executing the virtual machine software.

Further, when the new common master node joins, the master node has to further update the membership relationship: Step 47: The new normal master node multicasts the join request to the remaining common master nodes, and the heartbeat detection modules of the remaining common master nodes After detecting the join request, the member relationship update request is sent to the corresponding member management module (MembershipMgmt), and the membership update request carries the identifier information of the new common master node and the identifier information of the invalid ordinary master node.

For example, after the normal primary node B receives the join request of the new common primary node a1 multicast, the heartbeat detection module of the common primary node B sends the membership update to the member management module of the ordinary primary node B. The request carries the identification information of A and the identification information of al.

Step 48: The membership management module updates the membership list.

For example, the identification information of the new normal primary node al is added to the member list, and the identification information of the failed ordinary primary node A is deleted.

Referring to the above process, the corresponding module can be as follows:

Referring to FIG. 5, in this embodiment, a common master node 51 and a management master node 52 are involved. Further, for the common master node, the determining unit is specifically a first heartbeat module detection module (Heartbeat Sync) 511, and the processing unit is specifically a first membership relationship management module (MembershipMgmt) 512. For the management master node, the determining unit is specifically a second heartbeat detecting module 521, and the processing unit specifically includes a first high availability (HA) module 522 and a first resource management module (ResourceMgmt) 523.

The first heartbeat detection module 511 is configured to determine that a normal primary node that has failed after detecting a heartbeat stop of any other common primary node, and determine that the ordinary primary node that stops the heartbeat is a failed primary primary node;

The first member relationship management module 512 is configured to receive a first member relationship request message, where the first member relationship request message carries the identifier information of the new common master node and the identifier information of the invalid ordinary master node, and the new The identifier information of the common master node is added to the first member relationship list, and the identifier information of the invalid ordinary master node in the first member relationship list is deleted.

The new common primary node is reselected from the standby node in the partition where the failed primary primary node is located after the primary node receives the first fault message, and the first fault message is the common After the primary node determines that there is a failed normal primary node, the first fault message carries the identifier information of the failed normal primary node.

The second heartbeat detection module 521 is configured to: after receiving the first fault message, determine that there is a normal primary node that is in a failure, where the first fault message is sent by the ordinary master node after determining that the normal primary node is invalid, the first A fault message carries the identification information of the failed normal master node;

The first high availability module 522 is configured to receive a primary node fault message, where the primary node fault message carries the identifier information of the failed normal primary node, and is reselected in the standby node of the failed normal primary node. a new common master node, and carrying the identifier information of the new normal master node and the identifier information of the failed normal master node in the first migration virtual machine request, where the master node fault message is received Sent after the first fault message; The first resource management module 523 is configured to send, according to the first migration virtual machine request message, identifier information of the virtual machine on the failed normal primary node to the new common primary node, and restart the virtual machine. .

In this embodiment, the scalability of the cluster system can be achieved by partitioning. In this embodiment, a peer-to-peer architecture is adopted between the master nodes, and after a master node fails, the master node is known to be invalid and the new master node is reselected in time to improve the availability.

FIG. 6 is a schematic flowchart of a method according to a third embodiment of the present invention, and FIG. 7 is a schematic structural diagram of a system according to a third embodiment of the present invention.

Referring to Figure 6, this embodiment includes:

Step 601: When the cluster is working normally, the standby node of each partition sends heartbeat information to the heartbeat detection module of the common primary node in the partition through the heartbeat detection module.

For example, the heartbeat detection module of the standby node al sends the heartbeat information to the heartbeat detection module of the normal primary node A of the partition.

Step 602: If the heartbeat detection module of the normal primary node A detects the heartbeat stop of the standby node al, sends a heartbeat detection message to another standby node of the local node.

For example, if the normal master node A does not detect the heartbeat information of the standby node a within the set time, the normal master node A detects the heartbeat stop of the standby node a1 and sends heartbeat detection to another standby node a2 of the partition in which it is located. The heartbeat detection message carries the identifier information of the standby node a1.

Step 603: The standby node a2 detects the heartbeat condition of the standby node al.

For example, the standby node a2 sends a pmg message to the standby node al. If the response message returned by the standby node al is not received, the standby node a heartbeat stops.

Step 604: The standby node a2 sends a heartbeat detection result to the normal primary node A, where the heartbeat detection result of the standby node al is carried.

Step 605: If the heartbeat detection result also indicates that the heartbeat of the standby node al is stopped, the normal primary node A multicasts a fault message, and the fault message carries the identification information of the standby node al.

Among them, the rest of the common primary node and the management master node will receive the fault message.

Step 606: After receiving the fault message, the heartbeat detection module of the management master node sends a standby node fault message to the HA module in the management master node, where the standby node fault message carries the identifier information of the failed standby node a1. Step 607: The HA module of the management master node is in the partition where the standby node a is located, and the other standby node is selected as the standby node of the migration virtual machine.

Among them, another standby node can also be selected according to priority, load condition, and the like.

Step 608: The HA module of the management master node sends a migration virtual machine request to the resource management module of the management master node, where the identifier information of the new standby node and the identifier information of the invalid standby node are carried.

For example, if the standby node to be reselected is a2, the migration virtual machine request carries the identification information of al and the identification information of a2.

Step 609: The resource management module of the management master node migrates the virtual machine on the standby node al to the standby node a2.

For example, send the configuration information of the virtual machine on the standby node al to the standby node a2, and instruct a2 to re-run the configuration information to restart the corresponding virtual machine. The configuration information of the virtual machine is information that can be used to start the virtual machine, for example, virtual machine software, and the virtual machine can be started after executing the virtual machine software.

Further, the failed standby node can perform the following actions:

Step 610: After the standby node al finds that its heartbeat information is lost, the pmg gateway sends a ping message to its own gateway.

Step 611: If the ping fails, the response message corresponding to the ping message is not received, and the power is off. Referring to the above process, the corresponding module can be as follows:

Referring to FIG. 7, in this embodiment, a common master node 71, a management master node 72, and a standby node 73 are involved. Further, for the common master node, the judgment unit and the processing unit are the same module, specifically the third heartbeat module detection module 711. For the management master node, the determining unit is specifically a fourth heartbeat detecting module 721, and the processing unit specifically includes a second high-availability module 722 and a second resource management module.

(ResourceMgmt) 723. For the standby node, the determining unit and the processing unit are the same module, specifically the fifth heartbeat module detecting module 731.

The third heartbeat detection module 711 is configured to: after detecting a heartbeat stop of any standby node in the partition where the common primary node is located, determine that the standby node is inactive, and determine that the standby node that is stopped by the heartbeat is a failed standby node. ;

After receiving the second fault message, the fourth heartbeat detecting module 721 determines that there is a standby node that is invalid, and the second fault identifier information is sent by the normal master node after determining that the standby node is invalid. The second fault identification information carries the identifier information of the failed standby node;

The second high-availability module 722 is configured to receive a standby node fault message, where the standby node fault message carries the identifier information of the failed standby node, and reselects a new standby node in the partition of the failed standby node. And transmitting the identifier information of the new standby node and the identifier information of the failed standby node in the second migration virtual machine request, where the standby node failure message is sent after receiving the second fault message;

The second resource management module 723 is configured to send the identifier information of the virtual machine on the failed standby node to the new standby node according to the second migrated virtual machine request message, and restart the virtual machine.

The fifth heartbeat detection module 731 is configured to send heartbeat information when the standby node is not invalid, and does not send heartbeat information when the slave node is invalid, so that the ordinary master node of the partition where the standby node is located is determined according to the situation of the heartbeat information. Whether the standby node is powered off, and performs power-off processing when it is a standby node that is invalid, or detects whether the corresponding standby node is a failed standby node after receiving the detection request, and The detection result is sent to the common primary node, so that the normal primary node performs re-effectiveness of the standby node, and the detection request is sent by the ordinary primary node after receiving the heartbeat information of any standby node within a certain period of time. The detection request carries the identification information of the standby node that stops the heartbeat.

In this embodiment, the scalability of the cluster system can be achieved by partitioning. In this embodiment, the standby node and the primary node adopt a star architecture, and after a standby node fails, the primary node migrates the virtual machine on the failed standby node in time to improve availability.

FIG. 8 is a schematic flowchart of a method according to a fourth embodiment of the present invention, and FIG. 9 is a schematic structural diagram of a system according to a fourth embodiment of the present invention.

Referring to Figure 8, this embodiment includes:

Step 81: When the cluster is working normally, the virtual machine proxy module on each node sends heartbeat information to the heartbeat detection module of the node where it is located.

For example, the virtual machine proxy module of a standby node sends a heartbeat to the heartbeat detection module of the standby node.

Step 82: If the heartbeat detection module of the standby node detects the heartbeat stop of the virtual machine, send a fault message to the normal primary node of the partition. For example, if the heartbeat detection module on the standby node does not receive the heartbeat information sent by the virtual machine proxy module on the corresponding node within a certain period of time, it determines that the corresponding virtual machine heartbeat stops.

Step 83: After receiving the fault message, the normal master node multicasts the fault message, and the fault message carries the identifier information of the faulty virtual machine.

For example, when the virtual machine on the standby node is faulty, the heartbeat detection module on the primary node does not receive the heartbeat information sent by the virtual machine proxy module within a certain period of time, and then determines the primary node. The virtual machine on the fault, multicast failure message.

The above fault message can be received by the remaining common master nodes and the management master node.

Step 84: After receiving the fault message, the heartbeat detection module of the management master node sends a virtual machine fault message to the HA module of the management master node, where the virtual machine fault message carries the label of the faulty virtual machine? Self

Step 85: The ΗΑ module of the management master node sends a restart virtual machine request to the resource management module of the management master node, where the restart virtual machine request carries the identification information of the faulty virtual machine.

Step 86: Manage the resource module of the primary node to restart the virtual machine.

For example, the configuration information of the failed virtual machine is sent to the node where the virtual machine is located, and the corresponding node is instructed to re-run the configuration information to restart the virtual machine. Alternatively, the management master node reselects a node as the target node according to the priority, the load condition, and the like, and then sends the configuration information of the failed virtual machine to the target node, and instructs the target node to re-run the configuration information to restart the virtual machine. Specifically, the resource management module of the target node reselects and runs the configuration information.

Referring to the above process, the corresponding module can be as follows:

Referring to FIG. 9, in this embodiment, a common master node 91, a management master node 92, and a standby node 93 are involved. Further, for the common master node, the determining unit is specifically the sixth heartbeat module detecting module 911, and the processing unit is specifically the fourth resource management module 912. For the management master node, the determining unit is specifically a seventh heartbeat detecting module 921, and the processing unit specifically includes a third high availability module 922 and a third resource management module 923. For the standby node, the determining unit includes a virtual machine proxy module 931 and an eighth heartbeat module detecting module 932, and the processing unit is specifically a fifth resource management module 933.

The sixth heartbeat detection module 911 is configured to: after receiving the virtual machine fault message sent by any standby node in the partition where the common primary node is located, or detecting the heartbeat stop of the virtual machine of the virtual machine, determining the faulty virtual machine And virtualizing the virtual machine or heartbeat indicated by the virtual machine failure message The virtual machine that is determined to be faulty;

The fourth resource management module 912 is configured to: when the virtual machine of the virtual machine is faulty, receive configuration information of the virtual machine that manages the fault that is sent by the primary node, and re-run the configuration information to restart the faulty virtual machine, where the faulty The configuration information of the virtual machine is sent after the management master node receives the third fault message, and the third fault message is sent by the common master node after determining that the virtual machine is invalid. The message carries the identification information of the faulty virtual machine.

The seventh heartbeat detection module 921 is configured to: after receiving the third fault message, determine that the virtual machine has a fault, and the third fault message carries the identifier information of the faulty virtual machine;

The third high availability module 922 is configured to receive a virtual machine fault message and send a restart virtual machine request, where the virtual machine fault message is sent after receiving the third fault message, the virtual machine fault message and the restart virtual The machine request carries the identification information of the faulty virtual machine;

The third resource management module 923 is configured to send configuration information of the virtual machine corresponding to the faulty virtual machine to the node where the faulty virtual machine is located, and instruct the node to re-run the configuration information to restart the faulty virtual machine. .

The virtual machine proxy module 931 is configured to send heartbeat information when the corresponding virtual machine is normal, and does not send heartbeat information when the fault occurs;

The eighth heartbeat detection module 932 is configured to determine, after detecting a heartbeat stop of the virtual machine on the standby node, according to the sending condition of the heartbeat information, determine that the virtual machine that is faulty exists, and determine that the virtual machine whose heartbeat is stopped is a fault virtual Machine

The fifth resource management module 933 is configured to receive configuration information of the faulty virtual machine sent by the management master node, and re-run the configuration information to restart the faulty virtual machine, where the configuration information of the faulty virtual machine is the management master node. After the third fault message is received, the third fault message is sent by the common master node after receiving the virtual machine fault message, and the third fault message carries the identifier information of the faulty virtual machine. The virtual machine fault message is sent by the standby node after detecting a heartbeat stop of the virtual machine on the standby node, where the virtual machine fault message carries the identifier information of the faulty virtual machine.

In this embodiment, the scalability of the cluster system can be achieved by partitioning. In this embodiment, a peer-to-peer architecture is adopted between the primary nodes, and the standby node and the primary node adopt a star architecture, so that after the virtual machine fails, the virtual machine is faulty and the virtual machine is restarted to improve the availability. In summary, in the embodiment of the present invention, by setting a partition, the cluster size can be expanded by adding a partition; by adopting peer-to-peer management of multiple master nodes, the HA bottleneck can be eliminated; and the resource status information of the peer node is different. The resource usage rate information can make the fault monitoring communication overhead small, and the state has the same overhead; when the heartbeat of a standby node stops, the primary node of the partition selects other standby nodes in the partition for arbitration, which can reduce the misjudgment and improve the availability; The peer-to-peer architecture is adopted between the master nodes. Compared with the star architecture, the reliability of the master node is enhanced. By effectively utilizing the standby nodes and migrating the VMs, resource waste can be reduced and management overhead can be reduced.

It can be understood that related features in the above methods and devices can be referred to each other. Further, "first", "second", and the like in the above embodiments are used to distinguish the embodiments, and do not represent the advantages and disadvantages of the embodiments.

A person skilled in the art can understand that all or part of the steps of implementing the above method embodiments may be completed by hardware related to program instructions, and the foregoing program may be stored in a computer readable storage medium, and the program is executed when executed. The foregoing method includes the steps of the foregoing method embodiments; and the foregoing storage medium includes: a medium that can store program codes, such as a ROM, a RAM, a magnetic disk, or an optical disk.

It should be noted that the above embodiments are only for explaining the technical solutions of the present invention, and are not intended to be limiting; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those skilled in the art that: The technical solutions described in the foregoing embodiments are modified, or some of the technical features are equivalently replaced. The modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

Rights request

A method for processing a virtualized cluster system, comprising:

After determining that there is a failed normal primary node, the node re-enacts the new normal primary node; after determining that there is a failed standby node, re-active the new standby node; or, after determining the faulty virtual machine, restarting the virtual machine;

The common primary node and the standby node are divided into at least two partitions, each partition includes a primary node and at least one standby node; each primary node and each standby node are respectively provided with at least one a virtual machine; a peer-to-peer architecture between the master nodes in different partitions; a star schema is used between the master node and the standby node in each partition; the master node includes a management master node and at least one common master node .

2. The method according to claim 1, wherein if the node is a normal primary node,

Determining that there is a failure of the ordinary primary node includes: after detecting the heartbeat stop of any other common primary node, determining that there is a failed normal primary node, and determining that the ordinary primary node of the cardiac arrest is a failed normal primary node;

After the normal primary node is determined to be invalid, the new common primary node is validated, and the method includes: receiving the first membership relationship request message, where the first member relationship request message carries the identifier information of the new common primary node and is invalid. The identification information of the common primary node is added to the first member relationship list, and the identification information of the invalid ordinary primary node in the first member relationship list is deleted;

The method according to claim 1 or 2, wherein when the node is a normal master node, Determining that there is a failed standby node includes: determining that the standby node is inactive after determining that the standby node of the standby node in the partition where the common primary node is located is a failed standby node;

After the determined standby node is determined to be invalid, the new standby node is re-in effect, including:

Receiving a second member relationship request message, where the second member relationship request message carries the identification information of the new standby node and the identification information of the invalid standby node, and adds the identification information of the new standby node to the second member relationship. In the list, deleting the identification information of the failed standby node in the second membership list;

The new standby node is reselected from the standby node in the partition where the failed standby node is located after receiving the second fault message, and the second fault message is the common primary node. And sending, by the second fault message, the identifier information of the failed standby node.

The method according to claim 1 or 2, wherein when the node is a normal primary node,

Determining the faulty virtual machine includes: receiving a virtual machine fault message sent by any standby node in the partition where the common primary node is located, or determining that the faulty virtual machine is detected after detecting the heartbeat stop of the virtual machine of the virtual machine And determining, by the virtual machine fault message, the virtual machine or the heartbeat stop virtual machine as a faulty virtual machine;

After the virtual machine is determined to be faulty, the virtual machine is restarted, including:

Receiving configuration information of the faulty virtual machine sent by the management node, and re-running the configuration information to restart the faulty virtual machine, where the configuration information of the faulty virtual machine is After the third node receives the third fault message, the third fault message is sent by the normal master node after determining that the virtual machine is invalid. The third fault message carries the virtual fault. Identification information of the machine.

5. The method according to claim 1, wherein when the node is a management master node,

Determining that there is a failure of the normal primary node includes: after receiving the first fault message, determining that there is a failed normal primary node, the first fault message is sent by the ordinary primary node after determining that the normal primary node is invalid, The first fault message carries the identifier information of the failed normal master node; After the normal primary node is determined to be invalid, the new common primary node is re-initiated, including: receiving a primary node failure message, where the primary node failure message carries the identification information of the failed ordinary primary node, where A new normal primary node is re-selected in the standby node of the zone where the failed primary node is located, and the identification information of the new common primary node and the identification information of the failed normal primary node are carried in the first migration virtual machine request. Transmitting, the master node fault message is sent after receiving the first fault message;

And sending, according to the first migration virtual machine request message, identifier information of the virtual machine on the failed normal primary node to the new normal primary node, and restarting the virtual machine.

6. The method according to claim 1 or 5, wherein when the node is a management master node,

Determining that there is a failed standby node includes: determining that there is a failed standby node after receiving the second fault message, where the second fault identifier information is sent by the normal master node after determining that the standby node is invalid, the second The fault identification information carries the identifier information of the failed standby node.

Receiving a standby node fault message, the standby node fault message carrying the identifier information of the failed standby node, reselecting a new standby node in the partition of the failed standby node, and the new standby device The identifier information of the node and the identifier information of the failed standby node are sent in the second migration virtual machine request, and the standby node fault message is sent after receiving the second fault message.

And sending, according to the second migration virtual machine request message, the identification information of the virtual machine on the failed standby node to the new standby node and restarting the virtual machine.

The method according to claim 1 or 5, wherein when the node is a management master node,

Determining the faulty virtual machine includes: determining, after receiving the third fault message, the faulty virtual machine, where the third fault message carries the identifier information of the faulty virtual machine;

Receiving a virtual machine failure message and sending a restart virtual machine request, where the virtual machine failure message is sent after receiving the third failure message, where the virtual machine failure message and the restarting virtual machine request carry the faulty virtual machine Identification information;

Sending configuration information of the virtual machine corresponding to the faulty virtual machine to the faulty virtual machine Node, and instructing the node to re-run the configuration information to restart the failed virtual machine.

The method according to claim 1, wherein when the node is a standby node, it is determined that there is a failed standby node, and after determining that the standby node is invalid, the new standby node is re-in effect, including:

The heartbeat information is sent when the standby node is not invalid, and the heartbeat information is not sent when the standby node is invalid, so that the normal primary node of the partition where the standby node is located determines whether the standby node is invalid according to the situation of the heartbeat information, and is in itself When the standby node is a failed standby node, the device is powered off, or the standby node that is not the failed node and the detection request is received, and the corresponding standby node is detected as a failed standby node, and the detection result is notified to the common primary node. The normal master node is configured to perform the re-prioritization of the standby node, where the detection request is sent after the normal master node does not receive the heartbeat information of any standby node within a certain period of time, and the detection request carries the heartbeat stop. Identification information of the standby node.

The method according to claim 1 or 8, wherein when the node is a standby node,

Determining a faulty virtual machine includes:

The heartbeat information is sent when the corresponding virtual machine is normal, and the heartbeat information is not sent when the fault occurs; and after the heartbeat of the virtual machine on the standby node is detected to be stopped according to the sending condition of the heartbeat information, the virtual virtual fault is determined. The virtual machine that determines that the heartbeat is stopped is a faulty virtual machine; after the virtual machine that determines the fault is determined, the virtual machine is restarted, including:

Receiving configuration information of the faulty virtual machine sent by the management master node, and re-running the configuration information to restart the faulty virtual machine, where the configuration information of the faulty virtual machine is after the management master node receives the third fault message. And sending, the third fault message is sent by the common master node after receiving the virtual machine fault message, where the third fault message carries the identifier information of the faulty virtual machine, where the virtual machine fault message is After the standby node detects that the heartbeat of the virtual machine on the standby node is stopped, the virtual machine fault message carries the identifier information of the faulty virtual machine.

A processing device for a virtualized cluster system, comprising:

a processing unit, configured to re-enable a new normal primary node after determining that the normal primary node is inactive; to re-enable the new standby node after determining that the standby node is inactive; or, to determine that there is a faulty virtual node After the machine is started, restart the virtual machine;

11. The device according to claim 10, wherein when the device is a normal primary node,

The determining unit includes:

The first heartbeat detecting module is configured to: after detecting the heartbeat stop of any other common primary node, determine that there is a normal primary node that fails, and determine that the ordinary primary node that stops the heartbeat is a failed normal primary node;

The processing unit includes:

a first member relationship management module, configured to receive a first member relationship request message, where the first member relationship request message carries the identifier information of the new common master node and the identifier information of the invalid ordinary master node, and the new The identifier information of the common master node is added to the first member relationship list, and the identifier information of the invalid ordinary master node in the first member relationship list is deleted.

12. The device according to claim 10, wherein when the device is managing a main node,

The determining unit includes:

a second heartbeat detection module, configured to: after receiving the first fault message, determine that there is a failed normal master node, where the first fault message is sent by the ordinary master node after determining that the normal master node is invalid, A fault message carries the identification information of the failed normal master node;

The processing unit includes:

a first high availability module, configured to receive a primary node failure message, where the primary node failure message is carried The identification information of the failed normal primary node, reselecting a new common primary node in the standby node of the partition where the failed ordinary primary node is located, and identifying the new common primary node and the invalidation The identifier information of the common master node is sent in the first migration virtual machine request, and the master node fault message is sent after receiving the first fault message.

a first resource management module, configured to send configuration information of the virtual machine on the failed normal primary node to the new common primary node according to the first migration virtual machine request message, and restart the virtual machine .

The device according to claim 10 or 11, wherein when the device is a common master node,

The determining unit and the processing unit are located in the third heartbeat detecting module, and the third heartbeat detecting module is configured to determine that there is a failed device after detecting a heartbeat stop of any standby node in the partition where the common primary node is located. a node, and determining that the standby node of the heartbeat stop is a failed standby node;

The device according to claim 10 or 12, wherein when the device is a management master node,

The determining unit includes:

a fourth heartbeat detecting module, configured to: after receiving the second fault message, determine that there is a standby node that is invalid, where the second fault identifier information is sent by the common master node after determining that the standby node is invalid, the second The fault identification information carries the identifier information of the failed standby node.

The processing unit includes:

a second high-availability module, configured to receive a standby node fault message, where the standby node fault message carries the identifier information of the failed standby node, and reselects a new standby node in the partition where the failed standby node is located And transmitting the identifier information of the new standby node and the identifier information of the failed standby node in the second migration virtual machine request, where the standby node failure message is sent after receiving the second fault message;

a second resource management module, configured to invalidate the virtual machine according to the second migration virtual machine request message The identification information of the virtual machine on the standby node is sent to the new standby node and the virtual machine is restarted.

The device according to claim 10, wherein when the device is a standby node, the determining unit and the processing unit constitute a fifth heartbeat detecting module, and the fifth heartbeat detecting module is configured to The heartbeat information is sent when the standby node is not invalid, and the heartbeat information is not sent when the standby node is invalid, so that the normal primary node of the partition where the standby node is located determines whether the standby node is invalid according to the situation of the heartbeat information, and is The failed standby node performs power-off processing, or detects whether the corresponding standby node is a failed standby node after receiving the detection request, and notifies the normal primary node of the detection result. The normal master node performs the re-prioritization of the standby node, and the detection request is sent after the normal master node does not receive the heartbeat information of any standby node within a certain period of time, and the detecting request carries the heartbeat stop device. The identification information of the node.

The determining unit includes:

The sixth heartbeat detection module is configured to: after receiving the virtual machine fault message sent by any standby node in the partition where the ordinary primary node is located, or after detecting the heartbeat stop of the virtual machine of the virtual machine, determining the faulty virtual machine And determining, by the virtual machine fault message, the virtual machine or the heartbeat stop virtual machine as a faulty virtual machine;

The processing unit includes:

And a fourth resource management module, configured to: when the virtual machine of the fault is faulty, receive configuration information of the faulty virtual machine sent by the management master node, and re-run the configuration information to restart the faulty virtual machine, where the faulty The configuration information of the virtual machine is sent after the management master node receives the third fault message, and the third fault message is sent by the common master node after determining that the virtual machine is invalid. The message carries the identification information of the faulty virtual machine.

17. The device according to claim 10 or 12, wherein when the device is a management master node,

The determining unit includes:

a seventh heartbeat detecting module, configured to: after receiving the third fault message, determine the faulty virtual machine, where the third fault message carries the identifier information of the faulty virtual machine;

The processing unit includes: a third high-availability module, configured to receive a virtual machine fault message and send a restart virtual machine request, where the virtual machine fault message is sent after receiving the third fault message, the virtual machine fault message and the restart virtual The machine request carries the identification information of the faulty virtual machine;

a third resource management module, configured to send configuration information of the virtual machine corresponding to the faulty virtual machine to a node where the faulty virtual machine is located, and instruct the node to re-run the configuration information to restart the faulty virtual machine .

18. The device according to claim 10 or 15, wherein when the device is a standby node,

The determining unit includes:

a virtual machine proxy module, configured to send heartbeat information when the corresponding virtual machine is normal, and not send heartbeat information when the fault occurs;

The eighth heartbeat detection module is configured to determine, after detecting a heartbeat stop of the virtual machine on the standby node, the virtual machine that is faulty, and determine that the virtual machine with the heartbeat is a fault virtual Machine

The processing unit includes:

a fifth resource management module, configured to receive configuration information of a faulty virtual machine sent by the management master node, and re-run the configuration information to restart the faulty virtual machine, where configuration information of the faulty virtual machine is the management master node After the third fault message is received, the third fault message is sent by the common master node after receiving the virtual machine fault message, and the third fault message carries the identifier information of the faulty virtual machine. The virtual machine fault message is sent by the standby node after detecting a heartbeat stop of the virtual machine on the standby node, where the virtual machine fault message carries the identifier information of the faulty virtual machine.

19. A virtualized cluster system, comprising:

The master node includes a management master node and at least one common master node, and the management master node is used to locate the failed normal master node or standby node after the normal master node or the standby node fails. Re-select a new common primary or standby node in the zone, or restart the virtual machine when the virtual machine on the common primary or standby node fails.

20. The system of claim 19, wherein:

The common master node is the device according to claim 11; the management master node is the device according to claim 12;

Or,

The common master node is the device according to claim 13; the management master node is the device according to claim 14; and the standby node is the device according to claim 15;

Or,

The common master node is the device according to claim 16; the management master node is the device according to claim 17; and the standby node is the device according to claim 18.