WO2012028013A1

WO2012028013A1 - Implementing method for main/standby configuration of board cards, and board card

Info

Publication number: WO2012028013A1
Application number: PCT/CN2011/075989
Authority: WO
Inventors: 黄文伟; 周海山; 杨骐
Original assignee: 中兴通讯股份有限公司
Priority date: 2010-09-01
Filing date: 2011-06-20
Publication date: 2012-03-08
Also published as: CN101938417A

Abstract

The present invention discloses an implementing method for main/standby configuration of board cards, and a board card, wherein the method includes: for multiple standby board cards in multiple board cards, in the process of competition, each standby board card sends a competition message to other standby board cards and receives competition messages from other standby board cards, wherein the competition message sent by each standby board card carries hardware information of the standby board card; each standby board card compares its hardware information with the hardware information in the received competition messages and determines the competition winner standby board card according to competition principles; an only main control node is determined according to the competition conditions of the multiple standby board cards and the main control node determines one standby board card from other standby board cards as a standby control node. The above solution solves the problem of unreasonable choosing mechanism of the main control node and no guarantee of the chosen main control node performance in current cluster router system.

Description

The present invention relates to the field of communications, and in particular, to a method and a board for implementing an active/standby configuration of a card. Background technique

Due to the rapid development of the router technology, the routing capacity has been greatly improved, which is mainly reflected in the increase in the number of slot slots, the improvement of the exchange capacity of each slot, and the increase in the density of the card ports.

At present, the types of routers are increasing, and the functional division of each type of router is becoming more and more obvious. In a variety of routers, for example, core layer routers need to have fast forwarding capabilities, high security, high stability, and sufficient capacity. Although router technology is constantly improving, the development of Internet traffic will always exceed the speed of device capacity updates. For a single router, if you need to expand it, you need to consider the development maturity of the optical device, power supply, heat dissipation performance, and load-bearing capacity of the equipment room. Currently, due to the above limitations of router expansion, the development of single-box routers Technology has gradually approached the limit, and the emergence of cluster routers has solved the problem of router expansion.

Through centralized and integrated control and management, the cluster technology enables the routers in the cluster system to work together well, thus breaking the limitation of single-chassis routers in terms of switching capacity, power consumption, and heat dissipation. The single-box router system smoothly expands to a larger capacity routing switching system.

Although cluster routers have advantages in capacity, the complexity of cluster routers is relatively high, and the application of cluster router systems has high requirements for uninterrupted online upgrade performance, disaster tolerance performance, automatic fault monitoring performance, and alarm functions. Therefore, it will increase the difficulty of implementing the cluster router system.

In a cluster router system, multiple routers in the system are divided. Generally, the cluster router system includes the master node and the standby node. The master control node is a centralized control point in the cluster router system, and is mainly used to implement version loading, user operation and maintenance, centralized calculation, and decision making of the entire cluster router system. For a cluster router system, the selection and monitoring of the master node directly affects the reliability and availability of the entire system. Therefore, considering the importance of the master node, the master node must be designed as a 1+1 master.

Generally, when the master node is selected in the cluster router system, the master node selection mode of the single-frame router system can be used, that is, the main control board on a certain chassis is fixed as the master node, if This choice, The software flow of the single-box system can be directly used, so the implementation process is relatively simple, but the entire cluster router system completely depends on the chassis where the selected master node is located, and the performance of the selected master node cannot be guaranteed. Both can meet the performance requirements of the master node, which will affect the normal operation of the entire cluster router system, and the reliability is poor.

In view of the problem that the selection mechanism of the master node in the current cluster router system is unreasonable and the performance of the selected master node is not guaranteed, an effective solution has not been proposed yet. Summary of the invention

Aiming at the problem that the selection mechanism of the master node in the current cluster router system is unreasonable and the performance of the selected master node is not guaranteed, the present invention provides a method for implementing the master/slave configuration of the board and the board, which can reasonably select the optimal performance. The master node.

In order to solve the above technical problem, the technical solution of the present invention is implemented as follows:

According to an aspect of the present invention, a method for implementing a board active/standby configuration is provided for determining a master node and a standby node from a plurality of boards located in a plurality of chassis.

The implementation method of the card master/slave configuration according to the present invention includes: for each of the plurality of boards, each of the candidate boards sends a competition message to the other candidate board, where The competition message sent by each candidate card carries the hardware information of the candidate card; during the competition, each of the candidate cards receives the competition message from the other candidate card; Each of the candidate boards compares the hardware information with the hardware information in the received competition message, and determines an optional board that wins according to the competition principle, wherein the competition principle is used for Determining, as a result of the comparison result of the hardware information, an optional board as a master node; determining a unique master node according to the contention of the plurality of candidate boards, and the master node is in another candidate board An optional board is identified as a standby node.

The determining, by the plurality of candidate boards, the unique master node includes: determining, among the plurality of candidate boards, an optional board that has never failed to be determined as the The master node.

Moreover, the implementation method further includes: if the candidate card fails to compete in the competition, the candidate card stops sending the competition message to the other candidate cards.

The method may further include: setting a competition flag to the candidate card in advance, and setting the competition flag to true, the setting indicating that the candidate card is allowed to become a master by transmitting the competition message. Node; for each candidate board, if the candidate board determines that it has failed in the competition with any of the other optional boards, then the content flag of the candidate board is set to False, this setting is used to indicate that the candidate card is prohibited from becoming a master node by sending a competition message. Moreover, determining, among the plurality of candidate boards, only one of the candidate boards that have never failed to be determined as the master node includes: broadcasting or multicasting to all other candidate boards on one of the candidate boards If the number of times the contention message reaches the predetermined number of thresholds, if the candidate card does not continue to receive the competition message from the other candidate card, the candidate card is determined as the master node. .

Preferably, the hardware information includes the physical location information of the card and the primary standby status information of the board in the chassis. The primary standby status information is used to indicate that the board is the main control board in the chassis. Card or standby control board.

Moreover, the above competition principle includes: preferentially determining an optional board as a main control board in the respective chassis as the winning candidate card; if the two selected boards are in the respective chassis The main control board determines the candidate board with a smaller or larger physical position as the winning candidate board in the two candidate boards for comparison.

Determining, by the master node, one of the other candidate boards as the standby control node includes: the master node notifying the other candidate boards to report the respective hardware information, and according to the reported hardware information An optional board that is located in a different chassis with the master node is determined as a standby node.

Preferably, the determining, according to the reported hardware information, an option board that is located in a different chassis of the master node as the standby node further includes: if there is a different chassis in the chassis and in the respective chassis As a plurality of candidate boards of the main control board, randomly selecting one of the plurality of optional boards is determined as the standby node, or the physical position of the plurality of candidate boards is minimized Or the largest candidate card is determined as the standby node.

In addition, after the master node and the standby node are determined, the method further includes: the master node sending a service allocation request to the main control card of the multiple chassis, and receiving the Hardware information and load information of each main control board returned by the main control board of the multiple chassis; the main control node is based on the hardware information and load information of the main control board of the multiple chassis Determining, by the main control board of the plurality of chassis, the main service node and the standby service node, the hardware information and load information reported by the main control board of the multiple chassis from the plurality of Determining the primary service node and the standby service node in the main control board of the chassis includes: the primary control node first determines the primary control card with the smallest load according to the load information as the primary service node, and the candidate board One of the cards in the card except the master node, the standby node, and the main service node is determined as a standby service node.

Determining, in the candidate board, one of the main control board except the main control node, the standby control node, and the main service node as the standby service node includes: A main control board located in different chassis is determined as the standby service node.

In addition, after determining the master node and the standby node, the method further includes: the master control section The point is sent to the standby control node by the keepalive message in a predetermined period; if the master control node meets the switching condition, the master control node and the standby control node are switched, wherein the switching condition includes: The standby control node does not receive the keep-alive message from the master control node for a predetermined period of time, and the standby control node learns the master control node by using other boards in the same chassis as the master control node. The working status is abnormal.

Optionally, the optional board may include part or all of the main control board and the standby control board of the multiple chassis.

According to another aspect of the present invention, a card is provided.

The board includes: a sending module, configured to send a competition to the other board, wherein the contention carries the hardware information of the board; the receiving module is configured to send the module to the other The board receives the contention message from the other board in the process of sending the contention message; the first determining module is configured to send the hardware information of the board and the competition information received by the receiving module from other boards. The hardware information in the text is compared, and the winning board is determined according to the competition principle, wherein the competition principle is used to determine the board as the master node according to the comparison result of the hardware information; the control module uses When the result of the determination is that the board of the first determining module fails to compete, the sending module stops controlling to send the contention message; and the second determining module is used to be the master node of the board where the board is located. Next, one of the other candidate boards is determined as the standby node.

The hardware information includes the physical location information of the board and the primary and backup status information of the board in the chassis. The primary standby status information is used to indicate that the board is the main control board in the chassis. Card or standby control board.

The board may further include: a notification module, configured to notify other boards to report respective hardware information after the board is determined to be the master node; and the second determining module is specifically configured to report according to the other board The hardware information is determined as a standby node with an optional board in which the master node is located in a different chassis.

The present invention determines the master node by making a plurality of candidate boards capable of being used as the master node to select the master node, so that the determination range of the master node is not limited to a fixed chassis, and The performance of the finalized control node can be ensured, and the problem that the system reliability caused by the master node can be determined in a certain fixed machine frame and the determination process is unreasonable can be avoided.

Further, with the above technical solution of the present invention, the advantages of the multi-chassis multi-control board of the cluster router system can be fully utilized, and the main control node, the standby control node, the active service node, and the standby service are implemented between multiple chassis. The determination of the node allows the master node to run on any of the frames, and can avoid the board in which the active/standby relationship is located in the same chassis, and effectively improve the system under the premise of ensuring the performance of the master node and the active service node. Reliability and throughput; In addition, the monitoring functions of the master node and the standby node can be realized by means of hardware signals and logic signals. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic flowchart diagram of a method for implementing a card master/slave configuration according to an embodiment of the present invention;

2 is a schematic diagram of processing for determining a master node in an implementation method of a board master/slave configuration according to an embodiment of the present invention;

3 is a schematic diagram of processing for determining a primary service node and a standby service node in an implementation method of a board master/slave configuration according to an embodiment of the present invention;

FIG. 4 is a schematic diagram showing the distribution of nodes in the active/standby relationship determined by the implementation method of the active/standby configuration of the board according to the embodiment of the present invention;

FIG. 5 is a schematic diagram of processing performed by a master control node and a standby control node according to an implementation method of a board master/slave configuration according to an embodiment of the present invention; FIG.

6 is a schematic structural view of a card according to an embodiment of the present invention. The present invention proposes that, in the current cluster router system, the selection mechanism of the master node is unreasonable and the performance of the selected master node is not guaranteed, the present invention proposes that multiple candidate boards are used in multiple chassis frames. According to the result of the competition, the master node is determined, and the determined range of the master node is not limited to a fixed chassis, and the performance of the final master node can be guaranteed.

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

According to an embodiment of the present invention, a method for implementing a card master/slave configuration is provided, which is used to determine a master node and a standby node from a plurality of boards located in multiple chassis.

FIG. 1 is a schematic flowchart of a method for implementing a master/slave configuration of a card according to an embodiment of the invention. As shown in FIG. 1 , an implementation method of a board active/standby configuration according to an embodiment of the present invention includes:

Step 101: For multiple candidate cards in multiple boards (in this embodiment, the performance of the optional board should be able to meet the performance requirements of the master node), and each candidate board is provided to other devices. The selection card sends a competition message, where the competition message sent by each candidate board carries the hardware information of the candidate board;

Step 103: In the process of competition, each candidate board receives a competition message from another candidate board; Step 105, each candidate board sends its hardware information and the received competition message The hardware information is compared, and the candidate board that wins the competition is determined according to the competition principle, wherein the competition principle is used to determine the candidate board as the master node according to the comparison result of the hardware information (wherein each time When comparing, because of the two comparisons For the board, the winning board is not necessarily the final board that is finalized as the master node. Therefore, for a comparison, the purpose of the competition principle is actually to determine the participation in comparing two candidate boards. More suitable as an optional board for the master node);

Step 107: Determine a unique master node according to the competition condition of the multiple candidate boards, and the master node determines one of the other candidate boards as the standby node.

By means of the above processing, the master control node is determined by making a plurality of candidate boards capable of being used as the master control node to select the master control node, so that the determination range of the master control node is not limited to a fixed chassis. The performance of the finalized master node can be ensured, and the problem that the system reliability caused by the master node can be determined in a certain fixed machine frame and the determination process is unreasonable is avoided.

Example 1

Preferably, the optional board should include all the normal running main control boards and the standby control boards of the multiple chassis, or only part of them, but it should be ensured that the optional boards have other boards in the local frame. The function or ability to perform the control.

Moreover, the foregoing hardware information includes physical location information of the card (for example, the chassis identifier where the card is located, the location identifier of the card in the chassis, the address information of the card), and the primary backup of the card in the chassis. Status information, where the primary and backup status information is used to indicate that the board is the main control board or the standby control board in the chassis.

Since the competition sent by the board may be lost, and based on the competition method used in this embodiment, if a board judges that it wins in the competition, the board should continue to transmit. For example, you can send a contest message at intervals of 1 second (other time lengths can be used as needed).

Moreover, the above-mentioned competition principle may include: preferentially determining an optional board as a main control board in the respective chassis as the winning candidate card; if the two boards being compared are the main players in the respective chassis The control board selects one of the two boards to be compared to determine the winning candidate board, or selects the smaller or larger physical board as the winning board. .

Generally, each mainframe is pre-configured with a main control board and a standby control board. Among them, for the reasons of reliability and management efficiency, it is necessary to add some functions to the main control board (including the board). The hardware of the card is configured to have a function of managing by hardware signals), so that the main control board can control the boards in the frame by hardware signals (for example, performing reset operations, etc.), therefore, it should be considered The optional board for communication with hardware signals is more suitable as the master control node; since the main control board needs to control other boards through hardware signals, the requirements for the main control board are higher, and the standby control board in the chassis It can have the function of controlling other boards in the chassis through hardware signals and logic signals, or can only have the function of controlling by using logic signals (in progress) During control, if the board can only use logic signals, the board can send logic signals to boards that can be managed by hardware signals, and then the boards send corresponding hardware signals to other boards) In addition, the performance of the standby control board in a chassis is not higher than that of the main control board. Therefore, selecting the main control board as the master node can effectively guarantee the performance of the master node.

For example, suppose the card Α of the chassis 1 receives the content of the card B from the chassis 2, if the card A is the main control card in the chassis 1, the card B is the chassis 2 On the standby control board, board A compares its own hardware information with the hardware information of board B carried in the competition message. Since board A is the main control board, board A will judge it. It itself won the competition and continued to send the competition message.

It is assumed that in the above example, the board A and the board B are the main control boards in the respective chassis. At this time, the board A can determine whether it wins according to a predetermined principle, for example, the address can be small ( The larger card is determined as the winning party. At this time, since the card A belongs to the chassis 1 and the address is 1XXX, the card B belongs to the chassis 2 and the address is 2XXX, so the address of the card A is small. When board A judges that it wins, it will continue to send the competition message. Among them, the principle of determining the party with the smaller address as the winner is only a specific example. In the practical application, other comparison methods can be used, as long as one unique among a plurality of candidate boards can be determined. The optional board is OK, and when judging, all boards should use this principle to judge, avoiding multiple master nodes at the same time.

For an optional board, if the board fails to pass, the board can always send the competing message at a predetermined interval, and once an optional board receives any other optional board. After competing for the message and judging that the card itself has failed, the card will immediately stop sending the competition message to the other board. In this way, after the competitive message has been sent multiple times on the candidate board that has not yet failed, there will be a unique candidate card that fails to compete.

In an actual application, a threshold of the number of transmissions and a time interval for transmitting the contention packets may be set, and each of the candidate boards may be in the period after the board is started in the chassis and after entering the steady state. The other board sends a competition message (once the board fails to compete, the board immediately stops sending), and the number of times that an optional board broadcasts or multicasts the competition message to all other optional boards reaches In the case of a predetermined number of thresholds, if the candidate card does not continue to receive the contention from other candidate boards, the board may be considered as the only candidate board that has not failed. At this point, the candidate board can be determined as the master node.

When the master node determines one of the other candidate boards as the standby node, the master node can notify other candidate boards to report the respective hardware information, and the hardware information according to the report will be located with the master node. An optional board of a different chassis is determined as a standby node.

In addition, in the related art, the cluster router system limits the selection range of the master node and the standby node to In the same chassis, hardware signals can be used to monitor the master node and the standby node, which reduces the complexity of monitoring. However, if the chassis is abnormal, it will affect the normal operation of the entire cluster system.

In order to avoid the problem, in the embodiment of the present invention, the master node may notify other candidate boards to report respective hardware information, and according to the reported hardware information, an optional board that is located in a different chassis from the master node. The device is determined as a standby node, and if there are multiple candidate boards that are in different chassis from the master node and are used as the main control board in the respective chassis, one of the multiple candidate boards is randomly selected. The board is determined as the standby node, or the candidate board having the smallest or largest physical position in the plurality of candidate boards is determined as the standby node, and similarly, one board is selected from the plurality of candidate boards. As a standby node, other selection methods can be used as long as one card can be uniquely determined from a plurality of boards.

In addition, in this embodiment, the system may be used to set a competition flag that is qualified as an optional card, wherein, in advance, if the contention flag is set to true, the permission of the candidate card is allowed to pass. The sending of the competition message becomes the master node. If the flag is set to false, it means that the candidate card is prohibited from being the master node by sending the competition message.

Specifically, in practical applications, in order to enable the operator to specify the range selected by the master node, the competition flag may be preset before the board sends the competition message; in addition, the competition flag may also be in the process of competition. According to the failure of the card competition, there will only be a single candidate board that wins the competition. At this time, the board can be determined as the master node according to the setting of the competition flag. For example, for each candidate board, if the candidate board determines that it has failed in the competition with any of the other optional boards, then the content flag of the candidate board is set to False, this setting is used to indicate that the candidate card is forbidden to become the master node by sending the competition message, so that after multiple times of the message transmission and competition, there may be only one competition. The flag is an optional board, and the board can be directly determined as the master node. In addition, the competition flag may not be set. By default, the main control board and the standby control board in each chassis participate in the competition.

If it is necessary to determine the need to set the competition flag as a true board before starting the competition, the operator or system can pre-specify the optional board that needs to be set to true. Moreover, it is also possible to make a determination based on historical data, for example, preferentially setting the competition flag of the candidate board that was once the master node to be true.

In addition, after the master node and the standby node are determined, the master node can send a service allocation request to the main control board of the multiple chassis, and receive the main control boards returned by the main control board of multiple chassis. The hardware information and the load information; in this way, the master node can determine the active service node and the standby service from the main control boards of the multiple chassis according to the hardware information and load information reported by the main control board of the multiple chassis. node.

The main control node is based on hardware information and load information reported by the main control board of the multiple chassis from multiple chassis. The main service node and the standby service node are determined in the main control board, and the specific manner is as follows: The main control node firstly determines the main control board with the smallest load according to the load information as the main service node, and the main service is based on the hardware information. The main control board whose nodes are located in different chassis is determined as the standby service node.

In addition, after the master node and the standby node are determined, the master node sends a keep-alive message to the standby node at a predetermined period; if the master node meets the switching condition, the master node and the standby node are switched, where The switching condition includes: the standby control node does not receive the keep-alive message from the autonomous control node within a predetermined time period, and the hardware working state of the master control node is abnormal (the standby control node can obtain the master through the partner node of the master control node) If the standby node learns that the hardware working state of the master node is normal through the partner node of the master node, the partner node of the master node may be notified by a logic signal, and the master node is indicated by the partner node. Reset. In this way, when the master node fails, the standby node is switched to the master node, and a main control board that is not the local chassis is selected as the standby control node, thus implementing 1+1 protection switching.

In addition, in practical applications, the board may be configured to stop sending the contention message after the number of times the board fails to reach a certain threshold, and the threshold value should be less than the reference when determining the master node. The threshold for the number of times the message is sent.

Example 2

In this example, the method for determining the standby control node, determining the primary service node, determining the standby service node, and monitoring the master control node may refer to the description of the previous example. The difference between this example and the example 1 is that When determining the master node, if a board fails in a competition, the board records the candidate board that wins in this competition, and continues to send the competition message, which is sent on the board. When the number of times of the message reaches the predetermined number of times, the other candidate boards that win out of the board are counted in all the optional boards. According to the statistics of each board, all the optional boards can be ranked. At this time, one of the first few candidate boards can be selected to determine the master node, and the selection process can be a random selection.

On the basis of this example, the setting of the competition flag can be further optimized. For example, after obtaining the ranking of all the optional boards, the ranking can be recorded. When the next time the main control board is required, the ranking can be prioritized. The competition flag of the previous multiple boards is set to true.

The method according to the present embodiment will be described below with reference to Fig. 2 and with reference to the above-described example 1.

As shown in Figure 2, the following processes are specifically included:

Because the master node is the centralized control point of the cluster router, the master node needs to be responsible for the version loading and service configuration of the entire system. Therefore, step 21 needs to be performed first, that is, when the system starts, the competition process of entering the master node is performed. . Step 22: When the main control board is started, the competition flag is read from the configuration information, and the main control board can perform the competition of the main control node only if the competition flag is true. The competition flag can be configured by the operator at startup, which gives the operator an opportunity to specify the location of the master node in the cluster router system; and the competition flag can be set or modified during operation. When the main control board synchronizes all the data of the master node and can take over the function of the master node at any time, the content flag of the main control board can be set to true. In this way, through the judgment of the competition sign, it is guaranteed that the main control board of the main control node can complete the main control function. Before the competition begins, if it is determined that a certain main control board does not satisfy the condition of the competition master node, step 23 is performed; if yes, step 24 is performed; step 23, the main control board will abandon the competition master. The control node waits for the main control node to perform service allocation on the board, and ends the current process;

Step 24: The board that satisfies the competitive condition periodically sends the master control node to compete for the 4th message, and the hardware information of the board is included in the 4th article, including the physical location, the hardware active/standby state, and the like. Information is the condition for judging when the master node competes.

Step 25: Each main control board determines whether it has received a competition message from another main control board, and the hardware information carried in the received message is obtained for the main control board that has not failed. Conducting a competitive judgment, if the board wins, it can continue to send the competition message, and for the main control board that has failed in the competition, it will not continue to send the competition · 艮艮, and may also be wrong The other main control boards sent the contest 4 essays for processing.

If the board receives the competition of other boards, go to step 26, otherwise go to step 29;

Step 26: According to the competition message sent by the other board, the priority is judged as the main control board in the respective chassis (that is, the board with the hardware signal as the main board) wins, if both sides are the main control board Card or standby control board, continue to determine which board has a smaller physical address, if the board wins, proceed to step 28; if the board fails, go to step 23;

Step 28, continue to send the competition message, and return to step 25 to continue to determine whether to receive the competition from other boards.

Step 29: If the board fails to compete, and the content of the other board is not received, the board is determined as the master node, and the master node collects other main control boards. Hardware information (for example, may be a physical address), and perform step 30;

Step 30: Select a main control board whose physical address is not the local frame as the standby control node according to the reported physical address. In step 29, if a board continuously transmits 60 times of the master node's contention and no longer receives the competing message of the other board, it indicates that the board is currently running normally. The control board competes with the master control node, and the board can enter the processing flow of the master control node. In addition, in the above process, the master node will continuously send the master node announcements, so that the master board that is started later does not need to perform master control when receiving the master node notification message. The node competes and waits directly for the master node to allocate traffic on the board. At the same time, the master node competes. The 艮艮也是 is also used as a kind of keep-alive 艮 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,

When the master node notifies other boards to report their physical information, the master node can sort the physical information collected in 5 seconds (or 10 seconds, etc.) according to the priority, according to the non-frame and hardware signals. The priority principle of the primary and physical locations is selected to select the standby node. If the information reported by other main control boards is not received within 5 seconds, it means that the other main control boards in the current system are not running normally. The main control node is in 1 minute (may be 2 minutes, half minutes, etc.) After that, the process of physical information collection is resumed until the qualified standby node is generated. After the master node and the standby node are selected, the system enters the control node 1+1 backup state. The standby control node does not carry the control service, but all the data configurations of the master node are synchronized on the standby control node. When the master control node fails, the standby control node can immediately take over.

After the main control node and the standby control node of the system are selected, the service distribution process is entered.

As shown in Figure 3, the master node sends all the cards to the board when it selects the service node (which can be the board that participates in the competition, as shown in Figure 3, which can be board C and board D). After requesting, board C and board D will return the hardware information and load information related to the board to the master node. After the master node collects the information, the board with the lowest load is determined as the active service node. When there are multiple main control boards with the same load and the smallest, the priority of the hardware and the physical location is small. The specific judgment method is similar to the previous method of selecting the standby node. After the primary service node is selected, the board with the lower load is preferentially selected as the standby service node in the other chassis except the chassis where the active service node is located. If the board cannot be uniquely determined according to the above conditions, , you can select the alternate service node by referring to the physical location.

In addition, in the above process, if the information of the other board is not collected within 5 seconds (or may be 10 seconds or other time period) after the master node sends the service node allocation request, the current other board is indicated. The card is not working properly. The master node needs to set a 1-minute timer (which can also be a timer of other duration). After 1 minute, it resends the service node allocation request for information collection.

According to the above processing, a master standby node configuration such as that shown in Fig. 4 can be obtained. As shown in FIG. 4, in the three chassis connected to the control plane switching network, the master node and the standby service node are located in the chassis X, and the standby node is located in the chassis ,, and the chassis 还 includes The main control board of the subrack (without service commitment), the main service node is located in the subrack, and the main control board of the subrack is also included in the subrack (no service commitment).

This ensures that the 1+1 backup nodes are not in the same chassis, so there will be no single in the entire system. The failure point of one can effectively improve the reliability of the system.

As shown in FIG. 5, the standby control node and the primary/secondary service node need to periodically send a keep-alive message to the primary control node, and the primary control node can also periodically send a keep-alive message to the standby control node, which can be classified into the following two types. Happening:

(Case 1) If the master node does not receive a keep-alive message for a certain board for 10 seconds, it considers that the board has failed. At this time, the data of the faulty board can be deleted on the master node. If the backup node and the standby service node are faulty, the master node needs to perform the process of selecting the standby node again. If the active service node fails, the master node first notifies the standby service node to initiate the master and backup. The switching process allows the standby service node to take over the service, and then starts the standby service node selection process;

(Case 2) The standby control node periodically receives the keep-alive message sent by the master control node. If the keep-alive message of the master control node is not received within 5 seconds, the master control node may be in a fault state. The master node is the only control point of the system. If two master nodes appear in the system at the same time, the processing flow of the system is confusing. Therefore, the standby node needs to pass the logic flow (since the master node and the standby node are located differently The chassis, so the master node and the standby node can communicate with each other through logic signals. After monitoring the abnormality of the master node, hardware signals are also needed to ensure that the master node does not work. At this time, the standby control node will query the partner board of the master node (for example, the other board in the same chassis as the master node queries the status of the master node, the partner board of the master node, and the master node. There is an active/standby signal line, so the partner board can obtain the running status of the master node. If the partner board checks through the hardware signal line that the master node has been running abnormally, the standby controller can be notified to perform the switching immediately; After the hardware signal line checks that the main control node is still running normally, the running status of the main control node is notified to the standby control node, and the standby control node rechecks the status of the main control node, if the main control node is still not received within 1 minute. The keep-alive message indicates that although the hardware signal of the master node is normal, but the software flow is already in a fault state, the standby control node notifies the partner board of the master node to reset the master node through the hardware signal, and the partner board after the reset is completed. The reset result is notified to the standby control node; the standby control node receives the master control node of the master node partner board response After the result of abnormal, initiates switching the control system take over the processing flow, flow into the master node processing becomes a master node and select the standby control node.

Through the above processing, the advantages of the multi-chassis multi-control board of the cluster router system can be fully utilized, and the determination of the main control node, the standby control node, the active service node, and the standby service node is implemented between the multiple chassis, and the main control node is allowed. It runs on any box and can prevent the boards with the active/standby relationship from being located in the same chassis. This improves the reliability and throughput of the system while ensuring the performance of the master node and the active service node. It is also possible to implement monitoring functions of the master node and the standby node by means of hardware signals and logic signals.

According to an embodiment of the invention, a board is also provided. As shown in FIG. 6, the board according to an embodiment of the present invention includes:

The sending module 61 is configured to send a competition message to the other board, where the competition message carries the hardware information of the board;

The receiving module 62 is configured to receive the contention from the other board in the process of sending the contention message to the other board by the sending module;

The first determining module 63 is connected to the sending module 61 and the receiving module 62, and is configured to compare the hardware information of the board and the hardware information in the competing messages received by the receiving module from other boards, and according to the competition The principle determines the winning board, wherein the competition principle is used to determine a board that is more suitable as the master node according to the comparison result of the hardware information;

The control module 64 is connected to the sending module 61, and is configured to control the sending module to stop sending the competition if the result of the determination is that the board of the first determining module 61 fails.

The second determining module 65 is connected to the sending module 61 and the receiving module 62, and is configured to determine one of the other optional boards as the standby node when the board is the master node.

The hardware information includes the physical location information of the board and the primary and backup status information of the board in the chassis. The primary and backup status information is used to indicate that the board is the main control board or the controller in the chassis. Board.

Moreover, the board according to the embodiment of the present invention may further include: a notification module (not shown), configured to notify other boards to report respective hardware information after the board is determined to be the master node;

Moreover, the second determining module 65 is specifically configured to determine, as the standby control node, an optional board that is located in a different chassis of the master node according to the hardware information reported by the other board.

In addition, the sending module 61 can also send the content in a broadcast or multicast manner, and can implement the sending of the keep-alive message, and the receiving module 62 can also receive the keep-alive messages of other boards and the reported load information and hardware. information. The transmission of the service node allocation request may be performed by the sending module 62 or by the notification module. If necessary, the notification module and the sending module 62 may be set in one.

Moreover, the selection of the primary service node and the selection of the standby service node may be performed by the first determining module 63 and the second determining module 65, respectively, or may be performed by one of them. If necessary, the first determining module 63 and The second determining module 65 is set in one.

Moreover, the control module can also be configured to notify the sending module to stop transmitting the competing message if it is determined that the number of failures of the card competition has reached a predetermined value.

The processing of the board shown in FIG. 6 in the process of transmitting the competitive message, the judgment of the winning, the selection of the standby control node, the selection of the active service node, the selection of the standby service node, and the monitoring of the master node are already in progress. Method embodiment It is described in connection with Example 1 and Example 2, and the specific process will not be repeated here.

In summary, with the above technical solution of the present invention, the advantages of the multi-chassis multi-control board of the cluster router system can be fully utilized, and the main control node, the standby control node, the main service node and the main service node are realized between multiple chassis. The determination of the standby service node allows the master node to run in any one of the frames, and can prevent the board in which the active/standby relationship is located in the same chassis, and effectively improve the performance of the master node and the active service node. System reliability and throughput; In addition, the monitoring functions of the master node and the standby node can be realized by means of hardware signals and logic signals.

The above description is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., which are included in the spirit and scope of the present invention, should be included in the present invention. Within the scope of protection.

Claims

Claim

A method for implementing a master/slave configuration of a card, configured to determine a master node and a standby node from a plurality of boards located in a plurality of chassis, wherein the method includes:

For each of the plurality of boards, each of the candidate boards sends a competition message to the other candidate board, wherein each of the candidate boards sends the contention message Hardware information of the optional board;

During the competition, each of the candidate boards receives a competition message from another candidate board;

Each of the candidate boards compares the hardware information with the hardware information in the received competition message, and determines an optional board that wins according to the competition principle, wherein the competition principle is used for Determining an optional board as a master node according to the comparison result of the hardware information;

A unique master node is determined according to the contention of the plurality of candidate boards, and the master node determines one of the other candidate boards as the standby node.

The implementation method according to claim 1, wherein the determining the unique master node according to the competition condition of the multiple candidate boards comprises:

Only one of the plurality of candidate boards that has never failed to be determined is the master node.

The implementation method of claim 1, further comprising:

If the candidate card fails, the candidate card stops sending the competition message to the other candidate cards.

The implementation method of claim 1, further comprising:

Setting a competition flag for the candidate card in advance, and setting the competition flag to true, the setting indicates that the standby card is allowed to become a master node by sending a competition message;

For each candidate board, if the candidate board determines that it has failed in the competition with any of the other optional boards, then the content flag of the candidate board is set to false. The setting is used to indicate that the candidate card is prohibited from being a master node by sending a competition message.

The implementation method according to claim 3, wherein the determining, by the only one of the plurality of candidate boards, the candidate card that has never failed to be determined as the master node comprises:

In the case where an optional board broadcasts or multicasts a number of competing messages to all other candidate cards for a predetermined number of thresholds, if the candidate card does not continue to receive competition from other candidate boards For the message, the candidate card is determined as the master node.

The implementation method according to claim 1, wherein the hardware information includes physical location information of the card, and primary standby state information of the board in the chassis, where the primary standby status information is used. Indicates that The board is the main control board or the standby control board in the chassis.

The implementation method according to claim 6, wherein the competition principle comprises:

It is preferred to determine the candidate board as the main control board in the respective chassis as the winning candidate board;

If the two candidate boards being compared are the main control boards in the respective chassis, the candidate boards with smaller or larger physical positions are determined to be winning in the two candidate boards for comparison. Optional board.

The implementation method according to claim 6, wherein the determining, by the master node, one of the other candidate cards as the standby node comprises:

The master node notifies the other candidate boards to report the respective hardware information, and determines an optional board that is located in a different chassis of the master node as the standby node according to the reported hardware information.

The method according to claim 8, wherein the determining, according to the reported hardware information, an option board that is located in a different chassis of the master node as the standby node further includes:

If there are a plurality of candidate boards that are located in different chassis and are in the respective chassis as the main control board, randomly selecting one of the plurality of candidate boards is determined as The standby node, or an optional board that minimizes or maximizes the physical location of the plurality of candidate boards, is determined as the standby node.

The method according to claim 1, wherein after determining the master node and the standby node, the method further includes:

And the master control node sends a service allocation request to the main control board of the multiple chassis, and receives hardware information and load information of each main control board returned by the main control board of the multiple chassis;

The master control node determines the active service node and the standby service node from the main control board of the multiple chassis according to the hardware information and the load information reported by the main control board of the multiple chassis.

The implementation method according to claim 10, wherein the master control node controls the plurality of chassis according to hardware information and load information reported by the main control board of the multiple chassis The primary service node and the standby service node are determined in the board, including:

Determining, by the main control node, the main control board with the smallest load according to the load information as the main service node, and excluding the main control node, the standby control node, and the main It is determined as a standby service node by using one of the main control boards other than the service node.

The implementation method according to claim 11, wherein the board of the main control board except the main control node, the standby control node, and the main service node is determined in the candidate card. The standby service node includes: determining, as the standby service node, a main control board that is located in a different chassis from the active service node.

13. The implementation method according to claim 1, wherein in the determining the master node and the After the node is controlled, the method further includes:

The master node sends a keep-alive message to the standby node at a predetermined period;

If the master node meets the switching condition, the master node and the standby node are switched, where the switching condition includes: the standby node does not receive the A keep-alive message of the master node, and the standby node learns that the working state of the master node is abnormal through other boards in the same chassis as the master node.

The implementation method according to any one of claims 1 to 13, wherein the optional board comprises part or all of the main control board and the standby control board of the plurality of chassis .

15. A board, comprising:

a sending module, configured to send a competition message to another board, where the competition message carries hardware information of the board;

a receiving module, configured to receive a competition from another board in the process of sending the contention message to the other board by the sending module.

a first determining module, configured to compare hardware information of the board in which the hardware information of the board is received with the hardware information received by the receiving module from the other board, and determine the winning board according to the competition principle The competition principle is used to determine a board as a master node according to a comparison result of hardware information;

a control module, configured to control the sending module to stop sending a competition message if the result of the determination is that the board of the first determining module fails to compete;

The second determining module is configured to determine one of the other candidate boards as the standby node when the board is the master node.

The board according to claim 15, wherein the hardware information includes physical location information of the board, and primary standby status information of the board in the chassis, wherein the primary standby status information is used. Indicates that the board is the main control board or the standby control board in the same chassis.

17. The card of claim 15 or 16, further comprising:

The notification module is configured to notify other boards to report respective hardware information after the board is determined to be the master node;

And the second determining module is specifically configured to determine, as the standby control node, an optional board that is located in a different chassis of the master node according to the hardware information reported by the other board.