TWI633416B - Server fan control system and control method - Google Patents

Abstract

A server fan control system includes a first fan controller, a first fan controller, A fan, a first substrate management controller, and a second fan controller, a second fan, and a second substrate management controller disposed on the second node. Both the first and second substrate management controllers are communicatively connected to the first fan controller and the second fan controller. When the operation of the second substrate management controller is normal, the first and second substrate management controllers control the first and second fan controllers to drive the first and second fans, respectively. When the operation of the second substrate management controller is abnormal, the first substrate management controller simultaneously controls the first and second fan controllers to drive the first and second fans.

Description

Server fan control system and control method

The invention relates to a server fan control system and a control method thereof, in particular to a server fan control system having a plurality of nodes and a control method thereof.

In today's society that requires uninterrupted operations, server failures caused by component failures, human error, or system crashes can cause service interruptions and increase the cost of business operations. Therefore, uninterrupted availability is an important goal to reduce costs and management time. In the design of servers, high availability (HA; High Availability) has gradually become a trend, and the design of high availability is aimed at the possible hardware and software. When the error occurs, the server is designed to still maintain normal operation. The advantage of using HA is that it can avoid long-term hardware repair or system reset. Generally, in a multi-node server, as shown in the server fan control system 10 shown in FIG. 1, each of the nodes 1-1 to 1-N (node) is provided with a fan 3-1 for cooling. To 3-N, and each fan is directly controlled by the Baseboard Management Controller (BMC) on its node to output Pulse Width Modulation (PWM) signals, and the Baseboard Management Controller 2-1 controls the fans 3-1, and the board management controller 2-N controls the fan 3-N. Once the fan module on the node fails, the entire node must be replaced, which is quite costly and does not meet the design principles of HA. On the other hand, if an error occurs in the baseboard management controller on the node, the fan module on the node will also stop operating, which will cause the entire server to have a risk of insufficient heat dissipation and overheating.

Traditionally, in order to improve such problems, in the design of some multi-node servers, as shown in the server fan control system 20 shown in FIG. 2, the fans 3 are independently set outside the nodes 1-1 to 1-N, micro-processing The controller 4 compares the output pulse width modulation signals of the baseboard management controllers 2-1 to 2-N, and controls the fan 3 of the server with the signal of the fan rotation speed demander. However, this design method cannot provide proper fan speed according to the system status on each node, which reduces the overall cooling performance. In addition, when the baseboard management controller on one of the nodes fails, it cannot output pulse width modulation. When the signal is sent, the microprocessor 4 cannot adjust the required fan speed accordingly, which does not meet the design principles of HA. In order to achieve high availability (HA) of a multi-node server and simultaneously improve the heat dissipation of traditional multi-node servers, the field of multi-node servers still needs a method that can improve the cooling effect when the baseboard management controller fails, and can provide various Fan control system for optimal cooling efficiency of the node.

A server fan control system according to an embodiment of the present invention includes a first fan controller disposed on a first node, a first fan, a first substrate management controller, and a second fan controller disposed on a second node. Two fans and a second baseboard management controller. The first substrate management controller has a first connection port and a second connection port, which are respectively connected to the first fan controller and the second fan controller. The second substrate management controller has a third port connected to the second fan controller. When the operation of the second substrate management controller is normal, the first substrate management controller generates a first control signal to control the first fan controller to drive the first fan through the first port, and the second substrate management controller generates the first The two control signals control the second fan controller to drive the second fan through the third port. When the operation of the second substrate management controller is abnormal, the first substrate management controller generates a first control signal via the first connection. The port controls the first fan controller to drive the first fan, and generates a second control signal to control the second fan controller to drive the second fan through the second port.

In some preferred embodiments of the present invention, the first substrate management controller is communicatively connected with the second substrate management controller. When the second baseboard management controller operates normally, the second baseboard management controller sends a heartbeat signal to the first baseboard management controller, and the first baseboard management controller further determines whether the second baseboard management controller operates abnormally according to the heartbeat signal. When the first baseboard management controller does not receive a heartbeat signal, the first baseboard management controller determines that the second baseboard management controller is operating abnormally, and generates a first control signal to control the first fan controller through the first port, And generating a second control signal to control the second fan controller through the second port.

In some preferred embodiments of the present invention, when the first baseboard management controller does not receive the heartbeat signal, the first baseboard management controller further sends a reset signal to the second baseboard management controller and the second baseboard management controller. Restart according to the reset signal. In addition, after the reset signal is transmitted by the first baseboard management controller, it is further determined whether the second baseboard management controller is operating normally according to the heartbeat signal. When the first baseboard management controller receives the heartbeat signal, the first baseboard management controller determines that the second baseboard management controller has been operating normally, and stops generating the second control signal.

In some preferred embodiments of the present invention, when the operations of the first and second substrate management controllers are normal, the first substrate management controller generates a first control signal according to the first system state of the first node, and the second The baseboard management controller generates a second control signal according to the second system state of the second node. The first baseboard management controller is in communication connection with the second baseboard management controller, and the second baseboard management controller further transmits the second system state of the second node to the first baseboard management controller. In addition, when the operation of the second baseboard management controller is abnormal, the first baseboard management controller generates a first control signal according to the first state of the first node, and generates a second control signal according to the second system state of the second node.

In some preferred embodiments of the present invention, the second baseboard management controller further has a fourth port for connecting to the first fan controller. When the operation of the first substrate management controller is abnormal, the second substrate management controller generates a first control signal to control the first fan controller via the fourth port, and generates a second control signal to control the first fan controller through the third port. The second fan controller.

A server fan control method according to an embodiment of the present invention is applicable to a server fan control system having a first node and a second node. The first node includes a first substrate management controller, a first fan controller, and a first fan, and the second node includes a second substrate management controller, a second fan controller, and a second fan. The first fan controller drives the first fan according to the first control signal, and the second fan controller drives the second fan according to the second control signal. The first substrate management controller is connected to the first fan controller through the first port. The first substrate management controller is connected to the second fan controller through the second connection port, and the second substrate management controller is connected to the second fan controller through the third connection port. The control method includes: when the first substrate management controller determines that the operation of the second substrate management controller is normal, generating the first control signal through the first substrate management controller and controlling the first through the first port. A first fan controller, and generating the second control signal through the second substrate management controller and controlling the second fan controller through the third port; and when the first substrate management controller determines the second When the operation of the substrate management controller is abnormal, the first control signal is generated through the first substrate management controller, and the first fan controller is controlled through the first port, and the above is generated through the first substrate management controller. The second control signal controls the second fan controller through the second connection port.

In some embodiments, the method for controlling a server fan further includes: when the first substrate management controller determines that the operation of the second substrate management controller is abnormal, transmitting a reset signal to the first substrate through the first substrate management controller. Two substrate management controllers; and when the above first substrate management The controller determines that the second substrate management controller is operating normally, and the first substrate management controller stops generating the second control signal.

10, 20, 100‧‧‧ server fan control system

1-1, 1-N, 110-1, 110-2‧‧‧ nodes

2-1, 2-N, 112-1, 112-2‧‧‧ substrate management controller

3, 3-1, 3-N, 116-1, 116-2‧‧‧ fans

4‧‧‧ Microprocessor

114-1, 114-2‧‧‧fan controller

120‧‧‧ back plate

P1, P2, P3, P4‧‧‧ ports

Other features and effects of the present invention will be clearly presented in the embodiments with reference to the drawings, in which:

FIG. 1 is a schematic diagram showing an embodiment of a conventional server fan control system.

FIG. 2 is a schematic diagram showing another embodiment of a conventional server fan control system.

FIG. 3 is a schematic diagram showing a server fan control system according to an embodiment of the present invention.

FIG. 4 is a flowchart showing the operation of the server fan control system according to the embodiment of FIG. 3 of the present invention.

The embodiments or examples shown in the drawings are expressed below in a specific manner. It should be understood that this embodiment or example is not intended to be limiting. Any replacement and modification of the embodiments of the present invention, and any further application of the principles of the present invention, can be completed by those skilled in the art in the field of the present invention with reference to the description of the present invention.

FIG. 3 is a schematic diagram showing a server fan control system 100 according to an embodiment of the present invention. As shown in FIG. 3, the server fan control system 100 includes nodes 110-1 and 110-2 and a backplane 120. The node 110-1 includes a baseboard management controller 112-1, a fan controller 114-1, and at least one fan 116-1. Similarly, the node 110-2 includes a baseboard management controller 112-2, a fan controller 114-2, and at least one fan 116-2. The baseboard management controller 112-1 is disposed in the node 110-1, and is mainly used for monitoring the components on the node 110-1, and has ports P1 and P2 for communication connection to the fan controllers 114-1 and 114-2, respectively. Similarly, the baseboard management controller 112-2 is set in the node 110-2, which is mainly used for monitoring the components on the node 110-2, and has ports P3 and P4 to communicate with the fan controllers 114-1 and 114, respectively. -2. Since the baseboard management controller 112-1 communicates with the fan controllers 114-1 and 114-2 through different ports P1 and P2 (for example, an integrated circuit bus (I ² C bus)), the baseboard management controller 112-1 can provide different control signals to the fan controllers 114-1 and 114-2. Similarly, the baseboard management controller 112-2 can also provide different control signals to the fan controllers 114-1 and 114- 2. In addition, the fan controller 114-1 is disposed in the node 110-1, and is used to provide a driving signal (for example, Pulse Width Modulation (Pulse Width Modulation; PWM)) to drive the fan 116-1, and the fan controller 114- 2 is provided in the node 110-2, and is used to provide a driving signal to drive the fan 116-2.

In some embodiments of the present invention, the baseboard management controller 112-1 of the node 110-1 and the baseboard management controller 112-2 of the node 110-2 are connected through a backplane communication. For example, the baseboard management controller may be Information is exchanged through a serial Gigabit Media Independent Interface (GSMII) using an Ethernet Physical Layer Transceiver (PHY). In addition, the baseboard management controller 112-1 and the baseboard management controller 112-2 can share and judge each other's system status (for example, node operating status, node temperature information, etc.), and instruct to restart (for example, send a reset signal). And determine whether or not each other operates normally (for example, a heartbeat signal). In some embodiments of the present invention, when the substrate management controllers 112-1 and 112-2 are powered on and operating normally, they will provide heartbeat signals to each other. Therefore, once the substrate management controller 112-1 does not receive the substrate management control Heartbeat signal provided by the controller 112-2, it can be determined that the substrate management controller 112-2 is abnormal, and vice versa with. On the other hand, by obtaining the system status from each other, the baseboard management controllers 112-1 and 112-2 can generate control signals to the fan controllers 114-1 and 114-2 respectively according to the system status of the respective nodes.

In some embodiments, when the baseboard management controller 112-1 and the baseboard management controller 112-2 are operating normally, the baseboard management controller 112-1 will generate a control signal according to the system state of the node 110-1 and pass through the port P1 is transmitted to the fan controller 114-1, and the fan controller 114-1 generates a corresponding driving signal according to the received control signal to drive the fan 116-1. For example, if the baseboard management controller 112-1 determines that the system state of the node 110-1 (for example, the node temperature is higher than a preset value) requires a faster fan speed, the baseboard management controller 112-1 transmits the corresponding A control signal is sent to the fan controller 114-1 to notify the fan controller 114-1 to drive the fan 116-1 to rotate at a higher speed. In addition, since the baseboard management controller 112-2 operates normally, the baseboard management controller 112-1 does not need to control the fan controller 114-2, so the baseboard management controller 112-1 does not generate a control signal on the port P2. Similarly, the baseboard management controller 112-2 generates a control signal according to the system state of the node 110-2 and transmits it to the fan controller 114-2 through the port P3, and the fan controller 114-2 according to the received control signal A corresponding driving signal is generated to drive the fan 116-2. In addition, since the baseboard management controller 112-1 operates normally and the baseboard management controller 112-2 does not need to control the fan controller 114-1, the baseboard management controller 112-2 does not generate a control signal on the port P4. It should be understood that the method of adjusting the fan speed can be designed according to the needs of the user. The fan can have different speeds according to the system status. Correspondingly, the system status can also include different system parameters according to the user's needs. (For example, the temperature of each element in the node, the load of the node operation, the work schedule, etc.), because the focus of the present invention is not on the adjustment of the fan speed, it will not be described in detail here, any selection method of the fan speed should include In the present invention.

In some embodiments of the present invention, when only one of the substrate management controller 112-1 and the substrate management controller 112-2 is operating normally, and the other is abnormal, the substrate management is operating normally. The controller can temporarily take over fan control on another node to avoid overheating and causing damage to the node. This is in line with the design principles of HA. Specifically, if the substrate management controller 112-1 determines that the substrate management controller 112-2 is abnormal, the substrate management controller 112-1 may send a reset signal to the substrate management controller 112-2 to notify the substrate management control. The controller 112-2 restarts. On the other hand, the baseboard management controller 112-1 transmits a control signal to the fan controller 114-2 through the port P2 to continue to drive the fan 116-2. Thus, the baseboard management controller 112- 2 During the occurrence of the abnormality and its restart, the fan 116-2 can continue to operate to ensure the heat dissipation effect of the electronic components on the node 110-2 or around it (for example, a hard disk disposed on the front side of the server). At the same time, the baseboard management controller 112-1 also continuously transmits a control signal to the fan controller 114-1 through the port P1. The fan controller 114-1 generates a corresponding driving signal according to the received control signal to drive the fan 116- 1.

In some embodiments of the present invention, since the substrate management controller 112-1 and the substrate management controller 112-2 continuously and periodically obtain the latest system status of each other, when the substrate management controller 112-1 temporarily takes over the management of the substrate When the fan of the controller 112-2 is controlled, the baseboard management controller 112-1 can further use the port according to the system status provided by the baseboard management controller 112-2 before the abnormality (ie, the system status of the node 110-2). P2 sends a control signal to the fan controller 114-2, and the fan controller 114-2 generates a corresponding driving signal to drive the fan 116-2 according to the received control signal. In some embodiments, the node 110-1 and the node 110-2 are in a mutually redundant system architecture. When the baseboard management controller 112-2 is operating normally and the baseboard management controller 112-1 is abnormal, the baseboard management controller 112-2 can be operated in the same manner as the above-mentioned baseboard management controller 112-1, and temporarily takes over the fan control of node 110-1 through port P4, which will not be repeated here.

The operation flow of the substrate management controller 112-1 in the server fan control system 100 of the present invention will be described below with reference to FIG. 4 and FIG. 3. In step S202, the substrate management controller 112-1 determines Whether the substrate management controller 112-2 is still operating normally. If the substrate management controller 112-2 is still operating normally, proceed to step S204; otherwise, if the substrate management controller 112-2 is abnormal, proceed to step S206. For example, the baseboard management controller 112-1 continuously or periodically determines whether a heartbeat signal transmitted by the baseboard management controller 112-2 has been received. If not, it is determined that the baseboard management controller 112-2 has occurred. If abnormal, proceed to step S206. On the contrary, if it is received, it is determined that the substrate management controller 112-2 is still operating normally, and then step S204 is continued.

In step S204, the baseboard management controller 112-1 transmits a control signal to the fan controller 114-1 through the port P1, and the fan controller 114-1 generates a driving signal to drive the fan 116- according to the control signal on the port P1. 1. At the same time, since the baseboard management controller 112-2 is still operating normally, the baseboard management controller 112-2 will also send a control signal to the fan controller 114-2 through the port P3, and the fan controller 114-2 is based on the port P3 The control signal generates a driving signal to drive the fan 116-2. In other words, in step S204, the fan controller 114-1 and the fan 116-1 on the node 110-1 are still controlled by the baseboard management controller 112-1, and the fan controller 114- on the node 110-2 2 and the fan 116-2 are still controlled by the baseboard management controller 112-2.

In addition, in step S206, the substrate management controller 112-1 transmits a reset signal to the substrate management controller 112-2, thereby restarting the substrate management controller 112-2. Next, in step S208, the baseboard management controller 112-1 transmits a control signal to the fan controller 114-2 through the port P2, and the fan controller 114-2 according to the baseboard management controller 112-1 on the port P2 The control signal generates a driving signal to drive the fan 116-2. At the same time, the baseboard management controller 112-1 also continuously transmits control signals to the fan controller 114-1 through the port P1, and the fan controller 114-1 continues to control signals on the port P1 according to the baseboard management controller 112-1. A driving signal is generated to drive the fan 116-1. In other words, at step S208 In the embodiment, the fan controller 114-1 and the fan 116-1 on the node 110-1 and the fan controller 114-2 and the fan 116-2 on the node 110-2 are controlled by the baseboard management controller 112-1.

Next, in step S210, the baseboard management controller 112-1 determines whether the baseboard management controller 112-2 has been restarted and resumes normal operation. If the baseboard management controller 112-2 can normally operate, step S212 is continued. Otherwise, it returns to step S208, and the baseboard management controller 112-1 continuously controls the fan controllers 114-1 and 114-2. For example, the baseboard management controller 112-1 continuously or periodically determines whether a heartbeat signal transmitted by the baseboard management controller 112-2 has been received, and if it has been received, it is determined that the baseboard management controller 112-2 has The restart is complete and normal operation resumes.

After the board management controller 112-2 has been restarted and resumes normal operation, in step S212, the board management controller 112-1 stops generating control signals on the port P2, and returns to step S204. In other words, the fan controller 114-2 and the fan 116-2 on the node 110-2 are controlled by the baseboard management controller 112-2. In some embodiments, the baseboard management controller 112-1 further sends a reply signal to the baseboard management controller 112-2 in step S212 to notify the baseboard management controller 112-2 to continue to control the fan controller 114-2.

It should be noted that under a mutually redundant architecture, the operation process of the substrate management controller 112-2 in the server fan control system 100 is also the same as the operation method of the above substrate management controller 112-1, which is not repeated here. . In summary, once the baseboard management controller in any node of the multi-node server fails, the baseboard management controller in other nodes can take over the fan of the failed node through the server fan control system of the present invention. The control can avoid the risk of data damage caused by system overheating and crashes and hard disk overheating, especially improves the heat dissipation when a fault occurs, reduces the risk of system overheating damage before troubleshooting, and realizes substrate management in a multi-node server system Controller for high availability of fan control. In addition, through the present invention, the fans of each node can still System status and control accordingly, providing better heat dissipation efficiency. Furthermore, in some embodiments of the present invention, the baseboard management controller communicates with each fan controller through the integrated circuit bus, so that the baseboard management controller can take over the failure in time when other baseboard management controller failures occur. The fan controllers and fans on the nodes do not need to undergo complex communication authentication.

The method of the present invention, or a specific form or part thereof, may exist in the form of a code. The code may be contained in physical media, such as a floppy disk, CD-ROM, hard disk, or any other machine-readable (such as computer-readable) storage medium, or is not limited to an external form of computer program product. When the code is loaded and executed by a machine, such as a computer, the machine becomes a device for participating in the invention. The code can also be transmitted through some transmission media, such as wire or cable, optical fiber, or any transmission type. Where the code is received, loaded, and executed by a machine, such as a computer, the machine becomes used to participate Invented device. When implemented in a general-purpose processing unit, the code in combination with the processing unit provides a unique device that operates similar to an application-specific logic circuit.

Although the preferred embodiments of the present invention have been described above, it should be understood that the above disclosure is not intended to limit the embodiments of the present invention. Instead, it encompasses multiple variations and similar configurations (as will be apparent to those skilled in the art). In addition, the scope of patent application attached should be interpreted in the broadest sense to include all the above-mentioned changes and similar configurations.

Claims (8)

A server fan control system includes: a first fan provided at a first node; a second fan provided at a second node; a first fan controller provided at the first node; A control signal drives the first fan; a second fan controller is provided at the second node, and the second fan is driven according to a second control signal; a first substrate management controller is provided at the first node, It has a first connection port and a second connection port for connecting to the first fan controller and the second fan controller, respectively; a second substrate management controller is disposed on the second node, and has a third A port for connecting to the second fan controller; and a backplane; wherein when the operation of the second substrate management controller is normal, the first substrate management controller is based on a first system of one of the first nodes The state generates the first control signal to control the first fan controller through the first port, and the second substrate management controller is based on the second One of the second system states generates the second control signal to control the second fan controller via the third port, wherein the first substrate management controller and the second substrate management controller communicate through the backplane. Connection, the second substrate management controller periodically transmits the second system state of the second node to the first substrate management controller, and when the operation of the second substrate management controller is abnormal, the first A baseboard management controller generates the first control signal to control the first fan controller through the first connection port, and generates the second control signal to control the second fan controller through the second connection port. According to the server fan control system described in the first item of the patent application scope, wherein the first substrate management controller is in communication with the second substrate management controller, and when the second substrate management controller is operating normally, the first The two baseboard management controllers transmit a heartbeat signal to the first baseboard management controller, and the first baseboard management controller further determines whether the second baseboard management controller operates abnormally according to the heartbeat signal, and when the first baseboard is When the management controller does not receive the heartbeat signal, the first substrate management controller determines that the second substrate management controller is operating abnormally, and generates the first control signal to control the first fan controller through the first port. And generating the second control signal to control the second fan controller through the second port. The server fan control system according to item 2 of the scope of patent application, wherein when the first substrate management controller does not receive the heartbeat signal, the first substrate management controller further sends a reset signal to the second The substrate management controller, and the second substrate management controller is restarted according to the reset signal. According to the server fan control system described in item 3 of the scope of patent application, wherein when the first substrate management controller transmits the reset signal, it is further determined whether the second substrate management controller has been normally operated according to the heartbeat signal, And when the first baseboard management controller receives the heartbeat signal, the first baseboard management controller determines that the second baseboard management controller has been operating normally and stops generating the second control signal. According to the server fan control system described in item 1 of the scope of patent application, when the operation of the second substrate management controller is abnormal, the first substrate management controller generates the above according to the first state of the first node. A first control signal, and generating the second control signal according to the second system state of the second node. According to the server fan control system described in item 1 of the scope of patent application, wherein the second substrate management controller further has a fourth port for connecting to the first fan controller, when the first substrate management control When the controller operates abnormally, the second substrate management controller generates the first control signal to control the first fan controller through the fourth port, and generates the second control signal to control the first fan controller through the third port. The second fan controller. A server fan control method is applicable to a server fan control system having a backplane, a first node, and a second node, wherein the first node includes a first substrate management controller and a first fan control. And a first fan, and the second node includes a second substrate management controller, a second fan controller, and a second fan. The first fan controller drives the first fan according to a first control signal. And the second fan controller drives the second fan according to a second control signal, the first substrate management controller is connected to the first fan controller through a first port, and the first substrate management controller passes through A second port is connected to the second fan controller, and the second substrate management controller is connected to the second fan controller through a third port. The first substrate management controller and the second substrate management The controller is connected through the backplane communication, and the server fan control method includes: the second substrate management controller periodically Send a second system state of one of the second nodes to the first substrate management controller; determine whether the second substrate management controller is operating normally through the first substrate management controller; and when the first substrate management controller judges the above When the second baseboard management controller operates normally, the first baseboard management controller generates the first control signal according to the first system state of one of the first nodes and controls the first fan control through the first port. And generates the second control signal according to the second system state of the second node through the second substrate management controller and controls the second fan controller through the third port; and when the first substrate When the management controller determines that the operation of the second substrate management controller is abnormal, the first control signal is generated through the first substrate management controller and the first fan controller is controlled through the first connection port, and the first fan controller is controlled through the first port. A baseboard management controller generates the second control signal and passes the second connection. The controller controls the second fan. The method for controlling a server fan according to item 7 of the scope of patent application, further comprising: when the first substrate management controller determines that the operation of the second substrate management controller is abnormal, transmitting a heavy packet through the first substrate management controller. A signal is set to the second substrate management controller; and when the first substrate management controller determines that the second substrate management controller is operating normally, the first substrate management controller stops generating the second control signal.