A multi-blade equipment management system and method based on a domestic controller

Abstract

The application relates to a multi-blade equipment management system and method based on a domestic controller, which comprises a mainboard controller arranged on each blade mainboard and a backboard controller arranged on an equipment backboard; the backboard controller is connected with each mainboard controller to build a monitoring management network; through the monitoring management network, the backboard controller collects information from each mainboard controller and packages and distributes the collected information to each blade mainboard controller, so that the mainboard CPU on each blade monitors the states of other blades except the blade; each mainboard controller sends the state monitoring result of the blade mainboard CPU according to the monitored states of each blade to the backboard controller, the backboard controller determines new power-on control instructions for each blade according to the state monitoring result of each blade, and sends the new power-on control instructions to each blade to update the power-on state. The application realizes the monitoring and management of each blade equipment and improves the reliability and safety of the mainboard.

Description

Technical Field

[0001] This invention belongs to the field of computer information technology, and specifically relates to a multi-blade device management system and method based on a domestically produced controller. Background Technology

[0002] With the development of information technology and the popularization of Internet applications, blade servers are increasingly widely used in various fields, especially in enterprises with high data-intensive requirements, where the monitoring and management of blade servers are extremely demanding. The Baseboard Management Controller (BMC) can remotely maintain and manage blade servers through a dedicated data channel, completely independent of the device's operating system, and can even perform remote monitoring and management while the server is powered off.

[0003] Currently, existing blade servers on the market generally use foreign-made dedicated BMC control chips to implement motherboard health management functions, which poses a security risk to server data. There is a lack of truly domestically produced BMC controllers in China.

[0004] To achieve true self-reliance and controllability in domestically produced servers, many Chinese companies have conducted in-depth research and development in this field, developing products that can replace those based on domestically produced BMC firmware. With the continuous development of self-reliant and controllable servers, major domestic server manufacturers are also using the Aspeed BMC chip + Kunlun BMC firmware solution from China Electronics Technology Group Corporation (CETC) in their domestically produced server products. However, the BMC hardware in this solution is still a foreign chip, and it has not achieved complete self-reliance and controllability. Furthermore, there are still no truly commercially viable BMC firmware products available to the market in China. Summary of the Invention

[0005] Based on the above analysis, the present invention aims to disclose a multi-blade device management system and method based on a domestically produced controller, which realizes the monitoring and management of each blade device and improves the reliability and security of the motherboard.

[0006] This invention discloses a multi-blade device management system based on a domestically produced controller, comprising: a motherboard controller mounted on each blade motherboard, and a backplane controller mounted on a device backplane including slots for inserting each blade; the backplane controller establishes a communication connection with each motherboard controller through the slots; and a monitoring and management network is constructed with the backplane controller as the master device and each motherboard controller as the slave device.

[0007] Through the monitoring and management network, the backplane controller obtains information from each motherboard controller, including the power-on status, temperature, and voltage of the blade motherboards, summarizes the information, and packages and distributes the summarized information to each blade motherboard controller. This enables the motherboard CPU on each blade to monitor the status of other blades besides its own. Each motherboard controller sends the status monitoring results generated by its blade motherboard CPU based on the monitored status of each blade to the backplane controller. The backplane controller then combines the status monitoring results of each blade to determine new power-on control commands for each blade and sends them to each blade to update its power-on status.

[0008] Furthermore, the chassis of the multi-blade device includes N sets of computing blades, M sets of GPU blades, 1 set of switching blades, 1 set of power blades, a backplane, and a rear panel;

[0009] Among them, the motherboards of N sets of computing blades, M sets of GPU blades, 1 set of switching blades and 1 set of power blades are all equipped with motherboard controllers, and the backplane is equipped with a backplane controller and a whole machine temperature sensor; the rear panel is equipped with S sets of whole machine fans, which are controlled by S fan boards for fan on / off and speed control, and the fan board controller is installed in the fan board.

[0010] The backplane controller is connected to the motherboard controllers of N sets of computing blades, M sets of GPU blades, 1 set of switching blades and 1 set of power blades, the fan board controllers of S fan boards, and the overall temperature sensor.

[0011] Furthermore, the backplane controller, each motherboard controller, and the fan board controller all use the domestically produced controller GD3232F450;

[0012] The backplane controller and each motherboard controller communicate with each other via two IPMI I2C interfaces, I2C_A and I2C_B.

[0013] The backplane controller is also connected to the overall temperature sensor and the fan controllers of the S fan boards via the controller's main I2C interface;

[0014] The backplane controller also sets the I2C address partitions for each blade according to the slot IDs of the backplane, and determines the type of each blade according to the I2C address partitions to perform automatic blade identification and hot-swapping functions; it also controls the power on / off and reset of the whole machine through the I / O interface.

[0015] Furthermore, both the computing blade and the GPU blade are equipped with a motherboard controller, a motherboard CPLD, and a motherboard CPU on their motherboards;

[0016] The motherboard controller communicates with the backplane controller via two IPMI I2C interfaces, I2C_A and I2C_B, on the outside of the blade; and communicates with the motherboard CPU via the communication interface provided by the motherboard CPLD on the inside of the blade.

[0017] The motherboard controller also obtains the power supply voltage and current data of the blade motherboard through the ADC pin, and obtains the temperature information of the blade motherboard temperature sensor through the I2C interface;

[0018] The motherboard CPLD provides a communication interface for communication between the motherboard controller and the motherboard CPU; executes instructions from the motherboard CPU to perform power management and power-down protection for the board, including power-on, power-off, and reset; and provides status indication and control and monitoring of the board's fans.

[0019] The motherboard CPU is used to monitor the status of other blade devices besides this blade based on the summary information received by the motherboard controller, generate status monitoring results based on the monitored status of each blade, and send them to the backplane controller through the communication path provided by the motherboard CPLD and the motherboard controller.

[0020] Furthermore, the motherboard CPLD includes:

[0021] The frequency divider module is used to provide a global clock to other modules in the motherboard CPLD.

[0022] The top-level module is used for data processing of other modules in the motherboard CPLD;

[0023] The power management module is used to control the power-on / off and reset timing.

[0024] The power control module is used to parse the power control signals sent by the motherboard CPU, and then provide the parsed results to the power management module to execute power actions.

[0025] I2C slave controller module, used to parse the I2C protocol;

[0026] The I2C slave register module is used to provide a data area for communication and interaction with other modules in the motherboard CPLD;

[0027] The fan PWM control module is used to provide PWM signals to control the fan speed of this board.

[0028] The fan speed monitoring module is used to detect the fan speed of this board.

[0029] Serial port data receiving module, used to receive serial port data;

[0030] The serial port data transmission module is used to send serial port data.

[0031] The serial port data protocol processing module is used to parse the serial port protocol.

[0032] This invention also discloses a management method for a multi-blade device management system based on a domestically produced controller, as described above, including...

[0033] Step S1: After the device is powered on, the backplane controller obtains information from each mainboard controller, including the power-on status, temperature and voltage of the blade mainboard, and summarizes the information.

[0034] Step S2: After the backplane controller aggregates and processes the data from each board, it distributes the aggregated information to each blade motherboard controller.

[0035] Step S3: The motherboard CPU on the compute blade and GPU blade monitors the status of other blade devices besides this blade based on the received summary information; and sends the status monitoring results generated by the motherboard CPU of this board based on the status of each blade to the backplane controller.

[0036] Step S4: The backplane controller, based on the status monitoring results of each blade, determines a new power-on control command for the blades in the equipment and sends it to the corresponding blades to update the power-on status.

[0037] Further, step S1 includes:

[0038] 1) After the entire machine is powered on, each motherboard CPLD sends its power-on control status and power-on control information to its motherboard controller via serial port.

[0039] 2) Each motherboard controller acquires the temperature and voltage data of its own board, and puts the received power-on information into the data exchange area of ​​the I2C slave device for access by the backplane controller.

[0040] 3) The backplane controller actively polls the I2C data exchange area of ​​each motherboard controller to obtain the temperature, voltage and power-on information of each blade server; it polls the fan board controller to obtain the speed of the whole machine fan; and it polls the whole machine temperature sensor to obtain the whole machine temperature.

[0041] 4) The backplane controller will aggregate and process the data obtained after polling all boards to obtain packaged data.

[0042] Further, step S2 includes:

[0043] 1) The backplane controller distributes the packaged data of all boards to each motherboard controller;

[0044] The packaged data includes the power-on control commands for each blade, the temperature data of each blade motherboard and backplate, the voltage data of each blade motherboard, the overall fan speed and status, the power-on information of each blade, and the position of the blade slot and the type of blade in place.

[0045] 2) After each motherboard controller obtains the packaged data of all the boards sent from the backplane, it sends it to the motherboard CPLD of its own board through the serial data port;

[0046] 3) After the CPLD of each motherboard obtains the data from all the boards, it organizes and saves it to the data exchange area of ​​the I2C slave device for access by the motherboard CPU of this board.

[0047] Furthermore, step S3 includes:

[0048] 1) The mainboard CPU of each blade generates status monitoring results based on the status of each blade;

[0049] The motherboard CPU analyzes the status of each blade. When it detects abnormal temperature, abnormal voltage, or abnormal fan speed of a blade, it outputs the corresponding status monitoring results or outputs the corresponding status monitoring results according to the power requirements of the board. The status monitoring results include information on power-on / off and reset measures taken for other blades or the whole machine fan.

[0050] 2) The CPU of each blade writes the status monitoring results to the data exchange area of ​​the I2C slave device of the motherboard CPLD;

[0051] 3) The motherboard CPLD sends the written status monitoring results to the motherboard controller of this board through the serial data port; the motherboard controller puts the status monitoring results into the data exchange area of ​​the I2C slave device;

[0052] 4) The backplane controller actively polls the I2C data exchange area of ​​each motherboard controller to obtain the status monitoring results of each blade.

[0053] Furthermore, step S4 includes:

[0054] 1) The backplane controller, based on the status monitoring results of each blade and the preset selection conditions, determines the new power-on control command for the blades in the equipment.

[0055] 2) The backplane controller distributes new power-on control commands to each motherboard controller via packaged data;

[0056] 3) After receiving the packaged data from the backplane, each motherboard controller sends it to the motherboard CPLD of its own board through the serial data port;

[0057] 4) After each motherboard CPLD obtains the packaged data, it organizes and saves it to the data exchange area of ​​the I2C slave device. The motherboard CPU of this board accesses the new power-on control instructions in the packaged data and updates the power-on status of this board.

[0058] This invention can achieve one of the following beneficial effects:

[0059] This invention discloses a multi-blade device management system and method based on a domestically produced controller, featuring a modular and layered hardware design. This enables configurable and scalable blades, combined with software design.

[0060] The system of the present invention is composed of the same I2C circuit, with master-slave adaptive function. It sets its own address according to the slot ID of the backplane, determines and distinguishes the blade type of the module, and can automatically identify and realize hot-swapping.

[0061] The system of the present invention is designed with two physical paths, I2C_A and I2C_B. When one communication device fails, it can be switched to another device, which improves the overall robustness and fault tolerance of the system.

[0062] The system of this invention features a relatively simple BMC circuit design. It utilizes domestically produced GigaDevice chips with built-in RAM and flash memory, requiring fewer external components and resulting in lower chip cost and power consumption.

[0063] The system of this invention is feature-rich and flexible in use. It can control or collect data on the power on / off status of multiple server blades, multiple status information, multiple temperature channels, multiple voltage and current channels, and multiple fan speeds according to actual needs, thereby ensuring the safe and reliable operation of the entire system.

[0064] Furthermore, the hardware of this invention uses domestically produced controllers and related components, while the software independently defines the transmission data according to the standard IPMB protocol, making it independently controllable. Attached Figure Description

[0065] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0066] Figure 1 This is a schematic block diagram of a multi-blade device management system according to an embodiment of the present invention;

[0067] Figure 2 This is a schematic block diagram of a specific multi-blade device management system in an embodiment of the present invention;

[0068] Figure 3 This is a schematic diagram showing the composition and connection of the motherboard CPLD in an embodiment of the present invention;

[0069] Figure 4 This is a flowchart of the BMC management method in an embodiment of the present invention. Detailed Implementation

[0070] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and, together with the embodiments of the present invention, serve to illustrate the principles of the present invention.

[0071] Example 1

[0072] One embodiment of the present invention discloses a multi-blade device management system based on a domestically produced controller, such as... Figure 1 As shown, the system includes a motherboard controller mounted on each blade motherboard and a backplane controller mounted on a device backplane that includes slots for inserting each blade. The backplane controller establishes a communication connection with each motherboard controller through the slots. A monitoring and management network is constructed with the backplane controller as the master device and each motherboard controller as the slave device.

[0073] Through the monitoring and management network, the backplane controller obtains information from each motherboard controller, including the power-on status, temperature, and voltage of the blade motherboards, summarizes the information, and packages and distributes the summarized information to each blade motherboard controller. This enables the motherboard CPU on each blade to monitor the status of other blades besides its own. Each motherboard controller sends the status monitoring results generated by its blade motherboard CPU based on the status of each blade to the backplane controller. The backplane controller then combines the status monitoring results of each blade to determine new power-on control commands for each blade and sends them to each blade to update its power-on status.

[0074] In one specific solution of this embodiment, such as Figure 2 As shown, the chassis of the multi-blade device includes N=3 sets of computing blades, M=1 set of GPU blades, 1 set of switching blades, 1 set of power blades, a heat dissipation module, a backplate, and a front and rear panel.

[0075] The blade motherboards of the three compute blades, one GPU blade, one switching blade, and one power blade are all equipped with motherboard controllers. The backplane is equipped with a backplane controller and a system temperature sensor. The backplane is equipped with S=2 system fans, which are controlled by two fan boards for fan power-on / off and speed control. The fan boards are also equipped with fan board controllers.

[0076] The backplane controller is connected to the motherboard controllers of 3 sets of compute blades, 1 set of GPU blades, 1 set of switching blades and 1 set of power blades, the fan controllers of 2 fan boards and the overall temperature sensor.

[0077] Specifically, the backplane controller, each motherboard controller, and the fan board controller all use the domestically produced controller GD3232F450;

[0078] The backplane controller and each mainboard controller communicate via two IPMI I2C interfaces, I2C_A and I2C_B. If one communication device fails, it can be switched to another, improving the overall robustness and fault tolerance of the system.

[0079] The backplane controller is also connected to the overall temperature sensor on the backplane via two IPMI I2C interfaces, I2C_A and I2C_B, to collect the overall temperature, and to the fan controllers of the two fan boards to control the fan speed and detect the fan status.

[0080] The backplane controller also sets the I2C address partitions for each blade according to the slot IDs of the backplane, and determines the type of each blade server according to the I2C address partitions, and performs automatic blade identification and hot-swapping functions; it also controls the power on / off and reset of the whole machine through the I / O interface.

[0081] The backplane controller also controls the power on / off of the entire machine and the status of each indicator light.

[0082] The backplane controller also displays BMC information to the user through a gigabit management network port; or the user can transmit power on / off, reset, or information on various functional nodes to the backplane controller through the network port, and the backplane controller will then transmit the commands to each blade controller node to execute the response operation.

[0083] Based on the above introduction, the backplane controller operates as follows:

[0084] 1) Collect information about the board, including power-on status, temperature, and voltage of each blade motherboard, via the IPMB I2C interface;

[0085] 2) Collect the ACPI (Advanced Configuration Power Interface) status of each blade server controller through the IPMB I2C interface to realize power-on management, power-on, reset, power-off and other operations of each blade motherboard;

[0086] 3) Obtain backplane temperature sensor data through the backplane controller's main I2C, and control the operation and detection of the fan speed of the fan control board by controlling the power supply of the fan control board.

[0087] 4) Identify the online or offline status of each blade server, as well as the type of blade server, through hardware pin identification;

[0088] 5) Control the power on / off of the entire machine and the status of each indicator light;

[0089] 6) The gigabit management port is displayed to the user.

[0090] Specifically, the operation of the fan controller set on the two fan plates includes:

[0091] 1) Obtain fan speed and control fan speed;

[0092] 2) As an I2C slave interface, it collects or uploads fan sensor information;

[0093] Specifically, the operation of the motherboard controller in the blade swapping setup includes:

[0094] 1) The IPMB I2C switch acts as an I2C slave device, acquiring and controlling the health information of the switch board;

[0095] 2) The switching board uses the IPMB interface to obtain the temperature sensing information of the board.

[0096] Specifically, the motherboard of one of the three sets of computing blades, the GPU blade set, is equipped with a motherboard controller, a motherboard CPLD, and a motherboard CPU.

[0097] The motherboard controller communicates with the backplane controller via two IPMI I2C interfaces, I2C_A and I2C_B, on the outside of the blade; and communicates with the motherboard controller and the motherboard CPU via the communication interface provided by the motherboard CPLD on the inside of the blade.

[0098] The motherboard controller also obtains data such as the power supply voltage and current of the blade motherboard through the ADC pin, and obtains the temperature information of the blade motherboard temperature sensor through the controller's main I2C.

[0099] The motherboard CPLD provides a communication interface for communication between the motherboard controller and the motherboard CPU; executes instructions from the motherboard CPU to perform power management and power-down protection for the board, including power-on, power-off, and reset; and provides status indication and control and monitoring of the board's fans.

[0100] Specifically, the motherboard CPLD provides a serial communication interface to connect with the motherboard controller; the motherboard CPLD provides an I2C interface to connect with the motherboard CPU.

[0101] The motherboard CPU is used to monitor the status of other blade devices besides this blade based on the summary information received by the motherboard controller. It generates status monitoring results based on the status of each blade and sends them to the backplane controller through the communication path provided by the motherboard CPLD and the motherboard controller.

[0102] Based on the above introduction, the motherboard controller of the computing board and GPU board operates as follows:

[0103] 1) Obtain the temperature information of the blade motherboard temperature sensor through the main I2C of the motherboard controller;

[0104] 2) Obtain data such as power supply voltage and current of this blade motherboard through the ADC pin;

[0105] 3) The IPMB I2C interface is used as a slave interface to obtain data such as temperature, power supply voltage and current, single board ACPI status, and board SN number from the backplane controller. The software slot identification changes the I2C slave address.

[0106] 4) The controller interacts with the server CPU through the UART serial port. The CPU can obtain the data collected by the controller and send status or control signals to the backplane controller through this link.

[0107] More specifically, such as Figure 3 As shown, the motherboard CPLD includes:

[0108] 1: Frequency divider module (CLK_PRO module), used to provide a global clock to other modules in the motherboard CPLD;

[0109] 2: Top-level module (CPLD_TOP module), used for data processing of other modules in the motherboard CPLD;

[0110] 3: Power management module (POWER_CTRL module), used to control power-on / off and reset timing;

[0111] 4: Power control module (PWR_CTRL module), used to parse the power control signals sent by the motherboard CPU, and then provide the parsed results to the power management module to execute power actions;

[0112] 5: I2C slave controller module (I2C_CONTROL module), used for parsing the I2C protocol;

[0113] 6: I2C Slave Register Module (I2C_REG Module), used to provide a data area for communication and interaction with other modules in the motherboard CPLD;

[0114] 7: The fan PWM control module FAN_PWM is the fan PWM control module;

[0115] 8: The fan speed monitoring module FAN_TACH is the fan speed monitoring module;

[0116] 9: Watch_dog is the watchdog module;

[0117] 10: LED is the indicator light module, used to control the status of all indicator lights on the board;

[0118] 11: Serial data receiving module (UART_RX module), used to receive serial data;

[0119] 12: Serial data transmission module (UART_TX module), used to send serial data;

[0120] 13: Serial port data protocol processing module (UART_RW_CTRL module), used to parse serial port protocols.

[0121] The backplane business data communication supports the on-site identification of each functional module, monitoring of operating status information such as voltage, power consumption, and temperature, as well as alarms and location of faults such as overvoltage, overcurrent, and overtemperature. It supports remote power-on, power-off, and reset functions; and supports viewing information of each functional node through a remote network interface to achieve system debugging and diagnosis.

[0122] This embodiment also discloses a management method based on the multi-blade device management system using a domestically produced controller, as described above. Figure 4 As shown, it includes:

[0123] Step S1: After the device is powered on, the backplane controller obtains information from each mainboard controller, including the power-on status, temperature and voltage of the blade mainboard, and summarizes the information.

[0124] Step S2: After the backplane controller aggregates and processes the data from each board, it distributes the aggregated information to each blade motherboard controller.

[0125] Step S3: The motherboard CPU on the compute blade and GPU blade monitors the status of other blade devices besides this blade based on the received summary information; and sends the status monitoring results generated by the motherboard CPU of this board based on the status of each blade to the backplane controller.

[0126] Step S4: The backplane controller, based on the status monitoring results of each blade, determines a new power-on control command for the blades in the equipment and sends it to the corresponding blades to update the power-on status.

[0127] Specifically, step S1 includes:

[0128] 1) After the entire machine is powered on, each motherboard CPLD sends its power-on control status and power-on control information to its motherboard controller via serial port.

[0129] 2) Each motherboard controller acquires the temperature and voltage data of its own board, and puts the received power-on information into the data exchange area of ​​the I2C slave device for access by the backplane controller.

[0130] 3) The backplane controller actively polls the I2C data exchange area of ​​each motherboard controller to obtain the temperature, voltage and power-on information of each blade server; it polls the fan board controller to obtain the speed of the whole machine fan; and it polls the whole machine temperature sensor to obtain the whole machine temperature.

[0131] 4) The backplane controller will aggregate and process the data obtained after polling all boards to obtain packaged data.

[0132] Specifically, step S2 includes:

[0133] 1) The backplane controller distributes the packaged data of all boards to each motherboard controller;

[0134] The packaged data includes the power-on control commands for each blade, the temperature data of each blade motherboard and backplate, the voltage data of each blade motherboard, the overall fan speed and status, the power-on information of each blade, and the position of the blade slot and the type of blade in place.

[0135] 2) After each motherboard controller obtains the packaged data of all the boards sent from the backplane, it sends it to the motherboard CPLD of its own board through the serial data port;

[0136] 3) After the CPLD of each motherboard obtains the data from all the boards, it organizes and saves it to the data exchange area of ​​the I2C slave device for access by the motherboard CPU of this board.

[0137] Specifically, step S3 includes:

[0138] 1) The mainboard CPU of each blade generates status monitoring results based on the status of each blade;

[0139] The motherboard CPU analyzes the status of each blade. When it detects abnormal temperature, abnormal voltage, or abnormal fan speed of a blade, it outputs the corresponding status monitoring results or outputs the corresponding status monitoring results according to the power requirements of the board. The status monitoring results include information on power-on / off and reset measures taken for other blades or the whole machine fan.

[0140] The statement that the status monitoring results are output according to the power requirements of this board refers to the status monitoring results that will request other blades to be put into standby or turned off when one of the blades needs more power during operation.

[0141] 2) The CPU of each blade writes the status monitoring results to the data exchange area of ​​the I2C slave device of the motherboard CPLD;

[0142] 3) The motherboard CPLD sends the written status monitoring results to the motherboard controller of this board through the serial data port; the motherboard controller puts the status monitoring results into the data exchange area of ​​the I2C slave device;

[0143] 4) The backplane controller actively polls the I2C data exchange area of ​​each motherboard controller to obtain the status monitoring results of each blade.

[0144] Specifically, step S4 includes:

[0145] 1) The backplane controller, based on the status monitoring results of each blade and the preset selection conditions, determines the new power-on control command for the blades in the equipment.

[0146] The pre-set selection conditions include: setting priorities for 3 groups of computing blades and 1 group of GPU blades, and determining new power-on control commands for the blades in the device based on the status monitoring results of the blades with higher priority.

[0147] 2) The backplane controller distributes new power-on control commands to each motherboard controller via packaged data;

[0148] 3) After receiving the packaged data from the backplane, each motherboard controller sends it to the motherboard CPLD of its own board through the serial data port;

[0149] 4) After each motherboard CPLD obtains the packaged data, it organizes and saves it to the data exchange area of ​​the I2C slave device. The motherboard CPU of this board accesses the new power-on control instructions in the packaged data and updates the power-on status of this board.

[0150] In summary, the multi-blade device management system and method based on a domestically produced controller in this embodiment of the invention has the following advantages:

[0151] 1) Modular and layered hardware design. This enables configurable and scalable blades, combined with software design.

[0152] 2) The system is composed of the same I2C circuit, with master-slave adaptive function. It sets its own address according to the slot ID of the backplane, determines and distinguishes the blade type of the module, and can automatically identify and realize hot-swapping.

[0153] 3) The system is designed with two physical paths, I2C_A and I2C_B. When one communication device fails, it can be switched to another device, which improves the overall robustness and fault tolerance of the system.

[0154] 4) The system's BMC circuit design is relatively simple. It uses domestically produced GigaDevice chips with built-in RAM and flash memory, requiring fewer external components and resulting in lower chip cost and power consumption.

[0155] 5) The system is feature-rich and flexible in use. It can control or collect data on the power on / off status of multiple server blades, multiple status information, multiple temperature channels, multiple voltage and current channels, and multiple fan speeds according to actual needs, thereby ensuring the safe and reliable operation of the entire system.

[0156] 6) The hardware of this invention uses domestically produced controllers and related components, and the software independently defines the transmission data according to the standard IPMB protocol, making it independently controllable.

[0157] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-blade device management system based on a domestically produced controller, characterized in that, include: A motherboard controller is installed on each blade motherboard, and a backplane controller is installed on a device backplane that includes slots for inserting each blade. The backplane controller establishes a communication connection with each motherboard controller through slots. A monitoring and management network is constructed with the backplane controller as the main device and each motherboard controller as a slave device. The chassis of the multi-blade device includes N sets of computing blades, M sets of GPU blades, 1 set of switching blades, 1 set of power blades, a backplane, and a rear panel; wherein, the motherboards of the N sets of computing blades, M sets of GPU blades, 1 set of switching blades, and 1 set of power blades are all equipped with motherboard controllers, and the motherboards of the computing blades and GPU blades are all equipped with motherboard CPLD and motherboard CPU. The motherboard CPLD provides a communication interface for communication between the motherboard controller and the motherboard CPU; executes instructions from the motherboard CPU to perform power management and power-down protection for the board, including power-on, power-off, and reset; and provides status indication and control and monitoring of the board's fans. The back panel is equipped with a back panel controller and a whole-machine temperature sensor; the rear panel is equipped with S groups of whole-machine fans, each controlled by an S fan board for fan on / off and speed control, and each fan board has a fan board controller. The backplane controller is connected to the motherboard controllers of N sets of computing blades, M sets of GPU blades, 1 set of switching blades and 1 set of power blades, the fan board controllers of S fan boards, and the overall temperature sensor. Through the monitoring and management network, the backplane controller obtains information from each motherboard controller, including the power-on status, temperature and voltage of the blade motherboard, summarizes the information, and packages and distributes the summarized information to each blade motherboard controller, so that the motherboard CPU on each blade can monitor the status of other blades besides its own blade. Each motherboard controller sends the status monitoring results generated by the CPU of its blade motherboard based on the monitored status of each blade to the backplane controller; the status monitoring results include information on power-on and power-off and reset measures taken for other blades or the whole machine fan; The backplane controller, based on the status monitoring results of each blade and according to the preset selection conditions, determines a new power-on control command for each blade and sends it to each blade to update its power-on status. The pre-set selection conditions include: setting priorities for computing blades and GPU blades, and determining new power-on control commands for the blades in the device based on the status monitoring results of the blades with higher priority.

2. The multi-blade equipment management system according to claim 1, characterized in that, The backplane controller, each motherboard controller, and the fan board controller all use the domestically produced controller GD3232F450. The backplane controller and each motherboard controller communicate with each other via two IPMI I2C interfaces, I2C_A and I2C_B. The backplane controller also connects to the system temperature sensor and... via the controller's main I2C interface. S Connect the fan controller to the fan board; The backplane controller also sets the I2C address partitions for each blade according to the slot IDs of the backplane, and determines the type of each blade according to the I2C address partitions to perform automatic blade identification and hot-swapping functions; it also controls the power on / off and reset of the whole machine through the I / O interface.

3. The multi-blade equipment management system according to claim 2, characterized in that, The motherboard controller on the motherboard of the computing blade and the GPU blade communicates with the backplane controller outside the blade through two IPMI I2C interfaces, I2C_A and I2C_B; and communicates with the motherboard CPU inside the blade through the communication interface provided by the motherboard CPLD. The motherboard controller also obtains the power supply voltage and current data of the blade motherboard through the ADC pin, and obtains the temperature information of the blade motherboard temperature sensor through the I2C interface; The motherboard CPU on the motherboard is used to monitor the status of other blade devices besides this blade based on the summary information received by the motherboard controller, generate status monitoring results based on the monitored status of each blade, and send them to the backplane controller through the communication path provided by the motherboard CPLD and the motherboard controller.

4. The multi-blade equipment management system according to claim 3, characterized in that, The motherboard CPLD includes: The frequency divider module is used to provide a global clock to other modules in the motherboard CPLD. The top-level module is used for data processing of other modules in the motherboard CPLD; The power management module is used to control the power-on / off and reset timing. The power control module is used to parse the power control signals sent by the motherboard CPU, and then provide the parsed results to the power management module to execute power actions. I2C slave controller module, used to parse the I2C protocol; The I2C slave register module is used to provide a data area for communication and interaction with other modules in the motherboard CPLD; The fan PWM control module is used to provide PWM signals to control the fan speed of this board. The fan speed monitoring module is used to detect the fan speed of this board. Serial port data receiving module, used to receive serial port data; The serial port data transmission module is used to send serial port data. The serial port data protocol processing module is used to parse the serial port protocol.

5. A management method for a multi-blade device management system based on a domestically produced controller according to any one of claims 1-4, comprising: Step S1: After the device is powered on, the backplane controller obtains information from each mainboard controller, including the power-on status, temperature and voltage of the blade mainboard, and summarizes the information. Step S2: After the backplane controller aggregates and processes the data from each board, it distributes the aggregated information to each blade motherboard controller. Step S3: The motherboard CPU on the compute blade and GPU blade monitors the status of other blade devices besides this blade based on the received summary information. The mainboard CPU of this board will generate status monitoring results based on the status of each blade and send them to the backplane controller. Step S4: The backplane controller, based on the status monitoring results of each blade, determines a new power-on control command for the blades in the equipment and sends it to the corresponding blades to update the power-on status.

6. The management method according to claim 5, characterized in that, Step S1 includes: 1) After the entire machine is powered on, each motherboard CPLD sends its power-on control status and power-on control information to its motherboard controller via serial port. 2) Each motherboard controller acquires the temperature and voltage data of its own board, and puts the received power-on information into the data exchange area of ​​the I2C slave device for access by the backplane controller; 3) The backplane controller actively polls the I2C data exchange area of ​​each motherboard controller to obtain the temperature, voltage and power-on information of each blade server; it polls the fan board controller to obtain the speed of the whole machine fan; and it polls the whole machine temperature sensor to obtain the whole machine temperature. 4) The backplane controller will aggregate and process the data obtained after polling all boards to obtain packaged data.

7. The management method according to claim 5, characterized in that, Step S2 includes: 1) The backplane controller distributes the packaged data of all boards to each motherboard controller; The packaged data includes the power-on control commands for each blade, the temperature data of each blade motherboard and backplate, the voltage data of each blade motherboard, the overall fan speed and status, the power-on information of each blade, and the position of the blade slot and the type of blade in place. 2) After each motherboard controller receives the packaged data of all the boards sent from the backplane, it sends it to the motherboard CPLD of its own board through the serial data port; 3) After the CPLD of each motherboard obtains the data from all the boards, it organizes and saves it to the data exchange area of ​​the I2C slave device for access by the motherboard CPU of this board.

8. The management method according to claim 5, characterized in that, Step S3 includes: 1) The mainboard CPU of each blade generates status monitoring results based on the status of each blade; The motherboard CPU analyzes the status of each blade. When it detects abnormal temperature, abnormal voltage, or abnormal fan speed of a blade, it outputs the corresponding status monitoring results or outputs the corresponding status monitoring results according to the power requirements of the board. The status monitoring results include information on power-on / off and reset measures taken for other blades or the whole machine fan. 2) The CPU of each blade writes the status monitoring results to the data exchange area of ​​the I2C slave device of the motherboard CPLD; 3) The motherboard CPLD sends the written status monitoring results to the motherboard controller of this board through the serial data port; the motherboard controller puts the status monitoring results into the data exchange area of ​​the I2C slave device; 4) The backplane controller actively polls the I2C data exchange area of ​​each motherboard controller to obtain the status monitoring results of each blade.

9. The management method according to claim 5, characterized in that, Step S4 includes: 1) The backplane controller, based on the status monitoring results of each blade and according to the preset selection conditions, determines the new power-on control command for the blades in the equipment; 2) The backplane controller distributes new power-on control commands to each motherboard controller via packaged data; 3) After receiving the packaged data from the backplane, each motherboard controller sends it to the motherboard CPLD on its own board via the serial data port; 4) After each motherboard CPLD obtains the packaged data, it organizes and saves it to the data exchange area of ​​the I2C slave device. The motherboard CPU of this board accesses the new power-on control instructions in the packaged data and updates the power-on status of this board.

Images