TW201327144A - Method for managing cloud server system - Google Patents

Abstract

A method for managing a cloud server system is provided. In the method, a plurality of node apparatuses of the cloud server system are abnormal is detected. When one of the node apparatuses is abnormal, a hardware address of the abnormal node apparatus is obtained. A location information of the abnormal node apparatus is searched from a node database according to the hardware address. And the abnormal node apparatus is isolated from the cloud server system. Moreover, a light emitting unit of the abnormal node apparatus is enable according to the location information.

Description

Cloud servo system management method

The present invention relates to a cloud servo system, and more particularly to a method for managing a cloud servo system that utilizes a light unit to indicate an abnormality.

At present, the server is widely used by various enterprises. The development scope is not only combined with the application of the internet and the telecom industry, but also deepens into the lives of ordinary people, such as finance, finance, online banking, and the Internet. The use of credit cards, etc., all rely on the powerful computing power of the server to achieve a high degree of confidentiality and difficulty in being cracked.

There are many types of cloud-based servo systems today. The more common ones are rack servers and tower servers. Among them, the rack server is an optimized structure of the tower server, which is designed to minimize space occupation. Many professional network devices are rack-mounted (such as switches, routers, hardware firewalls, etc.), which are mostly flat, just like drawers. In general, the rack server has a width of 19 inches and a height in U (1U = 1.75 inches = 44.45 mm). There are usually 1U, 2U, 3U, 4U, 5U, 7U standard servers.

At present, the node devices in the cabinet generally have a High Availability (HA) function, which provides redundant fault-tolerant backup. After one node device fails, it can immediately take over related resources and continue to provide corresponding services. When a node needs to replace another node device or upgrade the original node device due to an abnormality of its node device or other reasons, since the replacement process is artificial, when the number of node devices is large, it is required to be numerous. It is quite difficult to find the node device to be replaced in the node device, and the node needs to be automatically added to the cloud servo system after replacing the node.

The invention provides a management method of a cloud servo system, which is convenient for a manager to know a node device to be maintained through a prompt of a light emitting unit.

Specifically, the present invention provides a cloud server system management method, which is applicable to a cloud server system, for example, a data center (Container) that provides IaaS (Infrastructure as a Service) service, wherein the cloud server system Includes multiple node devices. In the method, it is detected whether an abnormality has occurred in these node devices. When an abnormality is detected in one of the node devices, the first hardware address of the abnormal node device is obtained. Then, according to the first hardware address, the node related information of the abnormal node device is searched from the node database. Here, the location information of the abnormal node device is recorded in the node related information. Moreover, the abnormal node device is isolated from the cloud servo system. In addition, the light emitting unit of the abnormal node device is enabled according to the position information.

In an embodiment of the present invention, the location information is a node location of the abnormal node device located in the cabinet or a network address of the abnormal node device.

In an embodiment of the present invention, the step of enabling the illumination unit of the abnormal node device according to the location information may transmit a command to the Baseboard Management Controller (BMC) of the abnormal node device according to the location information. The light emitting unit is illuminated to the first color by the substrate management controller.

In an embodiment of the present invention, in the management method of the cloud server system, after detecting that the abnormal node device is replaced with another node device, the light-emitting unit of the node device after the replacement is lit is the second color.

In an embodiment of the present invention, after detecting that the abnormal node device is replaced with another node device, the node related information of the node device is retrieved through the network management module. Further, the second hardware address can be received from the replaced node device to reallocate the network address to the node device. Then, an instruction is transmitted to the node device according to the network address to obtain the node related information of the node device through the base device management controller of the node device. And, update the node related information to the node database.

In an embodiment of the present invention, in the step of obtaining the node related information of the node device by the baseboard management controller of the transmitting node device, when the baseboard management controller receives the command, the central processing unit of the node device is restarted. In the process of rebooting, the node related information is obtained through the substrate management controller.

Based on the above, the present invention utilizes the light-emitting unit to remind the manager where the abnormal node device is located, facilitating the manager to replace the node device.

The above described features and advantages of the present invention will be more apparent from the following description.

1 is a block diagram of a cloud servo system in accordance with an embodiment of the present invention. In this embodiment, the cloud server system includes at least one cabinet. Since the cabinets have the same composition, for convenience of description, in the embodiment, one cabinet 100 is taken as an example. The architecture of cabinet 100 generally includes a plurality of racks, each of which includes a plurality of slots, each of which includes a plurality of nodes. Moreover, a switch 140 is disposed in the cabinet, and the switch 140 is coupled to each node device.

Referring to FIG. 1, the cabinet 100 includes n nodes, and these nodes are respectively provided with n node devices. These node devices can be classified into three types of nodes, namely, a service pool 110, a computing pool 120, and a storage nodes pool 130. The service resource pool 100 includes i node devices 110_1~110_i, the computing resource pool 120 includes j node devices 120_1~120_j, and the storage resource pool 130 includes k node devices 130_1~130_k.

In addition, in this embodiment, each of the node devices is provided with a light emitting unit, including the light emitting units 111_1~111_i, the light emitting units 121_1~121_j, and the light emitting units 131_1~131_k. Here, the light-emitting unit is, for example, a Light Emitting Diode (LED), but is not limited thereto.

The service resource pool 110 is used to provide a database service, a virtual resource provisioning service, a physical installer service, a physical manager service, a virtual manager service, and an application interface. Programming Interface, API) Service types such as service, storage manager service, load balance, and security services. The computing resource pool 120 is used to provide computing services. The storage resource pool 130 is used to provide a storage service.

In the embodiment, an abnormality detecting module 112 is installed in the node device 110_2 of the service resource pool 110 to monitor whether an abnormality occurs in the cloud servo system by using the abnormality detecting module 112. In other embodiments, the anomaly detection module 112 can also be installed in other node devices of the service resource pool 110 or installed in another server independent of the cabinet 100. In addition, the node device 110_1 in the service resource pool 110 is configured to provide a database service, and is provided with a node database 113 for storing node related information of each node device in the cabinet 100.

Here, the node related information records related information of each node device, and includes a plurality of items, each of which represents data of one node device. Each project includes network card information, processor information, memory information, hard disk information, node location, node type information, and service type. Specifically, the node device generally has a system network card and a Baseboard Management Controller (BMC) network card. The network card information includes the media access control (MAC) address of the BMC network card, the Internet Protocol (IP) address of the BMC network card, and the bandwidth (in Mbps (megabit)). Per second)), as well as the MAC address of the system NIC, the IP address and bandwidth of the system NIC. Processor information includes the processor model and operating frequency. The memory information includes the size of the memory module. Hard disk information includes carrier number, hard disk type, hard disk capacity, Revolution Per Minute (RPM), and hard disk cache capacity. The node location includes the rack number, slot number, and node number. The node type information is used to indicate that the corresponding node device belongs to the service resource pool 110, the computing resource pool 120, or the storage resource pool 130. The service type records the type of service provided by the node device corresponding to it.

The above-mentioned cloud servo system is used to explain the management method. 2 is a flow chart of a method for managing a cloud server system according to an embodiment of the present invention. Referring to FIG. 1 and FIG. 2 simultaneously, in the embodiment, the abnormality detecting module 112 is used to monitor whether the node devices are abnormal. The reason for the abnormality of the node device is, for example, a node device failure, a prediction device node failure, and a node device failure. It is removed, node device maintenance, node device forced replacement, node device join, but not limited to this.

When the abnormality detecting module 112 detects that an abnormality occurs in one of the node devices, the hardware address of the abnormal node device is obtained as shown in step S205. For example, the MAC address of the BMC of the abnormal node device is obtained.

Next, in step S210, the node related information of the abnormal node device is searched from the node database 113 according to the hardware address. Here, the location information of the abnormal node device is recorded in the node related information, and the location information is, for example, a node location of the abnormal node device in the cabinet 100 or a network address of the abnormal node device. The node location is also the actual location in the cabinet 100, ie, the rack number, the slot number, and the node number. The network address is, for example, the IP address of the BMC.

And, in step S215, the abnormal node device is isolated from the cloud servo system. For example, the abnormal node device is isolated from the cloud servo system according to an isolation mechanism.

In addition, in step S220, the light emitting unit of the abnormal node device is enabled according to the position information. For example, a command is sent to the BMC of the abnormal node device according to the location information, so that the BMC lights the light-emitting unit to be the first color (for example, red). When it is detected that the abnormal node device is replaced with another node device, the light-emitting unit of the replaced node device is lit to a second color (for example, green).

Accordingly, the manager finds the abnormal node device by observing the color of the light emitting unit. After that, the manager can pull out the abnormal node device and then insert the good node device. For example, replacing the abnormal node device with a higher-efficiency node device or replacing the abnormal hardware in the abnormal node device with a functioning hardware.

It is worth noting that after detecting that the abnormal node device is replaced with another node device, the node related information of the node device can be retrieved through a network management module. The network management module is, for example, a server having a Dynamic Host Configuration Protocol (DHCP) Servo (Server) function. In this embodiment, the network management module is disposed in another host independent of the cabinet 100. In other embodiments, the network management module can also be configured by any node device of the service resource pool 110 in the cabinet 100.

3 is a flow chart of a method for obtaining node related information according to an embodiment of the invention. Referring to FIG. 3, in step S305, the network management module receives the MAC address of the BMC from the node device to allocate an IP address to the BMC of the node device.

Next, in step S310, the network management module transmits an instruction to the BMC according to the IP address. The above instructions are, for example, an OEM (Original Equipment Manufacturer) instruction of the Intelligent Platform Management Interface.

When the BMC receives the above instruction, as shown in step S315, the central processing unit of the node device is restarted to obtain the node related information of the node device through the BMC during the restarting process. This is because the processor information, network card information, memory information, hard disk information, and node location are dynamically acquired, so the central processing unit is rebooted to the Extensible Firmware Interface (EFI) shell. It is obtained by the Basic Input Output System (BIOS) during the Power On Test Self (POST) and then transmitted to the BMC.

Then, in step S320, the BMC responds to the above instruction, and transmits the node related information to the network management module, so that the network management module stores the node related information into the node database 113. In addition, the network management module can also determine the type of service to be deployed by the node device according to the information about the node, and transmit the node related information to the cloud deployment program for deployment of the cloud operating system.

As far as the current technology is concerned, the BMC can be preset to automatically send its MAC address to the network management with DHCP servo function when the initial code is executed. Module to get the IP address of the BMC. In this embodiment, the network management module obtains node related information. However, in other embodiments, the node related information may also be obtained by the anomaly detection module 112. For example, the anomaly detection module 112 can issue an IPMI OEM command to the BMC through the IP address of the BMC to obtain node related information.

For example, Table 1 shows the node-related information of the node device of the node position (6, 0, 0). The node position (6, 0, 0) represents the rack number 6, the slot number is 0, and the node number is 0. In addition, Table 2 shows the node related information after the node location (6, 0, 0) is replaced by another node device.

When the node detecting module 112 detects that an abnormality occurs in the node device set by the node position (6, 0, 0), it acquires the MAC address of the abnormal node device BMC, that is, "00: A0: D1: EC: F8: BA". Then, the node database 113 is queried according to the MAC address "00: A0: D1: EC: F8: BA", and the node related information as shown in Table 1 can be obtained. Then, according to the IP address "10.1.0.19" of the abnormal node device BMC, a command is sent to the abnormal node device to illuminate the illumination unit of the abnormal node device to red.

When the manager finds the red light, the abnormal node device is unplugged and another node device is installed. Assume that the MAC address of the BMC NIC of the newly replaced node device is "00:A0:D1:EC:F8: F0", its bandwidth is 100Mbps, and the MAC address of the system NIC is "00:A0:D1:EA :34:F0", its bandwidth is 1000Mbps, the processor model is "Intel(R) Xeon(R) CPU E5540", the working frequency is 2260 MHz, and the node type is storage resource pool. And, assuming that the node device has four hard disks, the hard disk information of one hard disk is represented by (bay number, hard disk type, hard disk speed, hard disk capacity), which is (1, SAS, 1TB, 7200, 16 MB), (1, SAS, 1TB, 7200, 16 MB), (2, SAS, 1TB, 7200, 16 MB) and (2, SAS, 1TB, 7200, 16 MB).

After the node device is installed in the cabinet, the BMC obtains the IP address "10.1.0.19" of the BMC network card through the DHCP protocol, and obtains the IP address "10.1.0.20" of the system network card. Then, through the above steps S305~S320, the node related information as shown in Table 2 can be obtained, and the IP address of the system network card is obtained through the system network boot (Network Boot).

The architecture of the cabinet 100 will be described below by way of an example to illustrate how the BMC obtains the node location. 4 is a schematic diagram of a cabinet architecture in accordance with an embodiment of the present invention. Referring to FIG. 4, the cabinet 100 includes r racks, that is, rack 0 to rack r-1. Here, the rack 0 is taken as an example, and the architecture of the other racks is similar to that of the rack 0, and details are not described herein again. . Rack 0 includes s slots, that is, slot 0 to slot s-1. For convenience of explanation, slot 0 is taken as an example here, and the architecture of the remaining slots is similar to slot 0, and will not be described again. Slot 0 includes n nodes, that is, node 0 to node n-1. Taking node 0 as an example, a node device 400 is provided in node 0.

Here, for convenience of explanation, only some of the components of the node device 400 are shown. The node device 400 includes a central processing unit 410, a control chip 420, a basic input output system (BIOS) chip 430, a BMC 440, a memory module 450, a node number module 460, a system network card 470, and a light unit. 480 and BMC network card 490. The central processing unit 410 is coupled to the memory module 450 and coupled to the BIOS chip 440, the BMC 440, the node number module 460, and the system network card 470 through the control chip 420. In addition, the BMC 440 is coupled to the light emitting unit 480 and the BMC network card 490.

The central processing unit 410 is configured to execute data in the hardware, the firmware, and the processing software in the node device 400. Control chip 420 is a bridge for external processing of central processing unit 410. In the present embodiment, the control wafer 420 includes a north bridge wafer and a south bridge wafer. In other embodiments, the control wafer 420 is, for example, a south bridge wafer, and the north bridge wafer can be integrated with the central processing unit 410. The memory module 450 is, for example, a dual in-line memory module (DIMM). The node number module 460 is used to store the node number where the electronic device 400 is located. Here, the node number in the node number module 460 is 0.

In addition, the rack 0 includes a rack number module 401 for storing the rack number. Here, the rack number stored in the rack number module 401 is 0. The slot 0 includes an expander 403 and a slot number module 405. The expander 403 is coupled to the rack number module 401, the slot number module 405, and the control chip 420 of the node device 400. In node 0, the control chip 420 of the node device 400 acquires the rack number 0 and the slot number 0 where the node device 400 is located through the expander 403, and the node number module 460 knows that the node number at which it is located is zero. A register can be set in the control chip 420 to store the rack number, the slot number, and the node number. For example, set up three scratchpads to store. The BIOS chip 430 or the BMC 440 can control the wafer 420 from the control chip 420 to obtain the node position of the electronic device 400 in the cabinet 100.

It is assumed that the node position (that is, the physical position in the cabinet 100) is represented by (r, s, n), r represents the rack number, s represents the slot number, and n represents the node number. Accordingly, the node position of the node device 400 is (0, 0, 0). By analogy, the location of the node where the other node device controls the wafer is also stored.

Here, the expander 403 is coupled to the rack number module 401 and the slot number module 405 through a General Purpose Input (GPI) interface, and through an internal integrated circuit (Inter-Integrated Circuit, I ² C). The bus bar is coupled to the control wafer 420 of the node device 400. In addition, the control chip 420 is also coupled to the node number module 460 via a GPI interface, for example. That is, the numbers of the rack number module 401, the slot number module 405, and the node number module 460 can be obtained through the GPI interface.

In summary, in the above embodiment, the lighting unit is used to remind the manager of the abnormal node device where the abnormality occurs, so that the manager can replace the node device. Moreover, after the new node device is replaced, the node related information can also be automatically obtained for subsequent management such as automatic deployment, for example, to install and install the required operating system and service of the node device. The operation of the cloud system.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the invention, and any one of ordinary skill in the art can make some modifications and refinements without departing from the spirit and scope of the invention. The scope of the invention is defined by the scope of the appended claims.

100. . . Cabinet

110_1~110_i, 120_1~120_j, 130_1~130_k, 400. . . Node device

110. . . Service resource pool

120. . . Computing resource pool

130. . . Storage resource pool

111_1~111_i, 121_1~121_j, 131_1~131_k, 480. . . Light unit

112. . . Anomaly detection module

113. . . Node database

140. . . switch

401. . . Rack number module

403. . . Expander

410. . . Central processing unit

420. . . Control chip

430. . . BIOS chip

440. . . BMC

450. . . Memory module

460. . . Node number module

470. . . System NIC

490. . . BMC network card

S205~S220. . . Cloud servo system management method steps

S305~S320. . . Method of obtaining node related information

1 is a block diagram of a cloud servo system in accordance with an embodiment of the present invention.

2 is a flow chart of a method for managing a cloud server system according to an embodiment of the present invention.

3 is a flow chart of a method for obtaining node related information according to an embodiment of the invention.

4 is a schematic diagram of a cabinet architecture in accordance with an embodiment of the present invention.

S205~S220. . . Cloud servo system management method steps

Claims (6)

A cloud server system management method is applicable to a cloud server system, wherein the cloud server system includes a plurality of node devices, and the cloud server system management method comprises: detecting whether the node devices are abnormal; when detecting the When an abnormality occurs in one of the node devices, a first hardware address of the abnormal node device that generates the abnormality is obtained; and according to the first hardware address, a node related information of the abnormal node device is searched from a node database, The location information of the node is recorded in the node related information; the abnormal node device is isolated from the cloud server system; and an illumination unit of the abnormal node device is enabled according to the location information. The method for managing a cloud server system according to claim 1, wherein the location information is a node location of the abnormal node device in a cabinet or a network address of the abnormal node device. The method for managing a cloud server system according to claim 1, wherein the step of enabling the light emitting unit of the abnormal node device according to the location information comprises: transmitting a command to the abnormal node device according to the location information The substrate management controller is configured to illuminate the light emitting unit by the substrate management controller to be a first color. The method for managing a cloud server system according to claim 3, further comprising: after detecting that the abnormal node device is replaced with another node device, lighting the replaced light emitting unit of the node device as a The second color. The method for managing a cloud server system according to claim 1, wherein after detecting that the abnormal node device is replaced with another node device, the node of the node device is re-acquired through a network management module. The step of information includes: receiving a second hardware address from the node device to allocate a network address to the node device; transmitting an instruction to the node device according to the network address to transmit through the node device The baseboard management controller obtains the node related information of the node device; and updates the node related information to the node database. The method for managing a cloud server system according to claim 5, wherein the step of obtaining, by the baseboard management controller of the node device, the node related information of the node device comprises: when the substrate management controller receives the When the command is executed, a central processing unit of the node device is restarted to obtain related information of the node through the baseboard management controller during the restarting process.