Bare metal server deployment method, deployment controller, and server cluster

Abstract

The present invention provides a deployment method, a deployment controller and a server cluster for a bare metal server, the method comprising uploading an initial configuration file corresponding to a central processor configured by the bare metal server to a mirror server, obtaining and saving a mapping relationship between the initial configuration file and the architecture information to which the central processor belongs; planning a deployment network for the bare metal server, obtaining configuration information of the bare metal server to determine the node record information of the node to which the bare metal server belongs after it is started; pre-configuring a startup configuration file adapted to the central processor and saving it; calling the startup configuration file by the ironic service, matching the startup configuration file adapted to the central processor according to the deployment network and node record information, so as to start the bare metal server. The present application realizes the deployment of multiple bare metal servers in a heterogeneous architecture scenario of the central processor, and realizes batch and rapid deployment of bare metal servers.

Description

Technical Field

[0001] The present invention relates to the technical field of server deployment, and in particular to a bare metal server deployment method, a deployment controller, and a server cluster deployed based on the deployment method. Background Art

[0002] Cloud computing has significantly impacted the way information technology infrastructure is consumed. With virtualization technology, workloads can be deployed using a variety of virtual infrastructures, from public cloud environments to on-premises data centers that rely on local hardware. New workloads are continuously created, deployed, and consumed for applications via virtual infrastructure. Bare metal servers (BMS), as a physical server that provides computing services with the elasticity of virtual machines and the performance of physical machines, can better undertake the aforementioned deployed workloads.

[0003] In cloud computing scenarios with high security isolation and high business response requirements, a large number of bare metal servers need to be deployed. The architecture of the central processing unit (CPU) contained in each bare metal server may be different. At present, the startup (deployment) and management of bare metal servers in cloud computing platforms generally adopt a combination of IPMI servers and PXE clients with cloud network configuration. The applicant pointed out that in the prior art, when bare metal servers with different CPU architectures are deployed to cloud computing platforms, there is a need to separately write network configuration, storage configuration, and inspection scripts for different CPU architectures and in combination with Ironic services. Deployment instances, resulting in the deployment (startup) process of bare metal servers with different CPU architectures in the prior art is relatively cumbersome, complex, and prone to deployment errors. Therefore, it is impossible to achieve large-scale and automated deployment of bare metal servers in scenarios with heterogeneous central processing units.

[0004] In view of this, it is necessary to improve the deployment method of multiple bare metal servers in a heterogeneous architecture scenario of the central processing unit in the prior art to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to disclose a bare metal server deployment method, a deployment controller and a server cluster, so as to solve the complex technical problem of deploying multiple bare metal servers in a heterogeneous architecture scenario of the central processing unit in the prior art, so as to realize large-scale and automated deployment of bare metal servers in a heterogeneous architecture scenario of the central processing unit.

[0006] To achieve one of the above-mentioned purposes, the present invention provides a method for deploying a bare metal server, comprising the following steps:

[0007] S1. Upload an initial configuration file corresponding to the central processor configured on the bare metal server to the image server, obtain a mapping relationship between the initial configuration file and the architecture information to which the central processor belongs, and save the mapping relationship;

[0008] S2. Plan a deployment network for the bare metal server, and obtain configuration information of the bare metal server to determine node record information of the node to which the bare metal server belongs after it is started;

[0009] S3, pre-configure a startup configuration file adapted to the central processing unit and save it;

[0010] S4. The ironic service calls the startup configuration file, matches the startup configuration file adapted to the central processor according to the deployment network and node record information, so as to start the bare metal server.

[0011] As a further improvement of the present invention, the step S2 further includes:

[0012] Plan the deployment network and business network of the bare metal server, and visualize the node record information. The operations of planning the deployment network and business network of the bare metal server are performed in a visualization interface or a Neutron client to determine the VLAN ID and IP address segment to which the bare metal server belongs through the operations of planning the deployment network and business network of the bare metal server.

[0013] As a further improvement of the present invention, the initial configuration file includes: a kernel image, an initialization memory disk and an operating system image;

[0014] The initialization memory disk deploys IPA and operating system image;

[0015] The startup configuration file includes: a kernel image and an initialization memory disk.

[0016] As a further improvement of the present invention, the mapping relationship is saved to the configuration file of the ironic service.

[0017] or

[0018] The mapping relationship is stored in a storage device, so as to initiate a request for invoking the mapping relationship from the storage device through a visual interface;

[0019] The configuration file of the ironic service is saved in a database mounted to the ironic service.

[0020] As a further improvement of the present invention, the step S1 further includes: writing the first unique identification number and the second unique identification number respectively returned by uploading the kernel image and initializing the memory disk to the image server into the configuration file of the ironic service, wherein:

[0021] The mapping relationship includes the user who uploaded the initial configuration file, the architecture information of the central processing unit corresponding to the initial configuration file, and the first unique identification number and the second unique identification number respectively returned when uploading the kernel image and initializing the memory disk to the image server.

[0022] As a further improvement of the present invention, the obtaining of the configuration information of the bare metal server includes: determining whether the user obtains the architecture information of the central processing unit corresponding to the bare metal server;

[0023] If yes, the user manually enters the CPU architecture information corresponding to the bare metal server in the visual interface or ironic service;

[0024] If not, obtain the CPU architecture information corresponding to the bare metal server through the monitoring process in the ironic service.

[0025] As a further improvement of the present invention, the configuration information includes: IPMI IPS setting information, user name, password, physical network card information and central processing unit architecture information corresponding to the bare metal server.

[0026] As a further improvement of the present invention, the method of obtaining the architecture information of the central processor corresponding to the bare metal server through the monitoring process in the ironic service includes the following sub-steps:

[0027] The monitoring process monitors the DHCP messages of the bare metal server's network card;

[0028] Parse the options in the DHCP message to obtain the architecture information of the CPU corresponding to the Metal Server;

[0029] Save to the database mounted to the ironic service.

[0030] As a further improvement of the present invention, the storage device includes: a shared storage space for storing the startup configuration file through an address link, the shared storage space opens permissions to all bare metal servers that have been started and / or not started, and the shared storage space is deployed on a TFTP server.

[0031] As a further improvement of the present invention, the operation of pre-configuring the startup configuration file adapted to the central processing unit in step S3 is performed in a visual interface or a Neutron client.

[0032] As a further improvement of the present invention, step S1 also includes: saving the third unique identification number returned by uploading the operating system image to the image server to the database, so that after the bare metal server is started, the user can select the third unique identification number in the visual interface and pull the operating system image adapted to the third unique identification number from the image server.

[0033] As a further improvement of the present invention, step S4 also includes: after the bare metal server is started, starting the IPA and kernel image of the initialized memory disk to obtain a link to the operating system image, so as to write the operating system image to a local disk of the started bare metal server.

[0034] As a further improvement of the present invention, after writing the operating system image to a local disk of the bare metal server, it also includes: the ironic service sets the local disk of the bare metal server to start the bare metal server from the local disk through an IPMI command, and restarts the bare metal server, and after the bare metal server restarts, loads the operating system image to the system disk of the bare metal server.

[0035] At the same time, based on the same inventive concept, the present invention also discloses a deployment controller, which runs the deployment method of the bare metal server as described in any of the above inventions.

[0036] As a further improvement of the present invention, the deployment controller deploys a Trunk port; the bare metal server configures a management network physical interface, a deployment network physical interface and a business network physical interface; a management network data plane, a deployment network data plane and a business network data plane are formed between the deployment controller and several bare metal servers.

[0037] Finally, based on the same inventive concept, the present invention also discloses a server cluster, including:

[0038] A deployment controller,

[0039] a data exchange device, and

[0040] At least one bare metal server started by the deployment controller through the bare metal server deployment method described in any of the above inventions;

[0041] The bare metal server is connected to the data exchange device through a virtual network.

[0042] As a further improvement of the present invention, the virtual network is a hybrid virtual network composed of one or any two of a VXLAN virtual network, a GRE virtual network, a VLAN virtual network, and a GENEVE virtual network.

[0043] As a further improvement of the present invention, the deployment controller deploys a Trunk port; the bare metal server configures a management network physical interface, a deployment network physical interface and a business network physical interface; a management network data plane, a deployment network data plane and a business network data plane are formed between the deployment controller and several bare metal servers.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] A bare metal server deployment method and deployment controller disclosed in the present invention can well adapt to the deployment of multiple bare metal servers in a scenario where the central processing unit has a heterogeneous architecture, reduce the difficulty of deploying bare metal servers with central processing units configured with different architectures, and realize batch and rapid deployment of bare metal servers. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is an overall flow chart of a bare metal server deployment method of the present invention;

[0047] Figure 2 A schematic diagram of a deployment controller of the present invention implementing a deployment process on a bare metal server through a bare metal server deployment method of the present invention;

[0048] Figure 3 Schematic diagram for Ironic service;

[0049] Figure 4 A topological diagram of a server cluster of the present invention;

[0050] Figure 5 for Figure 2 The database (DB) stores a plurality of schematic diagrams of mapping relationships formed by the initial configuration files and the architecture information to which the central processor belongs;

[0051] Figure 6 for Figure 4 Schematic diagram of the data plane. DETAILED DESCRIPTION

[0052] The present invention is described in detail below in conjunction with the various embodiments shown in the accompanying drawings, but it should be noted that these embodiments are not limitations of the present invention, and any equivalent transformations or substitutions in functions, methods, or structures made by ordinary technicians in the field based on these embodiments are all within the scope of protection of the present invention.

[0053] It should be noted that when a component is considered to be " connect "Another element, which may be directly connected to another element or may have a central element at the same time. Before elaborating on the technical solutions and inventive ideas contained in the present application in detail, the technical meanings of some terms or abbreviations involved in the present application are briefly described or defined. Unless otherwise defined, all technical and scientific terms used in this document have the same meanings as those commonly understood by technicians in the technical field of the present invention.

[0054] the term" and / or ” includes any and all combinations of one or more of the associated listed items.

[0055] the term" BMS ”: Bare Metal Server (BMS) is a physical server that provides computing services with the elasticity of a virtual machine and the performance of a physical machine.

[0056] the term" Ironic ": A component of the bare metal server used in the Openstack environment, responsible for the activation and management of the bare metal server.

[0057] the term" Neutron "(i.e. Neutron service component): a network component based on the OpenStack environment, used to provide a Layer 2 / Layer 3 virtual network and provide network connectivity for virtual machine (VM) instances. It specifically includes Neutron-Server (receives and routes API requests to the network plug-in in OpenStack), OpenStackNetworking plug-ins and agents (creates ports, networks and subnets, and provides IP addresses) and Messaging Queue (saves routing information between Neutron-Server and agents, and serves as a database to store the network connection status of plug-ins).

[0058] the term" Glance ": A component in the OpenStack environment that provides image services and is used to provide virtual machine images. Specifically, it includes: querying and obtaining image metadata and images, maintaining image information (including metadata and images), executing snapshot commands on virtual machine instances to create new images, or backing up the status of virtual machines.

[0059] the term" PXE”: Pre-boot Execution Environment (PXE), PXE enables computers to boot through the network instead of from local hard disks, optical drives and other devices. Currently, mainstream network cards are embedded with ROM chips that support PXE. When the computer boots, the BIOS calls the PXE client into the memory for execution and displays a command menu. After the user selects, the PXEclient will download the operating system placed on the remote end through the network to the local machine for execution.

[0060] the term" DHCP ”: Dynamic Host Control Protocol (DHCP), used to centrally and dynamically assign IP addresses to clients.

[0061] the term" TFTP ”: Trivial File Transfer Protocol (TFTP), TFTP is a protocol in the TCP / IP protocol family used for simple file transfer between the client (C) and the server (S), and can be used for OS (operating system) and configuration updates of network devices.

[0062] The specific implementation process of the present invention is exemplarily described below through several embodiments.

[0063] Embodiment 1:

[0064] Ginseng Figures 1 to 3 As shown, this embodiment shows a deployment method of a bare metal server (hereinafter referred to as "deployment method"), including the following steps S1 to S4. It should be noted that in this application, the so-called startup refers to the first startup of the bare metal server, which is different from restarting. The central processing unit referred to in this embodiment is the CPU deployed in the bare metal server.

[0065] Step S1, upload the initial configuration file corresponding to the central processing unit configured by the bare metal server to the image server 50, obtain the mapping relationship formed by the initial configuration file and the architecture information to which the central processing unit belongs, and save the mapping relationship. The initial configuration file includes: a kernel image (i.e., Kernel), an initialization memory disk (i.e., Initrd) and an operating system image (i.e., sysImg). The initialization memory disk deploys IPA211 and an operating system image 212. IPA (i.e., ironic python agent) is a program built into initrd (boot loader initialized RAM Disk, initialized memory disk), which will automatically run when Initrd starts, and communicates with the ironic-conductor process in the deployment controller 10 according to the kernel parameters passed in, so as to pull the operating system image of the bare metal server from the opposite end (i.e., a device or system or computer that is logically independent of the bare metal server to be started and the deployment controller 10), and write the operating system image to the operating system of the bare metal server. Initrd (or Initrd file, or initialization memory disk) is a compressed small root directory, which contains the driver modules, executable files and startup scripts required by the bare metal server in the startup phase (and restart phase). Figure 2 The image server 50 can be adapted according to the image file type supported by the bare metal server, and does not need to be specifically limited. Figure 4 As shown, in the actual environment, Glance 11 is deployed in the mirror server 50, and Neutron 12 is deployed in the network node 40. The Ironic service provides a process for initiating the deployment method of the bare metal server from the visualization interface 60 for the user or administrator through the NOVA computing service and NOVA API (not shown).

[0066] The mapping relationship is saved to the configuration file of the ironic service.

[0067] or

[0068] The mapping relationship is saved in the storage device, so as to initiate a request to call the mapping relationship from the storage device through the visualization interface 60 .

[0069] The configuration file of the ironic service is saved to the database mounted to the ironic service (i.e. Figure 2 DB16 in ). Combined with Figure 5As shown, the database 16 can store k mapping relationships, and the parameter k is a positive integer greater than or equal to 1. Specifically, in this embodiment, the database 16 can store a mapping relationship 161 based on the X86 architecture, a mapping relationship 162 based on the MIPS architecture, a mapping relationship 163 based on the ARM architecture, and a mapping relationship 16k of other architectures, which are not listed here.

[0070] In this embodiment, the storage device includes: a shared storage space for storing startup configuration files through an address link (e.g., a URL), the shared storage space is open to all bare metal servers that have been started and / or not started, and the shared storage space is deployed in the TFTP server 101. The shared storage space is deployed in the TFTP server 101. CPUs with different architecture information (e.g., bare metal servers with CPUs based on the X86 architecture or bare metal servers with CPUs based on the MIPS architecture, etc.) will generate startup configuration files dedicated to one architecture information and save them to the shared storage space of the TFTP server 101. The shared storage space references Figure 2 The shared directory of / tftpboot is stored in the storage device so that the physical interface 202 (ie, PXE Eth0) of the deployed network can be used to pull the Kernel and Initrd from the shared directory. The storage device also includes a cache and a non-volatile memory.

[0071] Step S1 also includes: writing the first unique identification number and the second unique identification number respectively returned by uploading the kernel image and initializing the memory disk 21 to the image server 50 into the configuration file of the ironic service, wherein the mapping relationship includes the user who uploaded the initial configuration file, the architecture information of the central processor corresponding to the initial configuration file, and the first unique identification number and the second unique identification number respectively returned by uploading the kernel image and initializing the memory disk 21 to the image server 50. Figure 4 As shown, during the deployment of the bare metal server BMS_1, the first unique identification number and the second unique identification number can be perceived by the user or administrator in the visualization interface 60, and the user or administrator issues an operation instruction or command in the visualization interface 60, and can be selected by the user or administrator.

[0072] Step S1 also includes: saving the third unique identification number returned by uploading the operating system image to the image server 50 to the database 16, so that after the bare metal server BMS_1 is started, the user selects the third unique identification number in the visual interface 60, and pulls the operating system image (sysImg) adapted to the third unique identification number from the image server 50. In this embodiment, the first unique identification number to the third unique identification number are all UUDI.

[0073] Step S2, plan the deployment network of the bare metal server BMS_1, obtain the configuration information of the bare metal server BMS_1 to determine the node record information of the node to which the bare metal server BMS_1 belongs after it is started. The deployment network of the bare metal server BMS_1 can be determined according to the network planning settings in the server cluster. The aforementioned network planning refers to the network load balancing (NLB) in the entire server cluster after the bare metal server is deployed, the number and topology of computing nodes (Note: the bare metal server is usually used as a computing node in a cluster or server cluster), IP address segment, VPN settings, default gateway and other network configurations to achieve secure isolation of the bare metal server.

[0074] In this embodiment, step S2 also includes: planning the deployment network and business network of the bare metal server BMS_1, and visually showing the node record information. The operation of planning the deployment network and business network of the bare metal server BMS_1 is performed on the visualization interface 60 or the Neutron client (see Figure 2 Neutron12) in is executed to determine the VLAN ID and IP address segment to which the bare metal server BMS_1 belongs by planning the deployment network and business network of the bare metal server BMS_1.

[0075] Obtaining the configuration information of the bare metal server includes: determining whether the user obtains the architecture information of the central processor corresponding to the bare metal server BMS_1;

[0076] If so, the user in the visualization interface 60 (refer to Figure 4 ) or ironic service (see Figure 3 ) manually enter the architecture information of the CPU corresponding to the bare metal server BMS_1;

[0077] If not, obtain the architecture information of the CPU corresponding to the bare metal server BMS_1 through the monitoring process in the ironic service.

[0078] Specifically, in this embodiment, obtaining the architecture information of the central processor corresponding to the bare metal server BMS_1 through the monitoring process in the ironic service includes the following sub-steps:

[0079] Sub-step S21: The monitoring process monitors the DHCP message of the network card of the bare metal server BMS_1. Figure 2 The data is forwarded in the deployed network data plane 302 and monitored through the Trunk port 102.

[0080] S22, parsing the options of the DHCP message to obtain the architecture information of the central processing unit corresponding to the metal server.

[0081] S23. Save to the database mounted to the ironic service (ie, DB16).

[0082] Configuration information includes: IPMI IPS setting information, user name, password, physical network card information and CPU architecture information corresponding to the bare metal server. Figure 2 The physical network physical interface 201 (such as IPMI interface), the deployment network physical interface 202 (such as PEX Eth0) and the business network physical interface 203 (such as Eth1).

[0083] Ginseng Figure 3 and Figure 4 As shown, the Ironic service includes Ironic-API14, Ironic-Conductor13 and Ironic-check15 that communicate with NOVAAPI, and the following monitoring process is executed by Ironic-check15. Ironic-API14 forwards the request to Ironic-Conductor13 through the RPC protocol (a remote procedure call protocol). Ironic-Conductor13 uses the MAC of the bare metal server to call the interface of Neutron12 to create a port. The process of creating a port will call the normative command of the data exchange device 30 with the RPC protocol to perform VLAN settings. The Ironic service and the bare metal server cluster 20 containing at least one bare metal server jointly access the data plane.

[0084] In particular, the applicant pointed out that in this embodiment, the architecture information of the bare metal server to be deployed is obtained by manual entry or by means of the bypass monitoring DHCP message of the monitoring process in the aforementioned ironic service. The aforementioned manual entry refers to manually entering the architecture information of the central processor corresponding to the bare metal server in the visual interface 60 or the ironic service. Through the above-mentioned technical means, it is possible to significantly adapt to the startup and batch deployment of bare metal servers with heterogeneous central processors (i.e., CPUs with different architecture information), reduce the difficulty of deploying bare metal servers with central processors configured with different architectures, thereby realizing batch deployment and rapid deployment of bare metal servers.

[0085] Step S3, pre-configure and save the startup configuration file adapted to the central processing unit. The startup configuration file includes: a kernel image and an initialized memory disk. The operation of pre-configuring the startup configuration file adapted to the central processing unit in step S3 is performed in the visual interface 60 or the Neutron client 60. The operation of saving the configuration file does not need to be re-prepared the next time the same bare metal server is started, thereby caching the kernel image, initrd image, and business image files required when the PXE information is activated, thereby solving the deployment simplicity of the bare metal server during repeated deployment.

[0086] Step S4, the ironic service calls the startup configuration file, matches the startup configuration file adapted to the central processing unit according to the deployment network and node record information, so as to start the bare metal server. In an embodiment, the startup configuration file is an NBP file, namely, a Network Bootloader Program, which is used to load the bare metal server during the deployment process to load the configuration files, kernel files, initrd and other files required for the required operating system startup. For example, the NBP file instance corresponding to the x86_64 architecture (i.e., a 64-bit x86 architecture central processing unit) is undonly.kpxe, and the NBP file instance corresponding to the ARM architecture is grubaa64.efi. Neutron DNSmasq121 replies with the corresponding IP address and TFTP address, and determines the logical location of the adapted NBP file (i.e., Figure 2 At the same time, Neutron12 will set the NBP file as the DHCP PXR option attribute of the corresponding port, so that the bare metal server can obtain the required PXE information when it is subsequently started, so as to perform the boot startup of the bare metal server through the PXE information. Obtain the Kernel and Initrd files 21 from the TFTP server 101 through the NBP file to start the kernel image and initialize the memory disk. When the bare metal server is started through PXE, the kernel image and initrd file are loaded to load the bare metal server with the driver (Drive) and file system (i.e. File System) required for normal operation after startup.

[0087] Step S4 also includes: after the bare metal server is started, starting the IPA 211 and kernel image (i.e., the Initrd file 21) of the initialization memory disk Figure 2Kernel in the bare metal server) to obtain the link of the operating system image to write the operating system image to a local disk of the bare metal server that has been started. The local disk is the only system disk of the bare metal server BMS_1. At the same time, after the operating system image (sysImg) is written to a local disk of the bare metal server, it also includes: the ironic service sets the local disk of the bare metal server to start the bare metal server from the local disk through the IPMI command, and restarts the bare metal server, and after the bare metal server is restarted, loads the operating system image (sysImg) to the system disk (Sysdisk) of the bare metal server, thereby completing the deployment operation of the bare metal server. The bare metal server should deploy at least one data disk to store business data.

[0088] In summary, the bare metal server deployment method disclosed in this embodiment can well adapt to the deployment of multiple bare metal servers in a heterogeneous architecture scenario of the central processing unit, reduce the difficulty of deploying bare metal servers with central processing units configured with different architectures, and realize batch and rapid deployment of bare metal servers.

[0089] Embodiment 2:

[0090] Based on the bare metal server deployment method disclosed in the first embodiment, this embodiment further discloses a deployment controller 10.

[0091] Ginseng Figures 2 to 6 As shown, a deployment controller 10 disclosed in this embodiment runs the bare metal server deployment method in Embodiment 1. The deployment controller 10 deploys a Trunk port 102. The bare metal server BMS_1 configures a management network physical interface 201 (e.g., an IPMI interface), a deployment network physical interface 202 (e.g., PEX Eth0), and a business network physical interface 203 (e.g., Eth1). A management network data plane 301, a deployment network data plane 302, and a business network data plane 303 are formed between the deployment controller 10 and a plurality of bare metal servers BMS_1.

[0092] The aforementioned management network physical interface 201, deployment network physical interface 202 and business network physical interface 203 are regarded as a physical network card in the example, and are respectively connected to the management network data plane 301, the deployment network data plane 302 and the business network data plane 303. The Trunk port 102 can allow multiple VLAN messages with Tag tags to pass through, and can receive and send multiple messages, and forward them to terminal devices that meet the Tag marking rules (such as a bare metal server BMS_1).

[0093] Figure 2The DHCP request, pulling of Kernel and Initrd, and the message generated by the configuration data exchange device 30 go through the deployment network data plane 302 through the Trunk port 102, and communicate with the deployment network physical interface 202. The message generated by the Ironic service calling IPA to write the operating system image and sending the IPMI request goes through the management network data plane 301 through the Trunk port 102, and communicates with the management network physical interface 201. The business message generated by the bare metal server BMS_1 after startup and restart goes through the business network data plane 303 through the Trunk port 102, and communicates with the business network physical interface 203. The data exchange device 30 establishes a data plane to meet the forwarding requirements of the bare metal server BMS_1 in the deployment stage and the actual operation stage after deployment for the business data generated by the bare metal server BMS_1 in response to the user forwarding request. The data exchange device 30 is specifically a configuration switch. The user can connect a computer through the Console port (i.e., the control port of the configuration switch) and configure the host name of the configuration switch, the SSH remote login password, the configuration device management address, and other operations in the computer interface. The data exchange device 30 is respectively configured with an IPMI interface to access a management network data plane 301 , a DEPLOY port to access a deployment network data plane 302 , and a DATA port to access a service network data plane 303 .

[0094] The deployment controller 10 disclosed in this embodiment has the same technical solutions as those in the bare metal server deployment method disclosed in the first embodiment, which are described in the first embodiment and will not be repeated here.

[0095] Embodiment three:

[0096] Combination Figure 2 , Figures 4 to 6 As shown, based on the technical solutions disclosed in the above-mentioned embodiment 1 and embodiment 2, this embodiment further discloses a server cluster, which includes:

[0097] A deployment controller 10,

[0098] a data exchange device 30, and

[0099] At least one bare metal server (BMS_1 to BMS_i, parameter i is a positive integer greater than or equal to 1) is started by the bare metal server deployment method described in Example 1 by the deployed controller 10. The bare metal servers BMS_1 to BMS_i are connected to the data exchange device 30 through a virtual network. Multiple BMSs constitute a BMS cluster 20 to provide elastically scalable high-performance computing services with physical isolation performance for the entire server cluster through one or more BMSs and a virtualization system (such as a Hypervisor).

[0100] In this embodiment, the virtual network is a hybrid virtual network composed of one or any two of the VXLAN virtual network, the GRE virtual network, the VLAN virtual network, and the GENEVE virtual network. The deployment controller 10 deploys the Trunk port 102. The bare metal server configures the management network physical interface 201, the deployment network physical interface 202, and the business network physical interface 203. A management network data plane 301, a deployment network data plane 302, and a business network data plane 303 are formed between the deployment controller 10 and several bare metal servers. The virtual network is determined by the type supported by the data switching device 30, and this embodiment does not specifically limit it.

[0101] The server cluster disclosed in this embodiment has the same technical solutions as those in Embodiment 1 and Embodiment 2. Please refer to Embodiment 1 and Embodiment 2 for details, which will not be repeated here.

[0102] The series of detailed descriptions listed above are only specific descriptions of feasible implementation methods of the present invention. They are not intended to limit the scope of protection of the present invention. Any equivalent implementation methods or changes that do not deviate from the technical spirit of the present invention should be included in the scope of protection of the present invention.

[0103] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for deploying a bare metal server, characterized in that: The following steps are involved: S1. Upload an initial configuration file corresponding to the central processor configured on the bare metal server to the image server, obtain a mapping relationship between the initial configuration file and the architecture information to which the central processor belongs, and save the mapping relationship; S2. Planning a deployment network of a bare metal server, obtaining configuration information of the bare metal server to determine node record information of a node to which the bare metal server belongs after being started, wherein obtaining the configuration information of the bare metal server includes: determining whether the user obtains architecture information to which a central processor corresponding to the bare metal server belongs; If yes, the user manually enters the CPU architecture information corresponding to the bare metal server in the visual interface or ironic service; if no, the CPU architecture information corresponding to the bare metal server is obtained through the monitoring process in the ironic service; S3, pre-configure a startup configuration file adapted to the central processing unit and save it; S4. The ironic service calls the startup configuration file, matches the startup configuration file adapted to the central processor according to the deployment network and node record information, so as to start the bare metal server.

2. The deployment method according to claim 1, characterized in that: The step S2 further comprises: Plan the deployment network and business network of the bare metal server, and visualize the node record information. The operations of planning the deployment network and business network of the bare metal server are performed in a visualization interface or a Neutron client to determine the VLAN ID and IP address segment to which the bare metal server belongs through the operations of planning the deployment network and business network of the bare metal server.

3. The deployment method according to claim 1, characterized in that: The initial configuration file includes: kernel image, initialization memory disk and operating system image; The initialization memory disk deploys IPA and operating system image; The startup configuration file includes: a kernel image and an initialization memory disk.

4. The deployment method according to claim 3, characterized in that: The mapping relationship is saved to the configuration file of the ironic service. or The mapping relationship is stored in a storage device, so as to initiate a request for invoking the mapping relationship from the storage device through a visual interface; The configuration file of the ironic service is saved in a database mounted to the ironic service.

5. The deployment method according to claim 4, characterized in that: The step S1 further includes: writing the first unique identification number and the second unique identification number respectively returned by uploading the kernel image and initializing the memory disk to the image server into the configuration file of the ironic service, wherein: The mapping relationship includes the user who uploaded the initial configuration file, the architecture information of the central processing unit corresponding to the initial configuration file, and the first unique identification number and the second unique identification number respectively returned when uploading the kernel image and initializing the memory disk to the image server.

6. The deployment method according to claim 1, characterized in that: The configuration information includes: IPMIIPS setting information, user name, password, physical network card information and central processing unit architecture information corresponding to the bare metal server.

7. The deployment method according to claim 1, characterized in that: The method of obtaining the architecture information of the central processor corresponding to the bare metal server through the monitoring process in the ironic service includes the following sub-steps: The monitoring process monitors the DHCP messages of the bare metal server's network card; Parse the options in the DHCP message to obtain the architecture information of the CPU corresponding to the Metal Server; Save to the database mounted to the ironic service.

8. The deployment method according to claim 4, characterized in that: The storage device includes: a shared storage space for storing the startup configuration file through an address link, the shared storage space is open to permissions for all started and / or unstarted bare metal servers, and the shared storage space is deployed on a TFTP server.

9. The deployment method according to claim 1, characterized in that: The operation of pre-configuring the startup configuration file adapted to the central processing unit in step S3 is performed in the visual interface or the Neutron client.

10. The deployment method according to any one of claims 3 to 5, characterized in that: The step S1 also includes: saving the third unique identification number returned by uploading the operating system image to the image server to the database, so that after the bare metal server is started, the user can select the third unique identification number in the visual interface and pull the operating system image adapted to the third unique identification number from the image server.

11. The deployment method according to claim 10, characterized in that: The step S4 also includes: after the bare metal server is started, starting the IPA and kernel image of the initialized memory disk to obtain a link to the operating system image, so as to write the operating system image into a local disk of the started bare metal server.

12. The deployment method according to claim 11, characterized in that: After writing the operating system image to a local disk of the bare metal server, the method further includes: the ironic service sets the local disk of the bare metal server to start the bare metal server from the local disk through an IPMI command, restarts the bare metal server, and after the bare metal server restarts, loads the operating system image to the system disk of the bare metal server.

13. A deployment controller, characterized in that: The deployment controller runs the bare metal server deployment method according to any one of claims 1 to 12.

14. The deployment controller according to claim 13, characterized in that: The deployment controller deploys Trunk ports; the bare metal server configures a management network physical interface, a deployment network physical interface and a business network physical interface; a management network data plane, a deployment network data plane and a business network data plane are formed between the deployment controller and a number of bare metal servers.

15. A server cluster, characterized in that: include: A deployment controller, a data exchange device, and at least one bare metal server started by the deployment controller through the bare metal server deployment method according to any one of claims 1 to 12; The bare metal server is connected to the data exchange device through a virtual network.

16. The server cluster according to claim 15, characterized in that: The virtual network is a hybrid virtual network consisting of one or any two of a VXLAN virtual network, a GRE virtual network, a VLAN virtual network, and a GENEVE virtual network.

17. The server cluster according to claim 15, characterized in that: The deployment controller deploys Trunk ports; the bare metal server configures a management network physical interface, a deployment network physical interface and a business network physical interface; a management network data plane, a deployment network data plane and a business network data plane are formed between the deployment controller and a number of bare metal servers.

Images