Linux Tutorial: Remote network installation of RedHat-Linux implementation

At present, many managers have a headache in network management, that is, installing the operating system for users. I once worked in a number of companies, many companies for the client configuration basically few with CD-ROM and not all clients have operating system installation CD. As a result, I often find that many maintenance personnel often use optical drives and discs to re-perform or upgrade the operating system for users. To solve this problem, the author tries to solve the boring problem of MIS personnel. At present, many managers have a headache in network management, that is, installing the operating system for users. I once worked in a number of companies, many companies for the client configuration basically few with CD-ROM and not all clients have operating system installation CD. As a result, I often find that many maintenance personnel often use optical drives and discs to re-perform or upgrade the operating system for users.

Therefore, in order to solve the boring and repetitive work of MIS personnel, I have established Windows Remote Installation server and Linux Remote Installation server for all enterprises that have served. Here I will discuss with computer enthusiasts about the Remote Installation process and implementation of Linux.

I. operating principle of Linux Installation

As we all know, during the installation process, Linux requires a boot program to enable the installation CD, and then implements configuration and installation. However, if the installation is converted to remote network installation, our installer needs to be remotely transmitted to the client through the network. Therefore, when setting up a remote Linux server, you must pay attention to the following issues:

1. the remote client is bare metal. The bare metal defined here is that there is no operating system or software program except standard hardware. Of course, you can also have an operating system or software. The reason for telling everyone that bare metal is the simplest way of thinking.

2. the remote client must be able to start the network. The problem described here is that the remote client must be able to send the installation request through the network to send the installation request to the server. Therefore, the client must have a nic, And the NIC has a pxe rom [pre-boot execution environment] chip, and the BIOS can support network startup.

3. after the remote computer is started, initialization settings will appear. At this time, because the client is not in the operating system or is in the bare metal status, it is impossible to bind the familiar TCP/IP address or other network protocol address. Therefore, the communication between them must automatically assign the service-DHCP through the IP address to distribute IP addresses for the Client, so as to achieve the connection and communication between the Client and Server.

4. After obtaining the IP address from the Remote Installation client, send a request to remotely install the server to a Linux Bootstrap program on the client to guide Linux and implement the installation interface on the client. However, if the TCP protocol is used for the transmission service in the TCP/IP protocol stack, the famous TCP three-way handshake process will appear, resulting in remote clients waiting for timeout and unable to complete the guiding process. Therefore, UDP protocol must be used for fast transmission to ensure remote client guidance. Therefore, the program used here is the TFTP [Trivial File Transfer Protocol] service as the server for remote transmission of Linux boot programs.

5. although UDP speed is very fast (this is due to its own structure), it is still one of the unreliable protocols and does not support mechanisms such as error correction and data arrival to the destination, therefore, you still need to use the TCP mode to transmit a large number of files or large files. This will not cause installation failure due to UDP transmission packet loss.

There are multiple ways to choose from:
1) NFS [Network File System]
2) HTTP
3) FTP
I chose the FTP method here, mainly because the FTP method is relatively simple.

Therefore, preparing to establish a Remote Installation server requires at least three services: DHCP, TFTP, and FTP. The three services can be implemented on multiple different computers or on the same computer.

Principles:
The remote client computer starts. Because the BIOS sets the NIC to start, the program in the pxe rom of the NIC is transferred to the memory for execution. First, the client finds the DHCP server in the network and then requests an IP address. At the same time, the DHCP server contacts the TFTP server and sends a bootstrap (bootstrap program) to the client ). After the client receives the bootstrap (File pxelinux.0) and runs the command, bootstrap requests TFTP to transfer the bootstrap configuration file (pxelinux. cfg ). Read the configuration file after receiving it. The client requests TFTP to transfer the kernel image file (vmlinuz) and the root file system file (initrd. img) based on the configuration file content and customer information ). Finally, start the kernel. This is a complete pxe build process. However, to enable the NIC to start and then install the system on the network, the FTP service also needs to place the required installation files in the FTP directory for transmission and installation.

The principles are illustrated as follows: