Mirror image file making and starting method of diskless system

Abstract

The invention discloses a mirror image file making and starting method of a diskless system. The mirror image file making method comprises the following steps: configuring a kernel basic module to compile and generate a kernel and a kernel module; Generating a root file system directory of the target operating system, after the kernel is added into the root file system directory to be started, theNFS service is mounted; obtaining the execution scripts of the software and the driver from the NFS server, a kdump tool is added and configured to be started and store system downtime logs to an NFSserver, a system log service script is established to redirect the system logs to the NFS server, and a root file system directory is packaged to generate a target operation system mirror image; thestarting method comprises the step of realizing diskless starting based on the mirror image file and PXE + DHCP + TFTP + NFS. According to the method, the size of the mirror image file can be greatlyreduced, the starting efficiency is improved, and system fault analysis during system downtime is facilitated.

Description

technical field [0001] The invention relates to a computer diskless startup technology, in particular to a method for making and starting an image file of a diskless system. Background technique [0002] Generally speaking, the operating system is installed in the hard disk, which is the so-called "boot with disk". However, some computer clients in high-performance computing, embedded systems, Internet cafes, hotels, education and many other fields do not have local disks. In these systems, computer clients need to load and run operating systems from remote servers, that is, "diskless start up". All the information of the computer client in the diskless boot system is stored in the memory, such as: the operating system, the logs generated during the running of the system, and the kernel running information are all stored in the memory. Information will be lost. [0003] In the diskless system, the server where the diskless system kernel and mirror image are located is cal...

AI Technical Summary

Problems solved by technology

[0010] In actual use, the traditional diskless computer startup method has the following defects: 1) When customizing the image of the target operating system, all files such as the kernel module, system software, library files, and header files required by the system must be organized according to the rules of the linux directory tree Packed into the image file, the image file is too large, and the system takes a long time to transfer the file through the network when the system is booted, and the network pressure is high. At the same time, the time for decompressing the larger image file to the memory will be longer

At the same time, the kernel has a limit on the image size

2) Due to the different needs of users, it is impossible to know all the files needed when customizing the image. Therefore, when the user's needs change frequently, the image file also needs to be modified frequently. Since the image file is relatively large, it takes a long time to package each time, which is time-consuming and laborious

Since all files are organized in the image according to the directory tree rules of the Linux system, the interdependence between library files is more complicated, and it is also difficult to delete unnecessary software and its dependencies

4) By default, the diskless system log service program rsyslog saves the log in the memory. When the system crashes or restarts, the log will be lost, and it is impossible to analyze and troubleshoot the system. In addition, when a kernel panic occurs, usually through The monitor can only see the information on the last screen, the real cause of the kernel error is no longer visible, and the exact cause of the kernel error cannot be analyzed

Images