In this chapter you will explore the process of OS installation, including seemingly minor details that present challenges and make life interesting. S. This chapter is not intended as the tell-all/end-all reference for installing an OS! Every OS release changes something. Usually a new release of an OS requires more disk space than the previous release. Sometimes a new release will come with a new installation tool. Always read your OS vendor’s documents before attempting to install the system.The following material covers the most problematic issues involved in a typical installation.
The Windows installer has a habit of wanting companionship, asking another question every 15 minutes or so just to make sure the person performing the installation stays nearby, and awake. This may be acceptable for a one- or two-host installation, but it becomes extremely inefficient if you have to install the OS on 100 computers. If you find that you need to perform many “cookie-cutter” installations, look into the network install-server options available with most of the high-end Windows operating systems. An example network install script is discussed in the “network installations” section (below). All versions of Windows provide a GUI-based installation tool. The first portion of the interaction between the user and the installation tool is primarily an information-gathering phase. The user is asked for information regarding the type of install, the disk partitioning, type of file system to install, and the system hardware. Once these questions are answered, the system formats the specified partition, and installs a minimal system on the hard disk. When that partition is booted, the system resumes the installation process. The next phase of the installation is an automated process in which the installer loads the required OS binaries onto the system disk. Once the first round of binaries are installed, the system reboots, and the installation continues, with the system running from the system hard disk. At this point, a minimal OS has been installed. However, this OS needs to be customized to fit into the local environment. The next phase of the installation prompts the administrator for local configuration information such as time zone, host name, network address information, domain membership, and so on. As the administrator answers questions, the installer configures services, and moves on to the next set of questions. Once all binaries are installed and configuration is complete, the system is ready for patches.
It is possible to build answer files that will install thin clients, thick clients, and some types of servers. Once the answer files are built, they must be tested to ensure they do the right thing. To test answer files, you often build a bootable diskette for the specific installation at hand. You boot the installer diskette on the system to be installed, and the installation is completely controlled by the answer files on the server. The configuration/testing portion of the network load process can be very tedious, but if you have many systems to install the time spent developing and testing these rules will pay off in the long run, as subsequent installations are very simple, and can often be handled by inexperienced personnel. The answer files are text files that contain answers to the questions asked by the installation program. A sample answer file is included on the Windows distribution media under the name \\i386\\Unattend.txt . The Windows distribution media also contains a program named Setup Manager (in the file Support\\Tools\\Deploy.cab ), which can be used to create custom answer files. Once the answer file has been created, you need to create the distribution folder that contains a copy of the CD distribution media. To populate the distribution folder, copy the content of the distribution media into the folder you just created. Refer to the information regarding unattended installations for more information, and a “checklist” procedure for this type of installation.
Most provide context-sensitive help, on demand, throughout the process. One notable exception to this generalization appears to be the Debian Linux distribution. Many Linux users love their Debian systems, once they get through the installation. However, the pain of installing the OS is almost enough to send them running for the latest RedHat media instead of waiting for the final result!
Under Solaris, the first task consists of creating several configuration files. Among these files are /etc/bootparams , /etc/ethers , and sysidcfg . The network boot process reads the configuration files, and then listens for an installation request from the network. The client that requires installation services sends a DHCP request packet, which asks for network information. Once the client has a valid network address, it asks if anyone knows how to boot it. The boot server replies by sending the client the installer image to be booted. The client loads and starts the installer, which handles the remainder of the install. Solaris provides a set of scripts that will assist you if you decide to create an installation server. The distribution CD contains a directory named Tools , which contains the setup_install-server , add_to_install_server , modify_install_server , and add_install_client scripts. These scripts are used, respectively, to create an install server, copy the installation media to the install server, add the WebStart interface to the server, and add client information to the server. The instructions for using these scripts are in the main pages, and in the Advanced Installation Guide distributed with the OS media. The distribution media also contains some sample configuration and rules files for your perusal.