Chapter 6

Making the Initial Server Installation

Previous chapterNext chapterContents


This chapter covers the following topics:

One of the primary selling points of Windows NT Server 4.0 is its ease of installation. A "standard" installation, using default values, on a new server computer takes less than 30 minutes. Windows NT Server 4.0 is such a powerful and flexible operating system that a one-size-fits-all "standard" installation procedure is sure to shortchange some users or, in the worst case, deprive all users of the potential capabilities of Windows NT Server 4.0.

The process of installing a network operating system starts long before you put the distribution CD-ROM into the drive. This is especially true for Windows NT Server 4.0, because you have a number of decisions that shouldn't be made while staring at an installation question that you must answer in the dialog on-screen. If your server is now running another operating system, you have many more decisions to make than if you're installing Windows NT on a brand-new computer. If you're installing Windows NT Server 4.0 for the first time, you should read this chapter thoroughly before touching the door of the CD-ROM drive.

Gathering Information and Making Decisions

You must consider several major issues before you start to install Windows NT Server 4.0. Are you installing Windows NT Server over another operating system? Are you familiar with all the hardware contained in or attached to your server? Is your computer already connected to a network (perhaps in a peer-to-peer configuration), or are you installing Windows NT Server 4.0 and a new network at the same time? The following sections cover each issue.

If you're upgrading Windows NT Server 3.5+ to Windows NT Server 4.0, the installation process is very close to automatic. All your existing settings-including Registry entries, users, groups, and services-are preserved. Thus, much of the advice in the following sections doesn't apply to a Windows NT upgrade. The most important change involves the new hardware device drivers required for version 4.0. Be sure to review the "Knowing Your Hardware" section before upgrading.

Preparing to Upgrade from Another Version of Windows

If you're running Windows 3.1, Windows for Workgroups 3.11, or Windows 95 on your server, it's a reasonably good bet that you plan to continue to use the files (and possibly applications) installed there after Windows NT is installed. For example, you may be running a shared-file database server that you plan to use under Windows NT Server 4.0 until upsizing to a client/server relational database management system, such as Microsoft SQL Server 6.5, that runs under Windows NT Server. In this case, you must tread very carefully during the installation procedure.

It's an uncommon practice to run conventional end-user Windows applications on a Windows NT server. If you're upgrading a PC used as a peer-to-peer server in a Windows 95 or Windows for Workgroups 3.1+ environment, it's recommended that you devote the server PC to entirely to Windows NT Server. Uninstall Windows 95 applications, such as Microsoft Office 95, before installing Windows NT to conserve disk space. (You must reinstall Office 95 applications to run them under Windows NT Server or Workstation.) If you're upgrading from 16-bit Windows or Windows 95, install Windows NT in its own folder (\Winnt), not your existing \WINDOWS directory or \Windows folder. By doing so, you can easily delete the old operating system files after completing the Windows NT Server installation.

In addition to accessing existing files, do you want to be able to continue to use the existing operating system? As a rule, running two operating systems on a server (called dual-booting) is very dangerous. You don't want people to be able to boot the server into DOS/Windows or some other operating system, bypassing all the security procedures that you've instituted under Windows NT Server. If someone with access to the room that holds the server can boot it into another operating system, they can access (read and change) any file that operating system can read. Further, dual-booting precludes the use of NTFS (NT File System) for volumes that other Windows versions must be able to access.

Assuming that you want to keep all the old files but not use the old operating system, do you want your old settings preserved and used in Windows NT Server? You need to know the answers to these questions before you begin the installation procedure.

Some decisions you make during installation, such as deciding whether the computer will be a domain controller or a server, can't be undone without completely reinstalling Windows NT Server. If you're installing Windows NT Server 4.0 in an existing Windows domain, you must know the role that the new server plays in the domain-Primary Domain Controller (PDC), Backup Domain Controller (BDC), or plain server. If you're installing Windows NT Server over an existing version, the server role is predefined. Before you start an upgrade, make sure that the server is connected to the network and the network connection is active. Later, the "Choosing to Install as a Domain Controller" section explains these roles.

Before proceeding with the Windows NT Server installation on a PC that shares folders, back up at least the shared folders to tape or to another PC with adequate disk space. Backing up to tape and to another PC is the best insurance against tape drive or media failure. Backing up to another PC is particularly important if you have a backup tape drive that isn't supported with a new Windows NT 4.0 driver. Later, the section "Backing Up Data on an Existing Computer" discusses backup operations in greater detail.

Knowing Your Hardware

The Setup program that installs Windows NT Server provides automatic hardware detection, but you need to answer questions about your hardware and make decisions before you start the installation process. The most important question about any hardware component you plan to use is this: Is the hardware component supported by Windows NT Server? If it isn't supported, you may have to install custom Windows NT drivers provided by your hardware manufacturer. The product documentation that accompanies the retail version of Windows NT Server 4.0 includes the Hardware Compatibility List (HCL) booklet. Updates to the HCL are provided at regular intervals at the following locations:

Be sure to check all the hardware in your system against this list before you begin to install Windows NT Server. If a hardware component you plan to use isn't on the list, you must obtain a 32-bit Windows NT driver. If such a driver isn't supplied with the hardware, contact the vendor or manufacturer to obtain a Windows NT driver.

Drivers for Legacy SCSI Host Adapters.

Some less common hardware that was fully supported in earlier versions of Windows NT is now slightly less supported. You need a driver disk for older SCSI host adapters, but you can build the driver disk from files supplied on the Windows NT Server 4.0 CD-ROM. The following SCSI host adapters require you to create a driver disk:

To create the driver disk, format a diskette and copy all the files from the driver folder to this diskette. The CD-ROM driver folder is one of the following:

Label the diskette as the driver diskette for retired storage drivers for Windows NT Server 4.0, and keep the diskette ready to use during the installation.

Fixed-Disk Drive(s).

Windows NT Server supports SCSI and IDE devices (primarily fixed-disk drives, CD drives, and tape drives), as well as other mass-storage devices, including the SCSI RAID drive arrays discussed in Chapter 7, "Setting Up Redundant Arrays of Inexpensive Disks (RAID)." Windows NT Server automatically detects mass storage devices during the installation processes and assigns these devices logical drive letters, beginning with C.

As part of the installation process, you can partition your fixed-disk drives, or ask Windows NT Server to respect the partitioning you've already done. After the drives are partitioned, you install a file system on each partition. When you partition a drive, it appears to the operating system as two or more smaller drives, usually called volumes. The first partition of each physical fixed disk is assigned a drive letter in sequence; the remaining partitions are then assigned drive letters.

There are two main reasons for partitioning a drive: it's to be accessed by more than one operating system, or you want to use drive letters as a convenient way to organize the drive. Partitioning a server drive for dual-boot operation is uncommon, because dual-booting servers isn't a recommended practice.

Some examples of partitioning for organizational reasons include the following:

If your drive is already partitioned and you plan to use the files that are kept on it, maintain the existing partitions. You can't repartition a drive without losing the information stored in the partition. If the first partition is smaller than 150M or the volume is compressed, you won't be able to maintain the existing partitions and will have to back up the files, repartition the disk, and then restore the files into the new partitions.

Each partition has an assigned file system, a method of storing and organizing files. The following two file systems can be used with Windows NT Server 4.0:

Microsoft has announced that a 32-bit version of the FAT file system, called FAT32, will be available for installation by suppliers of PCs that preload Windows 95 by late 1996. When this book was written, Microsoft stated that FAT32 wasn't intended for retail distribution with Windows 95 and won't be supported by Windows NT. The primary reason for providing FAT32 is to overcome the 2G partition size barrier and to allow the use of a smaller cluster size in order to reduce the percentage of slack on large fixed-disk drives. NTFS has neither the 2G partition limit nor the slack problem.

The C: volume, the active system partition on the first internal hard disk, is the Windows NT system partition. The system partition contains a specific set of files used to start Windows NT Server and can't be compressed. The system partition can't be part of a volume set or a stripe set. If this restriction poses a problem for you, set aside a 150M system partition; the rest of the drive can be partitioned more flexibly.

If you won't be using any other operating system and don't have any organizational reasons for partitioning a single drive, make each drive one partition that uses NTFS. If your drive contains files you want to use after the installation, don't change the partitioning or the file system, but convert the file system to NTFS after the installation.

It's possible to convert the file system to NTFS as part of Setup and preserve all your existing files. However, if you quit Setup without completing all the steps, the conversion won't be done. It's less confusing-and usually quicker-to leave the file system alone until Setup is complete, and then convert it to NTFS.

To convert a FAT (DOS, Windows 3.x) or HPFS (OS/2) partition to NTFS and preserve all the files on it, open a Command Prompt window (sometimes called a DOS box) from the Start menu by choosing Programs and MS-DOS Prompt. Type the command convert d: /fs:ntfs (where d: is the drive you want to convert) in the text box and press Enter. If you try to convert the system partition (typically C), you'll be warned that the Convert program can't get exclusive access, but that the conversion can be scheduled as part of the next restart. Choose Yes, and then restart Windows NT Server to proceed with the conversion. Several additional restarts are required before the conversion completes.

CD-ROM Drive.

Installing Windows NT Server 4.0 requires access to a CD-ROM drive, preferably with a SCSI interface. You don't need one of the high-speed 6X or 8X drives designed for multimedia applications; a 2X drive is quite adequate for installing server software, including Microsoft BackOffice components. It's possible to install Windows NT Server 4.0 over the network from a CD-ROM drive of another machine, but such a process is cumbersome.

Printer.

You must know the make and model of your printer, and the port to which it's connected. Check the hardware compatibility list to verify that Windows NT includes a driver for your printer. If you aren't ready to install your printer during Setup, you can add it later by using the Printers tool under My Computer.

Network Adapter.

The Setup software detects network adapter cards automatically, but it asks you to confirm the IRQ number, I/O base port address, memory buffer address, and other network card settings. If you installed the network card yourself, you have the required information that you need to accept the proposed values or enter new values. If you didn't install the network card, be sure to obtain the required network card settings from your computer vendor or the technician who installed the card.

If the PC has a network card, you can use Control Panel's Network tool or Windows Setup to display the current IRQ and I/O address range settings for your network card. Figure 6.1 shows the Resources page of the property sheet for an Intel EtherExpress 16/16TP network card. Click the Advanced tab to display additional settings for the network adapter. Windows 3.1+'s Network Setup dialog leads to dialogs that display similar information.


6.1

The Resources page of the property sheet for a typical network adapter displaying the current IRQ and I/O address range.

Providing Names and Identification

During the installation process, you must provide the following names and numbers:

The computer name must be unique on your network-it must not match a group or domain name on your network. The computer name also can't be longer than 15 characters. Make sure that you know the names of the computers, domains, the groups on your network, and the way that names are assigned. In some environments, computers are named for the person or department using them, for the function they serve, or according to a more whimsical pattern, such as names of flowers or precious stones. Make sure that the name for this server makes sense in your environment. If you aren't sure about it, contact the person who named the other machines to confirm your choice of a server name. If your entire network is new, take a moment to think of a logical scheme for naming machines on the network.

Choosing to Install as a Domain Controller

A domain is a group of computers that have the same administrative rules-for example, all the computers that belong to one department. There may be several domains on a network. Using domains makes administration much easier, because you don't need to set up an account for each user on each server. One machine, the Primary Domain Controller (PDC), maintains the domain account database. Other machines on the network that don't have copies of the account database are domain servers, also called plain servers. Computers with copies of the account database are called Backup Domain Controllers (BDCs). Chapter 16, "Distributing Network Services with Domains," discusses in detail the architecture of Windows NT Server domains.

If you plan to set up multiple Windows NT servers, you must decide which server acts as the PDC, which server(s) act as the BDC(s), and which server(s) act as the domain servers. (This decision isn't easily reversible-in fact, making some changes of domain status requires a full reinstallation of Windows NT Server.) For new installations, you install Windows NT Server 4.0 as the PDC first, and then add BDCs and domain servers while the PDC is operating and connected to the network.

When you install Windows NT Server on the first server of a domain, you must create a PDC and name the domain. Choose a name that isn't being used by any existing domain. Users of all client computers connected to the network must know the name of the domain controller(s) to which each client connects.

Backing Up Data on an Existing Computer

Before you undertake a step as significant as changing the operating system (or reinstalling your current operating system), you must back up the data on your system. Chapter 8, "Installing File Backup Systems," shows you how to organize backups and use the Windows NT backup system with tape drives. If the machine stores vital data, the safest approach is to make two backups, each on different media. For example, back up to tape, and then copy the crucial files over the network to a different hard drive in case the tape is unreadable. In many instances, backup tapes are found to be unreadable, either because of a defect in the tape or tape drive, or because the format of the tape is specific to the prior operating system. Another problem, mentioned earlier in the chapter, is the current lack of Windows NT 4.0 drivers for many low-end backup tape drives. If your computer is already running Windows NT and is a Primary Domain Controller, make sure that the Backup Domain Controller has an up-to-date copy of the domain account database.

After the backup, if the drive is compressed, you must decompress it. Whether the drive is compressed with Stacker, DoubleSpace for DOS 6, or DriveSpace for Windows 95, you must undo the compression before you try to install Windows NT Server 4.0 on the drive. Don't try to decompress before the backup; one of the disasters the backup ensures against is data loss during decompression. After you decompress the drive, however, back up the decompressed files that you want to preserve.

If the drive is compressed, it's because not all the files fit without compression. Thus, it's likely that the first step in the decompression process is to remove some of the files temporarily, preferably by moving the files to another server. Moving files without inconveniencing users of vital data isn't a simple task and requires advance planning. For example, users must be warned to save and close files by a certain time, and the files must be unavailable for a short time after that. Don't forget to clear at least 148M (preferably 200M) of free drive space to install Windows NT Server 4.0.

In addition to the 148M of free drive space, you need a blank 3 1/2-inch diskette that's labeled NT Server Emergency Repair Disk and is further identified with the name of the Windows NT server. You don't need to format the diskette; the Setup program formats the diskette for you. If your A drive accommodates 2.88M diskettes, use an unformatted diskette or a diskette formatted to 1.44M (don't use a diskette preformatted at 2.88M).

Starting the Basic Installation

After you make your decisions, gather the necessary information, and prepare your computer, you're ready to run Setup. The setup process involves many steps and a substantial number of Press Enter to Continue prompts. Most of the decisions don't need detailed explanations, but some steps are critical for a satisfactory installation. Following is a summary of steps that comprise the installation process:

  1. Create boot diskettes, if necessary.
  2. Start Setup from the boot diskettes, a network share, or from an existing operating system.
  3. Choose to install or repair.
  4. Detect mass storage.
  5. Choose to upgrade or install from scratch.
  6. Confirm basic system information.
  7. Set up your fixed-disk drive(s).
  8. Watch the copying process.
  9. Identify the user, company name, and licensing terms.
  10. Choose the type of domain controller.
  11. Select the locale.
  12. Set the administrator password.
  13. Check for the Pentium floating-point division bug.
  14. Fine-tune the installation.
  15. Join the network.
  16. Finish Setup.
  17. Configure virtual memory.
  18. Install IIS.
  19. Set the time and date.
  20. Configure your display.
  21. Create the emergency repair disk.
  22. Restart the computer for the last time.

The following sections describe each of these steps in detail.

Running the Setup Program

If you're installing Windows NT Server 4.0 on a computer with a CD-ROM drive supported by Windows NT Server, insert the first Setup diskette, labeled Setup Boot Disk, in the A drive; insert the Windows NT Server CD-ROM in the CD-ROM drive; reboot the computer; and skip to the section "Choosing to Install or Repair." Setup uses diskettes for the first part of the installation procedure, and then detects the CD-ROM drive automatically and uses the CD-ROM drive for the rest of the installation.

The following sections pertain to installing Windows NT Server over the network from another server or from a CD-ROM drive that isn't supported by Windows NT Server but is supported by DOS or Windows 95.

Creating Setup Boot Disks.

The Windows NT Server 4.0 distribution CD-ROM is bootable, but relatively few CD-ROM drives and/or system BIOSes support bootable CD-ROMs. To install Windows NT Server 4.0 on most PCs with a supported CD-ROM but no operating system installed, you must use Setup boot disks to install a minimal version of Windows NT on the PC. If you have the distribution CD-ROM but no boot disks, have three formatted, blank 3 1/2-inch diskettes ready and follow these steps to create the required boot disks:

  1. Insert the distribution CD-ROM in the CD-ROM drive of a PC running DOS, Windows NT, or Windows 95. If the Setup splash screen appears, close it.
  2. At the command prompt, log on to the CD-ROM drive and change to the \I386 folder (or the folder for the appropriate platform).
  3. Type winnt /ox (DOS or Windows 95) or winnt32 /ox (Windows NT) and press Enter to start the Setup program.
  4. Accept or correct the path to the CD-ROM files, and then press Enter or click Continue. Windows NT displays the dialog shown in figure 6.2.

    6.2

    The Windows NT 4.00 Server Installation/Upgrade dialog for creating Setup boot disks.
  5. Insert a diskette labeled Windows NT 4.0 Server Setup Disk #3 in the A drive and click OK to continue.
  6. When prompted, insert the remaining two diskettes, Windows NT 4.0 Server Setup Disk #2 and Windows NT 4.0 Server Setup Boot Disk, and press Enter or click OK.

The diskettes are necessary if your Windows NT installation becomes corrupt, requiring use of the emergency repair disk created later in the installation process. Leave the Setup boot disk in the A drive and restart your computer, booting from the A drive. Skip to the "Choosing to Install or Repair" section to continue with the description of the setup process.

Installing from a Network Server.

Installing over the network is more common for workstations, which may not all have CD-ROM drives, than for servers. You must have an operating system installed and a functional network connection to install Windows NT Server from a network server. To install Windows NT Server 4.0 over the network, you need another machine on the network with either the Windows NT Server CD-ROM in a shared CD-ROM drive, or a prepared Windows NT Server installation folder with a copy of one of the three folders from the CD-ROM:

To prepare an installation folder for network setup of Windows NT Server, follow these steps:

  1. Create a folder on the server with an appropriate name, such as Installnt.
  2. Share the folder, giving the Administrator group at least Read access.
  3. Create a subfolder with the same name as the one you copy from the CD-ROM-for example, \I386.
  4. Copy the installation files from the CD-ROM's subfolder to the new installation subfolder.

To start Setup from the new computer after the Windows NT installation folder is created, or to run from an unsupported CD-ROM drive, you follow almost identical steps:

  1. Label four blank 3 1/2-inch diskettes as Setup boot disk, Setup disk 2, Setup disk 3, and NT Server Emergency Repair Disk. Add the name of the computer to the emergency repair diskette. You can use the other three diskettes for installation on other computers; the emergency repair diskette is specific to the computer on which the installation is made.
  2. Start the machine where Windows NT Server is to be installed by using the existing operating system, which must either support an attached CD-ROM drive or have a connection to the network.
  3. If the machine is using DOS, change to the network drive and directory or to the CD-ROM drive and directory that holds the installation files. If the machine is using Windows 95, open an Explorer window for the network or CD drive. If the machine is using a previous version of Windows NT, open a File Manager window for the network or CD drive. (If the machine runs Windows 3.1x, exit Windows, if necessary, and perform the installation from DOS.)
  4. From DOS, run WINNT.EXE by typing winnt and pressing Enter. From earlier versions of Windows NT, run Winnt32.exe by double-clicking the file in File Manager. From Windows 95, run Winnt.exe by choosing Run from the Start menu, typing winnt, and then pressing Enter.
  5. When requested, provide the drive and directory name for the location of the installation files.
  6. Insert each formatted setup diskette when prompted. The WINNT or WINNT32 program copies the required setup files to the diskettes.
  7. Continue the setup process as described later, starting with the section "Choosing to Install or Repair."

Installing from an Existing Operating System.

If you have a compatible CD-ROM drive and a compatible operating system, you can start the Setup program from DOS or Windows as follows:


6.3

The Windows 95 and Windows NT splash screen for installation from the operating system.

In either case, you're prompted to confirm the location of the files on the CD-ROM; then Setup copies all the setup files to a temporary folder, requests you to reboot, and begins the installation process. You need approximately 250M of free disk space to store the installation copies and the working copies of the files. When the files are copied, remove the CD-ROM from the drive (and the diskette, if any, in your A drive), and reboot the computer. On restarting, the Boot.ini file automatically starts the Windows NT setup process.

Choosing to Install or Repair

If you use the setup diskettes, Setup runs from the setup boot diskette, requests a second diskette, and then displays a blue (DOS) screen titled Windows NT Server 4.0 Setup: Welcome to Setup, which asks whether you want to repair an existing Windows NT Server installation or continue with a full installation.

At any point in the setup process where you're prompted for input, pressing F1 provides Help. Pressing F3 exits Setup. Input options are summarized at the bottom of the screen.

The first part of this chapter covers installing Windows NT Server 4.0 for the first time, so press Enter to continue the installation process. You're requested to insert Setup Disk #2 and press Enter.

One of the last setup steps is building the emergency repair diskette, which you use to repair a damaged or corrupted installation of Windows NT Server. The repair process is described in the "Repairing the Windows NT Server Operating System Installation" section near the end of this chapter.

Detecting Mass Storage

The mass storage devices (SCSI and IDE adapters) on your system ordinarily are detected automatically during the setup process, and Setup automatically installs the required driver(s). You have the option of skipping the automatic detection process and selecting the driver yourself. The automatic detection process is somewhat slow, so if you're installing Windows NT Server on a number of similar machines, you might want to select the mass storage driver manually.

If your device isn't on the supported hardware list, the automatic detection process won't find it. Also, some hardware combinations cause the automatic detection to hang the Setup program. If Setup hangs, reboot the computer, start Setup again, and bypass the automatic detection process. Allow the automatic detection at least five minutes before assuming it has hung.

Press Enter to proceed or S to skip the automatic detection process. After Setup detects hardware and loads drivers, you're given an opportunity to confirm the hardware that was detected and to add more mass storage devices, if necessary. If you press S to add devices manually, a list of drivers known to Windows NT appears; choose a driver from the list. If you have a driver disk, choose Other and insert the disk when prompted. When all the drivers are loaded, press Enter to continue.

IDE and ESDI drives are detected at this point too, but not shown to you. You set up your fixed-disk drives later in the Setup process.

After selecting mass storage devices, the Windows NT End-User License Agreement (EULA) appears. To read the license agreement, press Page Down 13 times, and then press F8 to accept the terms of the agreement. If you don't press F8, installation terminates.

You're then requested to insert the distribution CD-ROM into the drive and press Enter.

Choosing to Upgrade or Install from Scratch

If an earlier version of Windows NT is detected on your hard drive, you have the opportunity at this point to upgrade, preserving as many of your old settings as possible. If you choose to upgrade by pressing Enter, skip to the "Watching the Copying Process" section. If you choose to install from scratch by pressing N, or if a previous version wasn't installed on your hard drive, the installation continues as described in the following section.

Confirming Basic System Information

Setup reports the basic computer hardware it has detected and asks for your confirmation of its findings. You may want to change the keyboard layout at this point (if, for example, you're using a Dvorak or other alternative layout), but the balance of the reported information is almost invariably correct. To change an entry, use the up and down arrows to highlight it, and then press Enter. A list of choices appears; choose the one you want with the up and down arrows and the Enter key. When the hardware summary is correct, highlight No Changes and press Enter to continue.

Setting Up Your Fixed-Disk Drives

After you watch the detection of mass storage devices, add more drivers if needed, and then confirm your hardware, you next must make Windows NT Server work with your fixed-disk drive(s). This involves the following basic steps:

  1. Partition the drive(s).
  2. Format with the right file system.
  3. Specify the installation folder.

The following sections describe each of the preceding steps in detail.

Partitioning the Drive(s).

The first partition is the system partition. On x86-based machines, the system partition is drive C. If you have more than one partition, make sure that the first one (the C partition) has sufficient room for the system files (about 148M).

Setup shows the partitions that already exist on your fixed-disk drive. Use the up and down arrow keys to highlight a partition or the unpartitioned space; then press D to delete it, C to create a partition in unpartitioned space, or Enter to choose it as the system partition.

Formatting with the Right File System.

Next, you choose a file system and optionally format the partition with that system. You can choose to format the partition with FAT or NTFS; keep in mind, however, that these two options destroy the data stored on the partition. If the partition is already formatted, you can elect to convert the partition to NTFS or retain the existing format. As noted earlier in the chapter, unless you have a compelling reason to use the FAT format, choose NTFS.

Use the up and down arrows to highlight your choice (for example, Convert to NTFS) and press Enter to continue.

Specifying the Install Folder.

Setup now needs to know where to put the Windows NT and system files. The default suggestion, C:\WINNT, ordinarily is satisfactory. You can, however, install these files in a folder of any name on a partition of the first physical (boot) drive. To change the folder name, use the Backspace key to remove the suggested name and type your chosen name.

At this point, Setup looks for previous versions of Windows on your machine. (If you formatted your system partition in the previous step, you wiped out any previous versions that were on the machine.) Rather than ask you to choose whether to upgrade or dual boot, Setup determines the answer based on whether you use the same folder. Following are your options when you choose the same or a different installation folder:

Watching the Copying Process

Now that Setup can access your partitioned and formatted drive(s), and has established where the system files are to go, it's time to copy the files to the fixed disk. Before doing so, Setup offers to examine your fixed-disk drive for defects. Press Enter to allow the examination to proceed, or Esc to skip it. (If you're having trouble installing Windows NT Server and find yourself at this point in the install repeatedly, it's not necessary to repeat the examination every time.) After the examination is performed or skipped, Setup copies files to the folder chosen in the previous step. The time required to create the copies depends on the speed of the CD-ROM or network connection, and the performance of your fixed-disk drive.

An error message during the copying process that indicates a checksum error between the file on the CD-ROM and that of your fixed-disk drive indicates a hardware problem. The most likely source of the problem is a damaged CD-ROM surface, dirt on the CD-ROM drive's optical components, a defective fixed disk, or a defective disk controller. Another less likely source of checksum errors is improper termination of a chain of SCSI devices. Improper SCSI termination usually results in an installation failure before the copying process starts.

Recovery from such an error depends on what happened and why. You may be able to correct the problem and choose Retry; more than likely you'll have to begin the Setup process again. If the CD is damaged, you must replace it.

When the copying process is complete, Windows NT Server is ready to run, but additional configuration information is required to finalize the installation process. At this point, you're prompted to remove a diskette still in the A drive. Also remove the CD-ROM. Press Enter to restart the system.

After the computer restarts, the rest of the process proceeds under the operating copy of Windows NT Server. Setup's simple character-based interface is replaced with Windows NT-style dialogs, and you indicate choices by clicking dialog buttons instead of pressing keys. The Help and Exit Setup options remain available, but they're activated by dialog command buttons. Also, you use the Back and Next buttons to move through the Setup Wizard (as Microsoft calls this next stage of Setup). You're requested to insert the CD-ROM into the drive indicated in the Copy Files From text box, and then click OK to copy the additional files. After the files are copied, click Next to continue.

Identifying the User, Company Name, Licensing Terms, and Computer Name

Provide the full (first and last) name of the main user (usually you) and the company name, as discussed earlier in this chapter, if you aren't upgrading an existing Windows NT installation. Click Next to continue.

You must type the product ID from the inside back cover of the installation guide or from the sticker on the Windows NT Server 4.0 CD-ROM's case. Click Next to continue. If you're upgrading an existing Windows NT Server installation, Setup copies files; skip to the "Specifying Creation of an Emergency Repair Disk" section later in this chapter.

The next step asks you to choose your licensing method-Per Server or Per Seat. Per Server licensing requires you to have a client license for every simultaneous connection to this server; Per Seat requires you to have a client license for every client machine in your installation. Which is better for you depends on how many clients and servers you have, and how many servers each client connects to at once. Here are some examples:

After selecting the licensing method, click Next to enter the computer name for the server, which is limited to a maximum of 14 characters. The name you assign at this point appears to all other computers on the network. Press Next to continue.

See "Licensing BackOffice Components," (Ch 21)

Choosing the Type of Domain Controller

The Server Type dialog offers you three options:

You must install the Primary Domain Controller before any other domain servers; this decision is difficult to reverse. The first Windows NT server you install in a network always is a Primary Domain Controller. Click Next to continue.

If this is the first server on your network, you're creating a new domain, so you name the domain at this point. (You should have already picked out a unique domain name.) If the domain already has a Primary Domain Controller, you can join the domain as a Backup Domain Controller. Make sure that the Primary Domain Controller is operational and that you know the administrator password for the domain. If you choose to install this server as a stand-alone server, the domain name is simply the domain you join.

No matter what server type you choose, Setup searches the network for existing domain names. For a Primary Domain Controller, Setup searches to verify that the new name is unique. For a secondary domain controller or a stand-alone server, Setup ensures that the domain name you supply exists.

Setting the Administrator Password

The Administrator account is used to manage this installation of Windows NT Server. Someone who knows the Administrator password can add and delete users, install and remove applications, and make any other system changes that might be required. When you install a Primary Domain Controller, you set this password. Choose a sensible password that's hard to forget yet difficult to guess. You enter it twice in this dialog; both entries must be the same, to rule out a slip of the fingers as you type it.

Dealing with the Pentium Floating-Point Division Bug

If your machine is an Intel Pentium-based computer, Setup checks for the known floating-point division bug. If you have a faulty chip, you're given the option of turning off the Pentium's floating-point operations and simulating them within NT instead. Although the simulated calculations are much slower, they're always right, so you should choose to disable the hardware floating-point calculations. If your Pentium chip doesn't have a faulty floating-point module, you won't see this screen.

Specifying Creation of an Emergency Repair Disk

At the end of a successful setup, Windows NT Server saves the computer's configuration information to the disk so that if your fixed-disk drive becomes corrupted, you can recover your configuration information. At this point in the Setup process, you're asked whether you want to make an emergency repair disk later. Doing so is strongly recommended. Using the emergency repair disk is covered later in the "Repairing the Windows NT Server Operating System Installation" section.

Fine-Tuning the Installation

If the computer is to be a dedicated server rather than a workstation, you can save disk space by skipping the installation of accessories such as CD Player, games, screen savers, wallpaper, and so on.

If you've installed Windows NT Server 4.0 on another computer, you can eliminate the Readme files from this installation. Before making this election, however, make sure that you have access to the Windows NT Server 4.0 Readme files on at least one computer.

Choose the components you want installed or not installed in the Select Components dialog. If you want part of a component, such as Multimedia, installed, click the Details button to select individual applications. Click OK on the Details dialog to return to the list of components. When that list has each component selected, unselected, or partially selected as you prefer, click Next to move to the next stage of Setup.

Joining the Network

Now all of Setup is complete, except for setting up the network. You confirm you want to go on to the network portion of Setup by clicking Next. You can't get back to earlier screens after moving to the network portion. You perform the following general steps for the network installation:

  1. Describe your connection to the network.
  2. Choose to install the Internet Information Server.
  3. Choose and configure a network adapter.
  4. Choose protocols.
  5. Choose services.
  6. Confirm network bindings.
  7. Start the network and join a domain.

Describing Your Connection to the Network.

The first question Setup asks is how you connect to your network. If your machine has a network adapter card (as it almost certainly does), choose Wired to the Network. If you dial up to a network (an unlikely choice for Windows NT Server), choose Remote Access to the Network.

Choosing to Install the Internet Information Server.

The Internet Information Server makes your information available over the Internet or a corporate intranet, and is discussed fully in Chapter 19, "Setting Up the Internet Information Server." At this point in the Setup process, you specify only whether you intend to install it.

Choosing and Configuring a Network Adapter.

Assuming that you have one or more network adapters, in the next step Setup detects them automatically. A functioning network adapter card is required for installation of a Primary Domain Controller or a Backup Domain Controller, but the network doesn't need to be operational at this point in the installation process.

Click Start Search to find the first adapter. If a second adapter is to be found, click Find Next to search for it. If you have an adapter that wasn't found, click Select from list to specify the card yourself. Click Next to move on.

Choosing Protocols.

Next, choose one or more network protocols: IPX/SPX, TCP/IP, or NetBEUI. (This decision is discussed in detail in Chapter 4, "Choosing Network Protocols.") You can configure multiple protocols by checking more than one box in this dialog. The most common combination is NetBEUI and TCP/IP, unless you have an existing Novell NetWare network, in which case you should select all three protocols at this point. You can add or remove network protocols with Control Panel's Network tool after installing Windows NT Server.

Choosing Services.

You can choose to install any of these five network services:

Adding new services after the fact is harder than adding them now, and you can use Control Panel's Services tool later to disable services you don't want to run. You also can add or remove network services with the Services page of Control Panel's Network tool (see fig. 6.4). For most server installations, it's best to install all the preceding services at this point.


6.4

Adding or removing network services in the Services page of the Network property sheet.

Even more services are available if you click Choose From List:

Network Settings.

Setup confirms that you're ready to install the adapters, protocols, and services that were selected over the previous few dialogs. Click Next, and you have the opportunity to confirm adapter settings, such as the interrupt (IRQ) number and I/O port address (see fig. 6.5). As discussed earlier in the chapter, you should know these settings before you start the installation procedure.


6.5

Setting the interrupt number, I/O port address, and other properties for your network adapter.

The Network Settings property sheet is next. You can display the same properties by using the Protocols page of Control Panel's Network tool (see fig. 6.6) after you install Windows NT Server 4.0. To configure a protocol, select it from the list on the Protocols page and click the Properties button. If you're not sure how to use this property sheet, you can leave it for now and bring it up again after Windows NT Server is completely installed.


6.6

The Protocols page of the Network property sheet for adding and configuring networking protocols.

Each network protocol is configured separately. Figure 6.7 shows the configuration dialog for TCP/IP, the Microsoft TCP/IP Properties sheet.


6.7

The IP Address page of the Microsoft TCP/IP Properties sheet.

On the IP Address page, set the IP address and subnet mask for your machine, or tell Windows NT to use DHCP (Dynamic Host Configuration Protocol) to assign an IP address dynamically. If you didn't establish these settings during your network planning process, get them from the person who did.

IP addresses uniquely identify machines on TCP/IP networks, such as the Internet, and are specifically assigned. You can't just arbitrarily choose an IP address; you must use an address that makes sense both within your network and, if applicable, on the Internet. The subnet mask is used to distinguish between IP addresses on your network and those that aren't. Like the IP address, the subnet mask is determined during the network planning process. The default gateway, if specified, must also be determined by a network administrator.

The DNS (Domain Naming System) page is used to control the way the server looks up domain names of other computers on a LAN or WAN (see fig. 6.8). You enter the IP addresses of one or more DNS servers accessible to your server. DNS servers translate a fully qualified domain name such as http://www.mcp.com. into an IP address. You can get these addresses from the same person who provided your server's IP address. Some networks use multiple DNS servers, checking the local one first and then asking a remote DNS server if the name wasn't found in the local one. To change the priority of an IP address within the list, click the Up[ua] or Down[da] buttons.


6.8

The DNS page of the Microsoft TCP/IP Properties sheet.

See "Understanding WINS and DNS Name Resolution," (Ch 15)

The WINS Address page describes the way that WINS (Windows Internet Naming Service) looks up the domain names of other computers (see fig. 6.9). Enter the IP addresses of your primary and secondary WINS servers as provided by the person who told you your own IP address. For machines that are local to your network, you may want to use DNS and LMHOSTS services; if so, select their check boxes. The Scope ID is usually left blank; provide a Scope ID only if you're told to by your network administrator.


6.9

The WINS Address page of the Microsoft TCP/IP Properties sheet.

See "Windows Internet Naming Service (WINS)," (Ch 17)

The DHCP Relay page identifies your DHCP servers. These servers manage IP addresses within your internal network, and you get their addresses from the person who told you to use DHCP rather than specify the IP address of your machine. If you need to adjust the other parameters on this page rather than accept the defaults, this person will tell you so.

See "Dynamic Host Configuration Protocol (DHCP)," (Ch 17)

The Routing page is relevant only to machines with more than one network adapter with more than one IP address. If you turn on IP forwarding, your server can route traffic between the two networks.

Confirming Network Bindings.

Setup gives you the opportunity to adjust your network bindings at this point. If you aren't sure what network bindings are or why you might want to adjust them, leave them alone. The default bindings usually are adequate for installation and network startup.

You can enable and disable communications between services and adapters or protocols by double-clicking the service name, clicking the adapter or protocol, and then clicking the Enable or Disable button.

Starting the Network and Joining a Domain.

After Windows NT is configured for your network hardware, Setup loads the network software and establishes a connection to the network. Confirm that you've made all the choices by clicking Next, and then wait while the network starts. Setup asks for a domain name. Provide the same domain name you used earlier, and wait while Setup searches the network. If you're joining a domain but not installing a Primary Domain Controller, provide the Administrator name and password here.

Finishing Setup

Click the Finish button to move to the final Setup steps. Setup creates Program menu groups and desktop icons, if you had an earlier version of Windows installed. Adding groups and icons doesn't install the applications or make any changes to the settings and configurations stored in Windows 3.1+'s .INI or REG.DAT files, or in Windows 95's Registry. You must rerun the application's Setup program if the application must be reconfigured for Windows NT or if you're upgrading from Windows 95.

To finish up the Setup process, you need follow these general steps:

  1. Install Internet Information Server 2.0.
  2. Set the time and date.
  3. Configure your display.
  4. Create the emergency repair disk.
  5. Restart the server for the last time.

Installing Internet Information Server

The dialogs and decisions involved in installing IIS are covered in detail in Chapter 19, "Setting Up the Internet Information Server." To install only World Wide Web services, clear the Gopher Service and FTP Service check boxes. Click OK to continue. Accept the default folders for the service(s) you install, unless you have a specific reason to do otherwise. Click OK to continue and then click OK again when asked to confirm your choices. Setup copies a number of files to your drive. Click OK when advised that you must establish an Internet domain name for the server.

In the Install Drivers dialog, select SQL Server in the Available ODBC Drivers list and click OK to continue.

Setting the Time and Date.

You now get a chance to set the time, date, and time zone. On the Time Zone page, select your time zone from the drop-down list. This list is arranged numerically according to the time difference from Universal Time (UT, formerly known as Greenwich Mean Time, GMT). Zones east of Greenwich, England, appear above it in the list. After you choose your time zone, the map of the world in this dialog scrolls so that your time zone is in the center. Windows NT Server knows the rules for use of Daylight Savings Time; make sure that the Automatically Adjust for Daylight Savings Time box is marked or cleared, as appropriate for your location.

On the Date & Time page, set the current date and time with the spin controls.

After Setup is complete and Windows NT Server is running, you can bring up this property sheet again with the Control Panel's Date/Time tool (see fig. 6.10).


6.10

Using the Date/Time tool to set the time zone, and the current time and date.

Configuring Your Display.

Until now, the setup program has used standard VGA resolution (640[ts]480 pixels) with 16 colors. Windows NT attempts to detect the type of chip on your graphics adapter card. If you're using an adapter with a popular Windows graphics accelerator chip for which a DirectDraw driver is included with Windows NT, click OK when the Detected Display message box appears. Otherwise, you must install a driver from a diskette provided by the graphics card supplier.

The Settings page of the Display Properties sheet (see fig. 6.11) lets you set a resolution and color depth suited to the combination of your graphics adapter card and video display unit (VDU). Some adapter cards provide additional features and controls on the Settings page. For a simple adapter, follow these steps to set your display properties:

  1. Click the Display Type button to confirm that the correct graphics adapter card has been detected, and change the selection, if necessary.

    6.11

    Configuring the color depth and resolution of your display.
  2. Adjust the number of colors, resolution, and refresh frequency. For servers, a resolution of 800[ts]600 pixels and a color depth of 256 colors is adequate. If you have a 15-inch or smaller display, the Large Fonts selection improves readability in 800[ts]600 resolution.
  3. Click the Test button to examine the results of the settings you choose.
  4. Click OK to close the Display Properties sheet.

Creating the Emergency Repair Disk

At this point, Windows NT Server 4.0 is installed, configured, and ready to act as a network server. The final step is to create an emergency repair disk to use in case of a catastrophic failure. Any 3 1/2-inch diskette will suffice, because Setup formats the diskette before copying the files. Label the emergency repair disk with the server name you assigned, and store it in a safe location.

Be sure to update your emergency repair diskette frequently to keep the configuration data up-to-date. Your server's configuration is likely to change appreciably during the first few hours of use as you install applications, change users, and groups, and so on. The pace of change slackens with time, but you should make a habit of updating your emergency repair disk regularly. To create an updated emergency repair disk, follow these steps:

  1. Insert your original emergency repair disk or a new diskette in drive A. From the Start menu choose Run, type rdisk in the Open text box, and click OK to open the Repair Disk Utility dialog (see fig. 6.12).

    6.12

    The Repair Disk Utility dialog.
  2. Click Update Repair Info to update the diskette's content. (If you've lost your emergency repair disk, insert a blank diskette and choose Create Repair Disk.)
  3. Confirm your intent to update the existing emergency repair diskette (see fig. 6.13). Click OK when asked if you want to create an emergency repair disk, and then click OK to format the diskette and create the updated version.
  4. After the updated emergency repair disk is created, click Exit to close the Repair Disk Utility dialog.

    6.13

    Confirming the update operation.

It's a generally accepted practice to create a duplicate of the emergency repair diskette for off-site storage, along with duplicates of driver disks you used during Setup. Driver diskettes also are needed during the repair process.

Add an entry to the Administrative Tools Program menu for Rdisk.exe to make the program easier to run after every significant configuration change.

Restarting the Server for the Last Time

After you make the emergency repair disk, Setup completes all the tasks required to install and configure Windows NT Server 4.0 and automatically restarts the server. Remove the emergency repair disk from the A drive and remove the CD-ROM to allow the system to boot from the fixed-disk drive. (The boot process takes more time if you converted a FAT partition to NTFS as part of the setup process, because it's at this point the format conversion occurs.) After Windows NT loads, press Ctrl+Alt+Delete to log on, and use the administrative account name and password you created earlier to log on to Windows NT Server. Your installation is complete.

Preparing the Server for Use

If you installed Windows NT Server onto a machine that was running Windows 3.1x, and you installed Windows NT into the same directory as before (typically, \WINDOWS), the Program Manager settings for a user (other than the administrator) are initialized. In this case, the Programs menu choices are based on the Windows 3.1+ settings that were present before Windows NT Server was installed. For example, if there was a program group called Invoices under Windows 3.1+, each new user sees a Programs menu group called Invoices with the same program items in it. Groups with names such as Main aren't migrated in this manner, because new Windows NT Server equivalents of these Programs menu groups were created earlier in the Setup process.

If the machine is to be a dedicated server, with no access by ordinary users, running existing applications is of no consequence. If the server is also used as a workstation (not a recommended practice), create local accounts for the workstation users. Log on to the other users' local accounts to oversee the migration of Windows 3.1+ applications and fix any problems that appear before releasing the computer for others to use.

See "Managing User Accounts," (Ch 12)

Repairing the Windows NT Server Operating System Installation

You can ruin your Windows NT Server software in various ways, and in some cases there's no way to recover. Making complete backups frequently is one way to reduce the recover work involved; having an emergency repair diskette is another. This section assumes that you have a problem that prevents your server from booting successfully. It further assumes that the Last Known Good choice during the boot process is of no assistance, and that you can't edit the Registry remotely from another computer to adjust your settings. The last resort is to reinstall Windows NT Server from scratch-but first try using your emergency repair disk.

Gather together (from off-site storage, if necessary) the emergency repair disk, originals or copies of the three Setup diskettes, any driver disks you used during the original installation, and the original CD-ROM used for installation. If any of these are missing, the repair process may be impossible. Follow these steps to attempt to repair the server:

  1. Correct hardware problems, if any, and install replacement hardware as needed.
  2. Boot from the Windows NT Server 4.0 boot diskette, switch to Setup Disk 2 when prompted, and select Repair at the first prompt. You can select one or more of these choices:
  3. Move the highlight up and down through this list with the arrow keys, and press Enter to select or deselect the highlighted item. If you have no idea what's wrong, leave all the items selected. When your list is complete, highlight Continue (perform selected tasks) and press Enter.
  4. Repair checks for mass storage devices using the same process as Setup. The steps and keystrokes involved are identical to those discussed in the "Detecting Mass Storage" section earlier in this chapter.
  5. When asked if you have the emergency repair disk, press Enter if you do and Esc if you don't.
  6. You're prompted for the original installation media (the CD-ROM) so that installed files can be compared to the originals.
  7. After a partial file examination process, you have a chance to restore Registry files. You should try at least one repair attempt without restoring Registry files. If that doesn't make the system bootable, and no backups are available, repeat the repair process and restore the Registry files. All changes made since the emergency repair disk was created, whether the installation of an entire application suite or a change to a user's desktop settings, will be lost.
  8. Repair examines the remaining files on your fixed-disk drive and compares them to files on the installation media. If a file is found that varies from the original, you're given four choices:
  9. When the Repair process is complete, you're prompted to remove any diskette still in the drive. Press Enter to restart the computer.

When a reboot is successful, you have the following problems to tackle:

If the Repair process doesn't work after several tries, you must reinstall Windows NT Server from scratch, and then reinstall applications as needed. In the event of serious corruption, you may lose all the user and group information for your server. A successful Repair operation takes about half as long as an install; always try a Repair first.

If you have another Windows NT server acting as a Backup Domain Controller, you can regain your user and group settings by promoting the BDC to a Primary Domain Controller, reinstalling Windows NT Server as a BDC, and resynchronizing the two domain controllers.

See "Promoting a Backup Domain Controller to a Primary Domain Controller,"
(Ch 16)

From Here...

After you have your server up and running, more work needs to be done before your server can be considered completely installed. The remaining chapters of this part of the book cover other aspects of setting up your server and the clients that connect to the server:

  • Chapter 7, "Setting Up Redundant Arrays of Inexpensive Disks (RAID)," explains how the RAID approach can make your system more reliable.
  • Chapter 8, "Installing File Backup Systems," explains how to protect the users on your network from personal mistakes, as well as from hardware trouble.
  • Chapter 9, "Understanding the Windows NT Registry," introduces you to the database that stores all the settings and configuration information you provide.
  • Chapter 10, "Configuring Windows 95 Clients for Networking," tackles the work involved in connecting Windows 95 workstations to your server.
  • Chapter 11, "Connecting Other PC Clients to the Network," describes the process of attaching machines running other operating systems to your server.
  • Two chapters in Part III, "Administering a Secure Network," and two chapters in Part IV, "Wide Area Networking and the Internet," also are relevant to a network administrator performing Windows NT Server 4.0 installations:

  • Chapter 12, "Managing User and Group Accounts," covers setting up users and groups on your server to make it usable.
  • Chapter 13, "Sharing and Securing Network Resources," describes some of the setting up that a successful network requires.
  • Chapter 18, "Managing Remote Access Service," covers the Remote Access Service, including installing it and getting it working.
  • Chapter 19, "Setting Up the Internet Information Server," describes the installation process to create a private intranet or to connect your server to the Internet.

  • Previous chapterNext chapterContents