Ask Dr. Bob Your Windows 2000 Questions - 15 Nov 1999


For answers to more of your NT questions, visit our online discussion forums at

If I want to add new Windows NT services to an existing installation, do I need to add them from the NT installation CD-ROM, then reload all service packs to update the core NT installation? Do I need to uninstall all service packs before to adding a new service?

You can add new services that require the NT installation CD-ROM at any time, and you don't need to remove a service pack before adding those services. However, you need to reapply the service pack whenever you use the installation CD-ROM to install new software. A service pack simply copies new versions of NT program files to the NT installation directories (after backing up the old versions if you've requested a backup). When you install a new service from the CD-ROM, the installation process might overwrite a file that the service pack has updated. Even more likely, the service pack includes updated versions of some files, but they don't copy to the computer when you install the service pack if the original files don't exist on the system.

To be safe, whenever you install new software that copies files from the original NT installation CD-ROM, reapply the service pack so that you know you're running the most recent versions of the programs. If you applied hotfixes after a service pack installation, remember to reapply the hotfixes if you reinstall the service pack. The service pack reinstallation most likely wrote over files that the hotfix replaced.

I don't recommend adding services just because you think you might use them in the future. Each service drains system resources, and some services have a significant impact on busy systems. For example, Network Monitor and the diskperf -y switch are two services that affect system performance, even when monitoring isn't active.

On my network, one PC runs Windows NT Server 4.0 and the other PCs run NT Workstation 4.0. The network's name is workgroup. When the server and workstations boot, the message Can't see workgroup appears. When the workstations aren't active, the server sees itself as workgroup. However, if a workstation is active, the server displays an error message. Any suggestions?

Your symptoms suggest that you've set up the server as a domain controller for the workgroup domain and the workstations as members of a workgroup also called workgroup. This conflicting use of the name workgroup explains why the server can boot alone but experiences an error upon boot when it finds a workgroup using its domain name.

I recommend you let the server be a domain controller, then join all the workstations to its domain. Alternatively, you can reinstall NT Server on the system so that the server isn't a domain controller but rather a member server. Then, configure the server to be a member of the workgroup called workgroup.

How can I determine the average bandwidth that Windows NT Server 4.0, Terminal Server Edition (WTS) client sessions use when I'm capacity planning?

In its "Terminal Server Capacity Planning" white paper (, Microsoft claims that you can use a figure of 2Kbps to 6Kbps per user to estimate the WTS client's impact on network utilization. However, this estimate isn't accurate for all customer environments. To get a more precise idea of how WTS will affect your network, you can use Network Monitor in conjunction with Performance Monitor.

On your NT server, go to the Services tab of the Control Panel Network applet and install Network Monitor Tools and Agent. This procedure adds the Network Segment object to Performance Monitor. Network Interface, the other object you'll need, installs with TCP/IP. You can use the newly provided counters under the Network Segment object to analyze the impact of your active sessions on the server. Specifically, you can use the Network Segment object's %Network Utilization counter, which displays the total bandwidth in use on a given network segment, and the Network Interface object's Bytes received/second and Bytes total/second counters, which provide the total number of bytes that the server's adapter receives or processes.

I've found that Network Monitor's %Network Utilization counter sometimes provides unreliable data, especially on Fast Ethernet/100Mbps networks. And NT Server's Network Monitor can monitor only the traffic to and from its adapter. If you want to more accurately monitor all traffic on a network segment, consider using an enhanced network monitor such as the Microsoft Systems Management Server (SMS) version of Network Monitor or a third-party product such as Data General's NetXRay.

I have a 10GB hard disk that holds Windows 98 on a FAT32 partition. How do I install Windows NT 4.0 to another physical 2GB disk and make it dual-boot on my computer?

I recommend using a third-party boot manager utility (such as V Communications' System Commander or PowerQuest's BootMagic) to manage your multiboot environment. These utilities create a special boot partition that becomes the launching point for booting multiple OSs from your various disk partitions. If you're managing a multi-OS environment, these products also offer the most flexibility and ease of use.

However, most third-party boot manager utilities require that you have an existing FAT12 or FAT16 partition within the first logical 2GB of the first hard disk. As a result, your existing disk configuration might not work. (A FAT32-formatted C drive isn't a suitable location for the NT boot manager, anyway.) You might need to repartition your existing disk scheme before you can install the boot manager. As with boot management, I recommend a third-party utility for disk partitioning.

I've been unable to install a particular application on a Windows NT workstation, even though the application installs on my Windows 95 machine with no problems. Why is this happening?

An inability to install applications on an NT workstation can occur for many reasons (e.g., conflicting DLLs, unsupported service packs). However, the most common reason is that the currently logged-on user, and the application installation utility running within that user's security context, has insufficient privileges to install the application. Typically, this situation results in an inability to write data to the HKEY_LOCAL_MACHINE portion of the Registry or to the system's NTFS volume. Try logging on as an administrator, then install the application.

I installed Windows 2000 (Win2K) beta 3 under the assumption that it was an upgrade to Windows 98. I performed a clean installation to an NTFS drive. Win98 resides on the primary FAT32 drive. Win2K didn't perform well with my 32MB of RAM, so I decided to erase Win2K and format the drive back to FAT32. How do I remove NT Boot Loader from my C drive and restore Win98's boot sector? How do I remove an entry from boot.ini?

I'm not surprised that you didn't like Win2K with 32MB of RAM. I've tested every version of Win2K through Release Candidate 2 (RC2), and I've found that you need at least 64MB (preferably 96MB or more) and a fast Pentium II processor (e.g., 400MHz or better) to get anything close to decent performance from a Win2K Professional (Win2K Pro) system.

You can remove NT Boot Loader by booting a Win98 boot disk that contains the Sys utility and running Sys C:. The contents of the boot.ini file in the system partition's root directory (i.e., C) control NT Boot Loader's choices. After you remove the read-only attribute from this file, you can edit NT Boot Loader's contents using any text editor. However, when editing boot.ini, be sure to keep the contents properly formatted or you could render your system unstartable. Screen 1 shows a sample boot.ini file. Microsoft provides several articles that describe the purpose and structure of the boot.ini file. I recommend the article "Purpose of the BOOT.INI File" ( kb/articles/q99/7/43.asp).

I'm repeatedly receiving a Kernel not found error message during the boot. The error renders my servers unrecoverable. I can't even get to the Windows NT Boot Loader screen. The problem forces me to reformat the disk and rebuild the server from scratch. I thought I might be experiencing a hardware failure, but the problem occurs across several servers. Any advice?

I suspect that something is altering the boot sector on your hard disk or another low-level disk structure. For example, you could attribute this event to software that has made low-level changes to the disk structure. If you're running a disk defragmenter (or other disk utility that automates repairs or maintenance on hard disks), try updating the software or temporarily disabling it to see if the problem goes away. However, assuming you aren't running software that makes low-level changes, I suggest you perform a thorough virus scan with the latest virus signatures on all your machines.

If your problem had occurred after the NT Boot Loader screen appears, then you'd probably need to look for a different cause. In that case, the problem would probably relate to the boot.ini file and the fact that something caused the enumeration order of the adapter, drive, or relative partition number to change on your system. This situation also invalidates the Advanced RISC Computing (ARC) pathname of the line in boot.ini that loads NT—for example, multi(0)disk(0)rdisk(0)partition(1)\winnt—and causes a black-screen failure that typically references an inability to load a kernel file.

Can you recommend a utility that examines and corrects a corrupted NTFS directory (e.g., a Norton Disk Doctor for NTFS)?

You can run Chkdsk, but you'll need to have an alternative method to access and run diagnostics on the volume. The best way to solve your problem is to use a utility such as Winternals' ERD Commander Professional Edition or simply move the disk into another system running Windows NT and run Chkdsk from there.

ERD Commander is helpful because it doesn't require you to move the disk. (Laptop hard disks with proprietary drive connectors are especially troublesome.) ERD Commander also lets you use a boot disk to run Chkdsk and perform other maintenance activities. Screen 2 shows a sample ERD Commander session.

Another Winternals utility, Remote Recover, lets you run the same diagnostics over the network. In addition to booting a special NT command/recovery shell from disk, Remote Recover loads network drivers so that you can administer the machine from anywhere on the LAN/WAN.

Editor's Note: Sean Daily and John Green contributed answers to this Tricks & Traps.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.