Send us your tips and questions and visit Bob Chronister's online Tricks & Traps at http://www.winntmag.com/forums/index.html.

In Office 97, connectivity is king. If you haven't set up an intranet or extranet (an intranet with connections from select clients), now is the time to gain experience with companywide communication. All major databases will have Web-like interfaces, and transactional servers will make contemporary database access and queries obsolete.

Drive Access and Corrupt Boot Records
I get many questions about problems with drive access and corrupt boot records. Let's examine the factors that can cause these problems.

Inaccessible Hard Disk
Several factors related to both hardware and software (e.g., a boot-sector virus) can lead to an inaccessible hard disk error. This type of error is always serious and needs immediate attention.

If you receive an inaccessible hard disk error on a SCSI hard disk, make sure all cables and terminators are correct and properly seated (for information on SCSI, see "Tricks & Traps," December 1996 and January 1997). Not surprisingly, heat and vibration can cause SCSI cables to move enough to either cause grounding (as when the cable grounds out to the case) or lose pin contact. As a result, this movement can cause the system to lose connections between devices and the controller card. You must attach today's SCSI-2 and SCSI-3 devices with cables that meet specifications. I vividly remember having a disk that occasionally stopped functioning. I always fixed it by reseating the connector, but the problem recurred. When I replaced the cable, the problem stopped.

With fast hard disks, heat can lead to an inaccessible hard disk error. If you have a hard disk that spins at 7200RPM, the disk can overheat and fail. This problem is common, especially with the typical small case size used today. Always make certain you have adequate ventilation and allow sufficient space above and below a hot disk.

Check the boot sector for viruses. A boot-sector virus can corrupt the boot sector and make the disk unreadable. If you have an NT boot floppy, you can boot to the floppy to access the NT installation. If you have to access NT in this way, your boot sector is corrupt or infected. You can scan for viruses and correct the problem from NT, or you can boot to a DOS floppy and run an antivirus application to fix the boot sector. You can also boot to a DOS floppy and run fdisk/mbr. This command will repair the boot sector, but you may still need to repair the boot sector with the NT Setup Repair Option.

If you get an inaccessible hard disk error on a large EIDE, enable Logical Block Addressing (LBA) and set the disk to the proper master and slave configuration. NT might not be able to read some proprietary software fixes for large hard disks, so the best solution is to enable LBA, re-run fdisk, and format the disk.

Boot Process Errors
Boot process errors can result from a boot-sector virus (the inaccessible hard disk error discussed above) or corrupt files. In the latter case, failing hardware or some other event, such as inadvertently deleting files, can cause file corruption. The most common boot process issues (in boot order) are

  1. If ntldr is missing, the following message appears:

    BOOT: Couldn't find NTLDR
    Please insert another disk.

  2. If ntdetect.com is missing, the following message appears on the boot loader screen:

    NTDETECT V4.0 Checking hardware
    NTDETECT V4.0 Checking hardware

  3. If ntoskrnl.exe is corrupt or missing, the following message appears immediately after the Last Known Good menu:

    Windows NT could not start because the following file is missing or corrupt:
    \winnt\system32\ntoskrnl.exe
    Please reinstall a copy of the file

  4. If bootsect.dos is missing or corrupt, you get the following message after the Last Known Good menu:

    I/O error accessing boot sector file
    multi(0)disk(0)rdisk(0)partition(1):\bootsect.dos

In all these cases, you can use NT setup's repair facility to examine and repair the disk's boot sector. Always have an up-to-date Emergency Repair Disk (for information on creating and using an Emergency Repair Disk, see Michael D. Reilly, "The Emergency Repair Disk," January 1997).

The system responds differently to boot.ini problems. If the boot.ini file is missing, ntldr immediately tries to boot NT. If NT is in the default \winnt directory, the system will probably boot. If the system uses another directory, the following message appears immediately following the Last Known Good menu:

Windows NT could not start because the following file was missing or corrupt:

\winnt\system32\ntoskrnl.exe

Please reinstall a copy of the above file

If the boot.ini file contains an improper path for NT, the following message appears:

OS Loader V4.0

NT could not start because of a computer disk hardware configuration problem. Could not read from the selected boot disk. Check boot path and disk hardware.

Please check the Windows NTâ„¢ documentation about hardware disk configuration and your hardware reference manuals for additional information.

To fix boot.ini file problems, you must edit and correct boot.ini. Remove its hidden and system attributes and change the improper syntax. If you can't access boot.ini, you must install a new version of NT (to a new directory) and access an NTFS partition containing boot.ini. You still need to manually edit the file.

Q: How do I know which file system to use with NT? What are the advantages and disadvantages of FAT, NTFS, and FAT32?

Deciding on a file system for NT is easier than you think. NT doesn't support FAT32, so you have to choose between FAT and NTFS. Table 1 summarizes the differences between the two file systems (for an in-depth comparison, see Sean K. Daily, "NTFS vs. FAT," October 1996).

TABLE 1: Advantages and Disadvantages of FAT and NTFS
Advantages Disadvantages
FAT Read by most operating systems Universal format No file security No disk restoration
NTFS File security Maintains logs Extended file attributes Read fully only by NT Cannot boot DOS to restore drives in case of problems Excellent cluster size management

I recommend NTFS. Emerging third-party NT utilities will make obsolete the argument that NTFS lacks the utilities to perform an NT repair. If you maintain an updated Emergency Repair Disk and have a boot floppy and a data backup, you can repair an NTFS volume unless you encounter a complete disk failure; in that case, you can't repair a FAT volume either. FAT can be faster than NTFS (e.g., with random reads). NTFS is difficult to use with small and removable drives. With small drives, you don't have enough room to justify the extra disk space NTFS requires. If you have many removable drives, you need to lock the drive (EZ-SCSI from Adaptec does this very well) before you can format it as an NTFS volume.

Some users advocate the use of a small, bootable FAT partition for ease of access to the drive in case you encounter problems with NTFS. This solution is acceptable, but you lose a considerable amount of security. In fact, having a FAT partition is a violation of C2 security, but I typically set up notebooks in this fashion.

Q: I recently bought an AST Ascentia P50 notebook computer as a result of a recommendation you made on the Windows NT Magazine online forums. It works with Windows 95, but I can't install NT. I can install the CD-ROM drive or the floppy drive but not both at the same time. To my disgust, the machine won't boot from the CD-ROM drive. I also have a Megahertz PC Card NIC. Will it work in NT?

A: First for the good news--NT 4.0 Workstation is selling well on notebooks. Most large notebook manufacturers will soon offer full or nearly complete power management and hot swapping modules (available now with Digital's UltraNote II).

To the case in point: I had the choice of installing Win95 or Windows for Workgroups (WFW) on my Ascentia notebook. I installed WFW because I wanted only a small FAT partition and I wanted a legal DOS/Windows software license for the notebook. I then used the AST utilities to copy the DOS 6.22 installation floppies. With these floppies in hand, I ran fdisk and removed the single partition. I then created a small (400MB) FAT partition and installed DOS. AST supplies a driver that lets you access the CD-ROM drive (teac_cdi.sys). I copied the \i386 directory to my hard disk and ran winnt/b to successfully install NT (for more information on running NT on AST's Ascentia P50, see Joel Sloss and Dean Porter, " Run NT on a Laptops? Yes, You Can!" on page 42).

Now to the NIC. I have a Megahertz XJ10BT NIC that uses the CC10BT/2 driver; however, several problems make using the NIC interesting. First, don't use the driver that comes with the NIC (dated 2/27/95). The driver will install successfully, but it is fatal. I couldn't fix it and had to reinstall NT. The only way to install the NIC so that it works in NT is to follow these steps:

  1. Don't let NT detect the NIC on installation. If NT attempts to detect the card, the process will return a null set for the inquiry and NT will blue screen.
  2. You can't manually detect the NIC because the path in NT setup is incorrect. Install NT without a NIC and manually add the card after NT is running. The proper path is <NT CD>:\drvlib\
    netcard\x86\mhzcpx where NT CD is the drive letter for the CD-ROM drive.

Once installed, the NIC works well in NT.

Q: I'm considering upgrading to Integrated Services Digital Network (ISDN). Several Internet Service Providers (ISPs) have told me that they don't support NT. Is this a problem? Is setting up ISDN in NT difficult?

The fact that many ISPs don't support NT is not a problem. Their connectivity software simply doesn't run under NT, so you'll have to configure your modem or ISDN device manually.

Setting up an ISDN device to work with NT is easy. I set up a 3ComImpact IQ ISDN device without trouble. The phone company will set up the necessary ISDN lines and provide you with the phone numbers and Service Profile Identifiers (SPIDs). After the phone line is in place, you connect an RJ-45 cable from the phone jack to the ISDN device.

You run 3Com's SPID Wizard (or comparable) and assign the phone lines and SPIDs. Be sure to enable multilink Point-to-Point Protocol (PPP) if you use dual B channels. Go to Dial-Up Networking and create a new phone book entry with the name, phone number, and ISDN device you plan to use. Screen 1 shows my settings. For TCP/IP, select Server assigned IP address and Server assigned name server addresses, as you see in Screen 2. NT will ask you to provide a username and password before you can dial your connection phone number. Make sure that the domain field is blank.

Q: I'm having problems with NT Explorer. Because of the way I use my system, I always have shares mapped from other systems. If I boot into NT and one of the mapped drives is not present, Explorer hangs for a long time. How can I avoid this behavior?

NT 4.0's Explorer is not my favorite app. I find it slow and awkward. In the case of mapped shares, Explorer can be very frustrating. If a share is mapped but not present, Explorer hangs. The easiest solution is to use File Manager to check shares before you use Explorer. If you use a server, you can also use Server Manager. You can't fix this problem, but you can manage it.

Q: This magazine might not be the proper place to ask, but how do I install non-Microsoft joysticks in NT?

My friend Larry Kahn figured this one out (I'm not much of a game player). Find the oemsetup.inf file in the \winnt\drvlib\multimed\joystick\x86 directory. You can't install this file directly. However, you can go to Control Panel, Multimedia Devices, Add, and select Add Other, Unlisted. Point the installer to the x86 directory to correctly install the file.

Q: I'm trying to decide on the best possible way to set up a network. I just read that you consider asynchronous transfer mode (ATM) to be one of the leading contenders for 1996 product of the year. Can you explain some of the advantages of ATM over Ethernet?

ATM offers some major improvements over Ethernet. First, ATM is a full-duplex network. So, for example, an ATM 155Mbits-per-second (Mbps) card can send and receive 155Mbps simultaneously. Only a handful of Ethernet cards offer full duplex. Although performance is not critical for workstations where only a few processes are active at the same time, performance is critical on a server. Full-duplex networking helps optimize network performance by increasing bandwidth.

Second, ATM is fully switched. Therefore, it optimizes performance on all ports to the maximum throughput rather than passively dividing the bandwidth over all active ports.

Finally (and most important), ATM is collision free. Ethernet has numerous packet collisions, and when utilization reaches 60 percent of the available bandwidth, Ethernet performance degrades dramatically. Under similar conditions, ATM functions without diffi-
culty (for more information on ATM, see Dan Blacharski, "ATM Boosts Network Speed," February 1997).

Of considerable interest is the ability to use Ethernet and ATM on the same network. You can consider an Ethernet LAN a connectionless technology (you don't have to establish a new logical or physical connection to send a message). The originator of the message simply attaches an address to the message, and the message travels through the network. The address can be a simple 48-bit media access control (MAC) address on the NIC when the station first boots onto the LAN. Every network station checks the destination MAC address of every packet moving over the network, but only the station to which the packet is addressed can remove the packet from the LAN. You simply change the bits of the MAC address to control who receives the message.

In contrast, ATM is a connection-oriented technology (a direct circuit is set up between the sending and receiving stations). The ATM cells transmit only through that circuit. This approach eliminates the collision of cells and enables private communication on the ATM LAN.

The ATM circuit is a virtual circuit (VC). During the VC setup, ATM uses an identifying string (160bits) that's larger than the Ethernet equivalent. However, after ATM establishes the VC, ATM places a 40-bit header in each ATM cell.

The solution that lets ATM and Ethernet communicate on the same LAN is quite elegant. First, you have to convert the Ethernet packets to ATM cells. Then you have to convert the addresses in the Ethernet packets to destinations that already exist or that you need to create. To make these conversions, you use LAN Emulations (LANEs). The client portion of the LANE provides a map that relates ATM VCs to MAC addresses. The server portion of the LANE keeps track of all the LANE clients. The client, server, and related bus form an emulated LAN.

The primary advantage of a combined Ethernet client and ATM backbone stems from the combined nature of ATM and Ethernet. With an ATM card on a server, you can load multiple instances of a driver and assign a different IP address to each instance (i.e., the subnet addresses are different). This capability lets you set up an emulated LAN for each IP address. Each Ethernet client on each emulated LAN can talk to clients on other emulated LANs or see the server, but they are independent of other emulated LAN connections. If you don't enable IP routing, you can easily segment your network and provide optimized, high-speed connections across the network (see http://www.whitetree.
com for details on ATM networking).