|
|
|
|
Windows 98 AnnoyancesBy David A. Karp1st Edition October 1998 1-56592-417-7, Order Number: 4177 464 pages, $24.95 Includes CD-ROM |
Chapter 6.
TroubleshootingIn this chapter:
General Troubleshooting Techniques
Error Messages
Preventative Maintenance and Data Recovery
Hardware Configuration
DOS to the RescueOften it's the entire Windows experience, rather than just Windows itself, that can really drive you nuts. In fact, Windows 98 goes farther than any previous version to help solve and even prevent problems, although it tends to cause many of the problems it alleviates.
Due to the sheer number of components that make up a modern computer system, it can be difficult to isolate the cause of a given problem. Hardware conflicts, buggy drivers, and poorly written software all contribute to any number of different problems. You may be experiencing frequent lock-ups, error messages, malfunctioning hardware, or data loss. These problems can be distilled into two basic forms: hardware or software malfunctions, and data loss due to these malfunctions.
While it's possible and even fun to blame Windows, it's not always productive. To deal with malfunctioning hardware or software, you need to first isolate the problem, and then take steps to solve whatever trouble you've uncovered. Unfortunately, isolating the problem is often the hardest part; a computer is a system of hundreds of different components all working together, and the symptoms aren't always obviously related to the problem.
Most hardware and software problems are caused by incompatibilities or conflicts, where two or more components don't work together--even though each may work perfectly well on its own. Faulty or out-of-date drivers frequently cause hardware problems, while incompatible .dll s can cause a myriad of software troubles. For example, one of the most common difficulties is trying to get a modem to work with a newly installed motherboard. The obvious tactics are to blame the modem or computer manufacturer, but there may be a conflict with the way the mouse port is configured in the system BIOS. If you don't know how to use the diagnostic components in Windows, finding problems like these can be ridiculously difficult.
While most Windows users will agree that these types of problems are annoying, nothing compares to trying to convince a technical support representative that the problem you're experiencing is not someone else's fault. Since most problems (even those that are hard to isolate) are easily fixed, it's always a good idea to do a little investigating before packaging the problem item up and bringing it back to the mega-super store where it was purchased.
On the other end of the spectrum is data loss, which I think is nicely summed up by the following haiku:
Your file is missing?
It might've been important,
But now it is gone.Coping with data loss requires two strategies: preventative maintenance and data recovery. Assuming you've isolated and solved the source of the data loss (usually one of the problems discussed above), there are several methods you can employ to make sure your work is safe, and to recover it when it isn't.
Regardless of the type of problem, there's no substitute for a full system backup. Even if your computer equipment is insured with Lloyds of London, once your data is gone, it's gone.
General Troubleshooting Techniques
Just a few words of advice before we begin: if it ain't broke, don't fix it. Many problems are actually caused by people looking for problems to solve. For example, installing a new device driver just for the sake of having the newest drivers on your system may introduce new incompatibilities. While the material presented here is intended to aid in general troubleshooting and data recovery, some of the solutions can cause other problems, so it is strongly recommended that you back up your entire hard disk before continuing.
Drivers and Installing Hardware
A driver is a piece of software that allows your computer to communicate with the various devices attached to it. For example, your printer has a driver that contains all the capabilities of your printer, such as paper sizes and print resolution. The beauty of this design is that an application like your word processor can simply send your document to Windows with the instruction "print," and Windows will take care of the rest. This way, each application doesn't need to know anything specific about your printer, or, for that matter, the specifics of any of your hardware.
Problems arise when a driver is either defective or outdated. Drivers designed either for a previous version of Windows or for a previous version of the device can create problems. Additionally, manufacturers must continually update their drivers to fix incompatibilities and bugs that emerge after the product is released. It's usually a good idea to make sure you have the latest drivers installed in your system when troubleshooting a problem. Newer drivers usually offer improved performance, added features and settings, better stability and reliability, and better compatibility with other drivers in your system.
Also, be aware that some drivers may not be the correct ones for your system. For example, when Windows is installed, it may incorrectly identify your video card or monitor and hence install the wrong driver, or even a generic driver. A common symptom for this is Windows not allowing you to display as many colors or use as high a resolution as the card supports. Make sure that Device Manager lists the actual devices you have installed on your system (double-click on the System icon in Control Panel, and click the Device Manager tab).
Device drivers worth investigating include those for your video card, monitor, sound card, modem, printer, network adapter, scanner, SCSI controller, tape drive, and any other drives or cards you may have. If you're not sure of the exact manufacturer or model number of a device, take off the cover of your computer and look, or refer to the invoice or documentation that came with your system. Most hard disks, floppy drives, CD drives, keyboards, mice, power supplies, memory, and CPU chips don't need special drivers (except in special circumstances).
Windows 98 comes with a significant number of drivers for hardware available at the time of its release, but as time passes, more third-party devices will require their own drivers.
It's possible to find out if Windows comes with a driver for a specific piece of hardware before you even install it. Double-click on the Add New Hardware icon in Control Panel. Click Next, and then Next again. Windows will take a few seconds (you won't see a progress bar here) to poll all of your Plug-and-Play devices. You'll then be given the option of having Windows search for your new hardware; since we're only looking for a driver, choose No and then click Next. Choose the type of hardware from the list shown that most closely matches what you're looking for, then click Next to display a list of manufacturers and their products (see Figure 6-1).
Figure 6-1. List the devices supported by Windows by using the Add New Hardware wizard
![]()
If you actually want to go ahead and install the device, or if the device is already physically installed in your computer, it's better to choose Yes to have Windows automatically search for the device. It may seem to take longer, but it's more reliable, since Windows will automatically detect the particular configuration settings as well (otherwise you have to configure them). The downside is that you don't get the list shown above.
Note that not all devices need to be installed in this way. Many devices, such as hard disks and CD drives, are automatically detected and configured when Windows starts; you should see their icons in the My Computer window without any fuss. Some other devices are detected by Windows during startup, but require that you answer a series of questions and sometimes insert a driver disk before you can use them. It all depends on the type of device and how well (if at all) the manufacturer has followed the Plug-and-Play specification. If you're in doubt, consult the documentation to see what the manufacturer recommends.
On the other hand, if a device is already installed and the driver has already been chosen, there are two ways to figure out if the correct driver is being used. First of all, the name used to identify the device in Device Manager is a good clue. For example, if, under the Display Adapter category, your video card is listed as a Diamond Viper V330, then that's the driver that's being used, even if that's not really the video card you're using.
However, there's more to the driver than just the name; to find the date and revision number of the driver, double-click on the device in Device Manager and choose the Driver tab. While Windows does come with plenty of drivers, very few of them are actually written by Microsoft even if Microsoft is listed in the Provider field; manufacturers simply submit their drivers for inclusion in the package. You can usually assume three things about the drivers included on the Windows CD versus those that come with your devices:
- The drivers are fairly stable.
- The dates are usually consistent with the release history of the manufacturer's drivers.
- Any special features or extras present in the manufacturer's version of the drivers have been left out. For example, many after-market display drivers include support for more colors, higher resolutions, hotkeys, panning windows, etc. The drivers on the Windows CD will usually not have any of these.
An easy but not foolproof way to tell if you're using the driver that came with Windows is to look at the driver date--it should be 5-11-98, if you're using the initial release of Windows 98. If the date is different, the driver probably came from another source, such as a driver disk, the Web, or a previous installation of Windows. Drivers with newer dates are usually--but not always--more recent. (Note that the Driver File Details button rarely has any useful information.) If you're trying to solve a problem, or take better advantage of your hardware, your best bet is to visit the home page of the device's manufacturer and download their latest and greatest driver.
To change the driver for the selected device, either by installing a newer version or by replacing it with a driver for a different device, click Update Driver.[1] You'll then see yet another wizard that allows you to show Windows where to get the new driver. Click Next, select Display a list of all the drivers, and click Next again. You can also use the Search for a better driver option, although it's usually more of a pain to do it this way.
At this stage, you'll see a list of "Compatible Hardware," which usually consists of the one currently installed driver. Any additional drivers listed here are those that Windows identifies as supporting the same device as the currently installed driver; if the currently installed driver is wrong, the other items in this list may also be wrong. Click Show all hardware to view all the devices in the current category; you'll see a window similar to Figure 6-1. Note that you can't change the category of the device; if it's a video card, you can't choose a driver for a SCSI adapter.
Even if your device is listed, your manufacturer may have a newer driver.[2] The nice thing about the Show all hardware window is that it shows all the installed drivers along with their dates. If you install a third-party driver, and then return to this window, you'll see the new driver alongside the driver that comes with Windows (if applicable).
You can also click Have Disk at this point to browse your hard disk for another driver. A common scenario involves downloading a zipped driver from the Web, unzipping it to a directory (see Appendix E, Interface Terminology and the Basics), and then using this dialog to instruct Windows to load the driver. Windows will accept any folder containing a valid driver, which is detected by the presence of an .inf file. Actually, all the drivers already installed on your system have a corresponding .inf file in the \Windows\INF folder. It's possible, however, that an installed driver won't show up in the Driver List window; in this case, you can sometimes use Have Disk and point Windows to the \Windows\INF folder to select the right driver.
The .inf file is the heart of each Windows driver. Sometimes it contains all the necessary device information; for example, most modems require only this single file. At other times, the .inf file contains information and links to .dll and .vxd files, which do the actual work of the driver. Unfortunately, each device is different--don't expect a set of tricks that worked for one driver to necessarily work on another driver.
The evolution of drivers
It's important that any drivers you use with Windows 98 be designed especially for the Windows 95/98 platform. Using drivers made for older versions of Windows or DOS may cause problems ranging from poor performance to crashes to the device not working at all. Newer 32-bit drivers offer much better performance and stability, as well as extra features like support for Plug-and-Play, and compatibility with other devices in your system.
The Win32 driver model, a specification introduced in Windows 98, is a new type of driver that is compatible with both Windows 98 and Windows NT 5.0. This type of driver is always preferable but is often (at the time of this writing) not available. Some manufacturers might even have both standard 32-bit drivers (for Win95 and Win98) and Win32 drivers (for Win98 and NT 5 only). Ideally, all drivers should offer the same functionality, stability, and performance, but more than likely, there will be a tradeoff somewhere along the way. You might even consider installing and trying out all driver varieties for your device so that you can be sure to have the best one.
If a new, 32-bit driver isn't yet available, you should be able to use an older Windows driver (designed for Windows 3.x), although this isn't recommended if you can avoid it. If you are loading DOS drivers in your Config.sys or Autoexec.bat files, you may be preventing Windows from installing proper 32-bit drivers--see "Do I Still Need Config.sys and Autoexec.bat?" later in this chapter for more information.
Some users may be disappointed to discover that a manufacturer of a discontinued product has stopped supporting the product, or that the company has gone out of business. If this happens, you may be out of luck and forced to replace the device if it isn't supported in your version of Windows; see Chapter 5, Maximizing Performance, for more information on upgrading your system. There is a way out, however. Many products, such as video cards, modems, and SCSI controllers, use similar components that are widely supported by the industry. For example, many varieties of S3's video card chips are used commonly in video cards today. By looking at your video card, you should be able to determine which variety of chipset it uses: look for the brand and model number. Even if the manufacturer of your video card has gone out of business, there may be other video cards that use the same chipset, and therefore may use the same driver.
Misbehaving drivers
Never install more than one new driver at a time. By upgrading one driver at a time, you can easily isolate any potential new problems and recognize when an existing problem has been solved. Wait for Windows to restart and try starting a program or two. If you install several new drivers at once, you'll have a hell of a time trying to find where you went wrong.
When you install a driver, Windows first copies the various driver files to as many as five different folders. Then the Registry is configured with the driver filenames, the specific resources used by the device (IRQs, I/O addresses, etc., all of which are discussed later in this chapter), and any special settings. A common problem is that the special settings can be incorrect, and no amount of fiddling with them can straighten out a misbehaving device.
This often happens with network cards and SCSI adapters, where the device doesn't function at all, or Windows doesn't recognize the device's resources correctly, or an attempt to use the device hangs the system. The solution is simply to reinstall the driver. The best way to go about this is to locate and select the device in Device Manager and click Remove, then close Device Manager and restart your computer. During Windows startup, you should then see a message to the effect that Windows is installing a new device, at which time it may ask for the driver disk. You can either point it to the appropriate location, such as your floppy drive or a folder on your desktop, or point it to your Windows\System folder, which forces it to use the old drivers. Windows then reinitializes the device and resets all its special settings, which, in many cases, will solve the problem.
More drastic measures include removing all the actual driver files from the hard disk before allowing Windows to install new ones. Since all drivers are different, there are no standard files to remove. Conscientious developers provide an uninstall utility for their drivers, or at least a list of the supported files so you can find them easily. If in doubt, contact tech support and ask them how to completely remove their driver.
My last piece of advice is to copy the latest drivers for all of your devices onto floppy disks for easy access the next time you need them. You'll be glad you did this when you realize that you can't download the right driver for your modem if your modem has stopped working.
Resolving Conflicts and Other General Problems
The most common type of hardware problem is a conflict. A conflict occurs when two devices try to use the same resource, such as an interrupt request line (IRQ) or memory address. The telltale signs of a conflict include one or more devices not working, one or more devices not showing up in Device Manager, or your system crashing every time one or more devices is used. The first step in diagnosing a conflict is to check the drivers (see "Drivers and Installing Hardware" earlier in this chapter).
Each installed device can use one, several, or even no resources. Usually, an expansion card, such as a sound card or modem, uses a single IRQ, a single I/O address range, and sometimes a direct memory access (DMA) address. Other devices can consume more than one of these resources, as well as other resources: memory addresses, SCSI IDs, IDE channels, and serial and parallel ports.
If two or more devices try to use the same resource, problems ranging from slow performance to system crashes can occur. Most older devices (called legacy devices) allow you to configure which resources they use by setting appropriate jumpers or switches on the devices themselves. Newer devices allow their settings to be changed with software. The newest Plug-and-Play products work with your Plug-and-Play BIOS to automatically configure themselves to work with other Plug-and-Play and legacy devices, theoretically avoiding all conflicts.
Note that some devices that connect to your computer's external ports, such as printers, don't technically use any resources of their own; however, the port to which they're connected does use resources. You can usually change the resources used by any given device (ports included). The trick is to configure all your devices to use different resources so that no conflicts occur. All devices are different; refer to the documentation included with the device, or contact the manufacturer for specific configuration instructions.
To determine which resources are still available in your system, as well as which devices are using the remaining resources, open the Device Manager by double-clicking on the System icon in Control Panel and choosing the Device Manager tab. Select Computer from the top of the list, and click Properties to display the Computer Properties dialog, as shown in Figure 6-2.
Figure 6-2. You can determine which resources are being used by looking at the Computer Properties in Windows's Device Manager
![]()
By choosing any of the four types of resources (IRQ, I/O, DMA, or Memory), you'll see how each is used by the various devices in your system. Any gaps in the numbers represent available system resources, which you should be able to assign to new devices. Most resource conflicts are shown here as well: if you see two cards assigned to IRQ 10, it's a safe bet that that's part of the problem.
Be aware that some devices can share resources. For example, your communication ports share IRQs (COM1 and COM3 both use IRQ 4, and COM2 and COM4 both use IRQ 3). Sometimes this is benign, such as having a mouse on COM1 and another pointing device (such as a graphics tablet) on COM3. However, most modems will complain (system hangs, slow performance, and other malfunctions) if they share resources with any other devices. If you have three or more devices connected to COM ports, you may have to either juggle them around or install configurable COM ports that can be configured more flexibly.
If you find some other type of conflict, start by either removing or reconfiguring one of the devices involved. You may be required to reconfigure several devices, literally shuffling resources around until all the conflicts are resolved. Again, the method used to change the resources used by a particular device depends upon the device itself. You should be able to see all the resources used by a given device by selecting it in Device Manager, clicking Properties, and choosing the Resources tab.
Note that the information presented in Device Manager may not necessarily reflect the current state of your system. If your computer was made before about the middle of 1995 and doesn't support Plug-and-Play entirely (or at all), you may have devices installed that don't show up, as well as devices displayed that aren't actually installed. The Refresh button is used to reread the devices in your system, but it won't detect anything more than is normally seen when Windows starts. Refresh is used primarily to detect devices attached to or disconnected from your system after Windows has started without having to restart. To force Windows to re-detect the hardware attached to your computer, use the Add New Hardware icon in Control Panel, and confirm that you want it to search for new hardware when asked.
Installing new hardware
If you're installing more than one device, do so one at a time; it's much easier to isolate problems when you know which device has caused them. You should expect installation of Plug-and-Play devices to be quick, automatic, and painless, at least in theory. However, many devices, while able to configure themselves automatically, may not be able to adapt entirely to your system. Be prepared to reconfigure or even remove some of your existing devices to make room for new ones.
If you're trying to get an existing device to work, try removing one of the conflicting pieces of hardware to see if the conflict is resolved. Just because two devices are conflicting doesn't mean that they are at fault. It's possible for a third, errant device to cause two other devices to occupy improper resources, and therefore conflict with each other or simply not function.
If removing a device solves a problem, you've probably found the conflict. If not, try removing all devices from the system, and then reconnect them one by one until the problem reappears. Although it may sound like a pain in the neck to remove all the devices from your system, it really is the easiest and most sure-fire way to find the cause of a conflict. Since there are so many different combinations of resource settings, it can be a laborious task to resolve conflicts. Some devices come with special software designed for this task; the software can either advise you of proper settings, or in some cases, even make the changes for you. Make sure to review the documentation for any mention of such a utility.
Firmware
User-upgradable firmware is a feature found in many new devices. Firmware is software stored in the device itself, and is used to control most hardware functions. While it's not possible to, say, increase a hard disk's capacity by upgrading its firmware, it is possible to improve performance slightly, as well as to solve any compatibility problems that may have been discovered after the product shipped.
The beauty of firmware is that if you purchase a peripheral, and the manufacturer subsequently improves the product, you can simply update the firmware to upgrade the product. While user-upgradable firmware can increase the initial cost of a product slightly, such an increase is dramatically outweighed by the money the manufacturer can save by not having users send in equipment to be updated. Naturally, user-upgradable firmware is also a boon to the end user, who can make simple updates in a matter of minutes, without having to send in the product or even open up the computer.
To find out the firmware version for a particular device, select the device in Device Manager, click Properties, choose the Settings tab if it's present, and look at the Firmware Revision field. If you're experiencing problems with a certain device, check with the manufacturer of the product for a newer firmware revision. In many cases, you'll be able to download a simple software "patch" that will update the firmware to the newest version, possibly fixing problems and even adding new features.
Devices that commonly have user-upgradable firmware include modems, CD drives, CD recorders, removable drives, removable and tape drives, motherboards (in the form of an upgradable BIOS), SCSI controllers, and network adapters, hubs, and routers.
Some older devices allow you to change the firmware by upgrading a chip. It's not as convenient as software-upgradable firmware, but it's better than nothing.
Plug-and-Play
Most new internal peripherals, such as cards and drives, and some external devices, such as printers and scanners, will be automatically detected when Windows boots up. Ideally, Windows should notify you that the new device has been identified, and give you the option of using the driver that comes with Windows, if available, or providing the driver on your own, either with a diskette or a folder on your hard disk. Windows should then load the driver, configure the device, and restart with no ill effects.
Problems arise when the new device either doesn't work or causes something else to stop working. Even the newest Plug-and-Play devices can sometimes cause conflicts, although with the passage of time, the PnP compliance of most new devices has generally improved. To aide in troubleshooting conflicts where PnP devices are involved, it's important to first realize exactly what Plug-and-Play technology is. PnP-compliant devices must have the following characteristics:
- The device must have a "signature" that is returned when Windows asks for it. Windows then looks up this signature in its driver database, and either finds a driver that matches it or asks you to insert a disk with a compatible driver. If a compatible driver is not found, no driver will be loaded for the device.
- All configurable resources of the device must be software-adjustable; that is, it is not necessary to physically set jumpers or switches on the hardware to reconfigure it. This doesn't mean, however, that the device can't come with jumpers; some cards let you disable their Plug-and-Play features and set resources manually, a very handy feature.
- The driver, if supplied, must be capable of instructing Windows which resources the device can occupy, and must be able to receive instructions from Windows and reconfigure the card accordingly. That way, Windows can read all the possible configurations from all the drivers, and then reconfigure each one so that there are no conflicts.
You can see, then, how dependent PnP devices are on their drivers, and why a buggy driver can cause problems with the entire system. One common bug in some drivers is that they are unable to reliably configure the corresponding device. For example, say a sound card can occupy IRQs 5, 7, 9, 10, and 11, but the driver is incorrectly programmed to also accept IRQ 13. When Windows attempts to shuffle all the devices around, it may then ask the sound card to occupy IRQ 13; since this is impossible, it will remain at its previous setting (or at no setting at all), and most likely cause a conflict with another device, say, a modem or parallel port. In this scenario, a tiny bug in a single driver has caused two separate devices to stop functioning.
Now, it's also possible that Windows will be unable to find a mutually agreeable configuration for all installed devices--even if one does exist--which means that Windows will simply boot with one or more conflicts. In most cases, Windows won't even tell you that PnP has failed. This is where you have to take matters into your own hands: learn to recognize the symptoms, which include crashing, hanging, and slow performance, and know how to look for conflicts. See the beginning of this section for details.
One of the loopholes that you can take advantage of is the way that Plug-and-Play systems assign resources, particularly IRQs, to PCI devices. Your BIOS will assign a different IRQ to each slot, rather than having each device try to grab an IRQ for its own; this ensures that PCI cards don't conflict. The funny thing about PnP BIOSes and Windows 98 is that sometimes some IRQs are neglected. If you have a full system and find yourself running out of IRQs, this can be a real problem. The good news is that you can enter your system's BIOS setup (see the next section) and manually assign an IRQ to each slot, often even specifying previously ignored IRQs, such as IRQ 12, 14, and 15.[3] This will then leave spaces open (usually lower IRQs), which other devices in your system can then occupy.
Finally, a common problem with Plug-and-Play is its propensity to detect devices that have been already configured. For example, after you've hooked up a printer, installed the drivers, and even used it successfully, Windows may inform you the next time you boot that it has detected a newly attached device, namely, your printer. This is almost always caused by an incorrect initial installation; for example, you may have connected your printer after Windows had started. The best course of action is to remove the drivers for the device (usually through Device Manager), reboot, and allow Windows to detect and setup the printer automatically. Naturally, you should check the printer's documentation for any abnormalities of the installation process.
Windows won't start
An all too common problem is that Windows simply won't start. This is an extremely broad problem, usually occurring without an error message or any obvious way to resolve it. Most of the many causes of this problem deal with hardware drivers, conflicts, or file corruption, all of which are discussed elsewhere in this chapter.
However, one solution that is fairly easy to implement is the following, which deals specifically with file corruption. Often by simply deleting your swap file and temporary files, you can allow Windows to load without incident.
TIP: Don't do this if you suspect that your hard disk has crashed, since it may make matters worse. See "Disk crash" in "Preventative Maintenance and Data Recovery" later in this chapter for more information. If you're worried about deleting potentially valuable temporary files that may contain data from previously open documents, try skipping step 4.
- If you find that Windows either doesn't complete its startup process, or does something weird like rebooting before it's finished loading, restart your computer by pressing the Reset button or the Ctrl-Alt-Del keys. Then press the F8 key immediately after the beep and the small message in black-and-white, "Starting Windows 98," appears.
- You'll then see the Windows startup menu, as described in "Create a Startup Menu" later in this chapter. Press Shift-F5 at this point to exit to the command prompt.
- The Windows 98 swap file (Win386.swp) is usually located in your \Windows directory, although your system may be different. Change to your Windows directory using the
CDcommand (usuallycd \windows), and then type the following:
attrib -r -s -h win386.swpdel win386.swp- Then change to your temp directory using the CD command (usually
cd\windows\temp) and type the following:
attrib -r -h *.tmpdel *.tmp- The last step is to run Scandisk by typing
scandiskat the prompt.
- When Scandisk is finished, restart your computer, and allow Windows to boot normally. If you see the Windows startup menu, make sure to choose the first option, 1. Normal, and then be patient as Windows loads.
Fixing Device-Specific Problems
For most components in the system, the number one rule for getting things to work is to make sure you have the latest and correct driver from the manufacturer, although this isn't always as easy as one would like. In many circumstances, obtaining the correct driver may be impossible, or simply not applicable. The next step in getting most hardware to work in Windows is to eliminate any hardware conflicts, which is discussed in "Resolving Conflicts and Other General Problems" earlier in this chapter.
But more often than not, problems are unique to a particular type of component. For example, modems often suffer the same types of problems, which don't necessarily affect any other types of hardware. The following guidelines should help you solve most component-specific problems, as opposed to general lockups or application error messages.
And don't forget that a nonfunctioning component can be a great excuse for an upgrade (see Chapter 5).
Video cards (also known as display adapters)
Even without the correct video driver installed, you should probably be able to use Windows at a resolution of 640 × 480 with 256 colors; this is a standard mode supported by nearly all VGA cards, and is Windows' default display mode. Most video card problems are caused by faulty or incorrect video drivers, however, so it's best to check with the card's manufacturer first.
Now, if you don't have a VGA-compatible card, you're much better off simply purchasing a new card (good ones can go for as low as $50), which will be much easier and less of a hassle than trying to get that old obsolete display adapter to work properly.
Most modern video cards are based upon a certain chipset, usually identifiable by the large, square chip in the center of the card. If the chip is covered with a sticker, remove the sticker to see what's printed on the chip surface. In fact, Windows may be able to detect the type of chip, even if it can't determine the make and model of the card. Common chipset manufacturers include S3, Cirrus Logic, ATI, Tseng, and Western Digital; each of these comes in several varieties as well. If you can determine the type of chipset your video card uses, you should be able to use either a generic video driver made for that chipset, or a driver for another card that uses the same chipset.
If you know you are using the correct video driver, but can't use all of the resolutions it supports, make sure Windows is identifying your monitor correctly.
If you're trying to use Windows 98's support of multiple monitors, there are a few things to be aware of. Your system BIOS chooses which video card is your primary adapter and which card is your secondary adapter. To switch their priority, either to resolve a problem or for personal preference, you'll have to swap their physical positions in your computer. Since your primary video card does not need to explicitly support multiple adapters, but the secondary card does, you may have to swap them to get multiple monitor support to work at all. One problem you may encounter is trying to negotiate one PCI card and one AGP card; since most BIOSes initiate PCI before AGP, your AGP card will never be the primary card. In this case, you'll either have to make do with what you've got or install two PCI cards.
Monitors
If Windows knows what type of monitor you're using, it can determine which resolutions and color depths it supports. In Windows 95/98, monitors have drivers, although they do little more than inform Windows of the monitor's capabilities. Newer Plug-and-Play monitors allow Windows to automatically identify your monitor, although a driver may still be required. To see if your monitor is specified correctly, double-click on the Display icon in Control Panel. Choose the Settings tab, click Advanced, and then choose the Monitor tab in the new window.
It's possible for your video card to generate video signals that your monitor isn't able to display, especially if your resolution, color depth, or refresh rate are set too high. While a video card and a monitor don't have to be matched precisely to work, it's worth investigating whether or not your monitor can support all of your video card's modes.
You may be able to use a higher resolution or color depth with your monitor by lowering your video card's refresh rate. You should also lower the refresh rate if you hear your monitor whistling, and raise the refresh rate if the display flickers.
For problems using multiple monitors, see "Video cards (also known as display adapters)" earlier in this chapter.
Hard disks and floppy diskette drives
These drives almost never need special drivers, unless they use a proprietary interface such as your parallel port. Windows will support virtually all integrated drive electronics (IDE) drives right out of the box, as well as many SCSI controllers and devices.
Most hard disks, with the notable exception of SCSI, require that you specify their parameters (number of heads, cylinders, sectors per track) in your computer's BIOS setup, usually accessible by pressing some key just after you first turn on your system. Newer systems with a Plug-and-Play BIOS can autodetect most drives.
If Windows does not recognize your floppy or hard disk, you'll need to obtain a driver specific to the controller to which it's connected. If you continue to have trouble accessing the drive, make sure the jumpers are set appropriately and the cables are connected correctly.
Hard disk controllers
Most hard drives available today are the IDE type. Since new IDE controllers can be purchased for around $10 these days, you should simply throw out your existing IDE controller if it needs a driver, or if it takes more than five minutes to get working.
Proprietary IDE controllers, such as caching controllers, usually perform worse and cause more problems than the standard controllers built into nearly all new motherboards. You're better off taking the extra memory from the controller and installing it on the motherboard, throwing away the proprietary controller, and using the one on the motherboard.
Windows comes with drivers for most types of hard disk controllers, including IDE, RLL, ESDI, and SCSI. If Windows doesn't support your controller, and you can't get a driver from the manufacturer, you're out of luck.
SCSI controllers
Most SCSI controllers are either supported by Windows out of the box, or have 32-bit drivers you can use that come with the SCSI card or are available from the manufacturer. For the most part, all SCSI controllers are fairly well supported, with recent drivers nearly always available. If you're having a SCSI problem, you should first check to see if newer drivers for your card are available.
If you're unable to find drivers for your SCSI card, you may still be able to use it in Windows if you can find a driver for another card that uses the same SCSI controller chip, sometimes called a miniport driver. Common miniport manufacturers include Adaptec, BusLogic, Future Domain, NCR, and Trantor. For example, you may have a sound card that has a built-in SCSI controller intended for your CD drive. If that SCSI controller just happens to be made by Adaptec, for example, you should be able to use a driver for the corresponding Adaptec product that runs off the same chip.
Next to drivers, the two most common problems with SCSI controllers and the devices that attach to them are bad cables and incorrect termination. When diagnosing any SCSI problems, it's best to have replacements for all the cables, so you can easily swap them to help isolate the problem. The use of improper or non-SCSI adapters and connectors is also a common culprit.
As for termination, a SCSI chain (the long string of devices connected by cables) won't work properly unless it's correctly terminated. Make sure that each end of the chain, but nothing in the middle, is terminated, either by using the built-in termination on your SCSI controller and SCSI devices, or by attaching stand-alone terminators (active terminators are best). You should never have more than two terminators unless you have more than one controller. The SCSI card itself should be terminated, unless you have internal and external devices, in which case the devices at the end of each side should be terminated.
For problems with specific devices connected to SCSI controllers, such as CD drives, hard disks, and removable drives, refer to the corresponding topic elsewhere in this section.
CD-ROM drives
Most CD drives don't need special drivers. In fact, if you plug in a CD drive and then start up Windows, it should automatically detect the drive and display an icon for it in My Computer. If your drive isn't detected, first check the controller. Most CD-ROMs connect to your IDE or SCSI controller; if your drive isn't recognized, most likely the controller isn't working or you don't have the right drivers for your controller installed. Some older CD drives connect to proprietary controllers or sound cards. For these, you also may need a driver made especially for your controller/drive combination; check the documentation for details.
Tape drives
Most tape devices don't require general-purpose drivers of their own, mostly because there is no standard for them. Any backup program compatible with your drive will come with its own drivers, which work with the installed drivers for the controller to which your drive is attached. So, if you have a SCSI tape drive, you'll need to first make sure the drivers for the SCSI controller are working. The backup program you use will come with generic drivers for SCSI tape drives, as does the backup software that comes with Windows 98.
Most tape devices come with their own backup software, which is usually guaranteed to work with the drive. If the manufacturer of your tape drive does not supply software, or if the supplied software is not a 32-bit application made especially for Windows 95/98, you'll need to use a different program, such as third-party commercial backup software or Microsoft Backup.
If you're trying to get Microsoft Backup or another backup program to recognize your drive, you should consider comparing the price of new backup software with the price of a whole new tape drive that includes its own 32-bit backup software.
If you must use old backup software with Windows 98, either because of monetary limitations or to maintain compatibility with other computers running older versions of Windows or DOS, you can still back up your long filenames with a separate utility. See "Restoring Windows After a Crash" later in this chapter for more information.
Note that you'll never see a drive letter for your tape drive in My Computer, unless you install a special utility designed for that purpose, such as Seagate Direct Tape Access. I've never found these utilities to work very well, however.
Removable, optical, and recordable CD drives
Removable cartridge drives, such as magneto-optical, Syquest, and IOmega drives, as well as CD recorders, traditionally connect to SCSI controllers. Therefore, as long your SCSI controller is functioning and the SCSI drivers are installed, you should be able to connect any SCSI device, make sure the chain is terminated, and it should work without a fuss, and without any special drivers.[4]
If you're having trouble getting a SCSI device like this to work, newer SCSI drivers or an update to the SCSI BIOS (contact the manufacturer) should solve the problem.
Some devices attach to IDE controllers via parallel ports. While these connections may seem more convenient or less expensive than SCSI, they're generally slower, less reliable, and more finicky. Additionally, parallel-port-connected devices usually require drivers provided by the manufacturer, which, of course, must be intended specifically for use with Windows 95/98.
Problems with external devices (SCSI and parallel-port connections) usually involve bad cabling. Make sure all cables are correct and seated firmly in their connectors, and that all thumbscrews are tightened.
Modems
If your modem is 9600bps or slower, it's not worth the time to get it to work. You can get a 56kbps modem for less money than it would cost in long-distance support calls to find drivers for the old one.
While choosing the appropriate driver is important, you can usually get by with one of the standard modem drivers included with Windows. In fact, Windows might simply call your modem a standard modem if it can't autodetect the make and model, even though a driver for your modem may be included with Windows. While a driver made especially for your modem will usually yield better performance and reliability, you can sometimes use a driver for another product by the same manufacturer, as long as it's the same speed.
Essentially, the only piece of configuration information Windows really needs to use your modem is knowledge of its maximum speed and its initialization string. This is a long string of seemingly nonsensical characters, beginning with AT, used to send commands to your modem to prepare it for dialing.
If Windows recognizes your modem, but you can't seem to get it to work, or simply can't find the appropriate driver, try entering your modem's initialization string into Windows. To obtain your modem's initialization string, either contact the manufacturer of your modem or refer to the documentation. If you currently have older software that works with your modem, a good trick is to snoop around the configuration section of the software to find the configuration string it's using.
Double-click on the Modems icon in Control Panel, select your modem from the list, click Properties, and choose the Connection tab. Click Advanced, and type your initialization string into the field labeled Extra settings.
If you can't find your modem in the list, double-click on Add New Hardware in Control Panel, and confirm that you want to it to search your system for newly attached devices. If Windows doesn't identify your modem, or it identifies it as an "unknown device," there are two possible causes of this problem. Either your modem is a proprietary model, such as some built-in modems found in portable computers, or the serial port to which it's connected is not functioning.
If yours is a proprietary modem (always something to avoid), you won't be able to use it without a proper driver from the manufacturer.
Serial port conflicts and misconfiguration are the most common causes of modem recognition problems. External modems are connected to serial ports, which usually are just cables that plug into the motherboard (see "Motherboards and CPUs," later in this chapter, for more information on serial ports).
Enter your computer's BIOS setup to verify that the serial port to which your modem is connected is enabled and configured correctly. Most likely you'll have two ports, one connected to your mouse, and the other to your modem (see Figure 6-3). Consult the documentation that came with your computer or motherboard for details.
Internal modems, on the other hand, are expansion cards that contain their own serial ports. Enter your computer's BIOS setup, and make sure any serial ports on your motherboard that you're not using are disabled, as they could otherwise conflict with your modem.[5] Commonly, only one serial port is used by the mouse (see Figure 6-3), but both are enabled on all new systems and motherboards by default.
Figure 6-3. Common connections for serial ports; disable port B if you have an internal modem, enable B if you have an external modem
![]()
If Windows recognizes your modem correctly or as the generic "standard modem," the next step in resolving the problem is to verify communication with the device. Start by running HyperTerminal (use Add/Remove Programs ➝ Windows Setup if you don't have it installed) to communicate with the modem. When prompted, type anything for the Name, and press OK. In the Connect To box that follows, choose Direct to ComX from the Connect using listbox, where ComX is the port to which your modem is connected, usually COM1 or COM2. Press OK; if your choice is correct, you'll be sent straight to HyperTerminal's main window.
Type
ATZ, a simple reset command, and press Enter. If you receive an "Ok" after the successful completion of this reset, it means communication between your computer and your modem is working, and the problem is probably an incorrect driver or configuration program in your software.[6] If you don't receive an "Ok," there's probably a hardware conflict, as described above.If you know the software is installed and configured correctly, there are external factors that can either prevent modems from working or slow modem performance. Start by removing all other electronic devices from the phone line, including answering machines, fax machines, autodialers, and standard telephone handsets. Any of these can actually interfere with the modem, preventing it from detecting the dial tone, or causing it to hang up prematurely. Other factors include bad phone cables and wall sockets. Try replacing your old phone cord with a brand new one just long enough to reach the wall jack.
Occasionally, a functioning modem can stop working temporarily. Since modems are constantly receiving commands from your computer, it's possible for the modem to become confused if it is sent a garbled or incomplete command. The easiest way to correct a confused modem is to turn it off and then on again. If the modem is an internal model, you'll need to completely power down your computer and then turn it on again; simply pressing the Reset button or restarting Windows may not be sufficient.
If you have an ISDN modem, these techniques won't necessarily apply. However, most external ISDN adapters connect through serial ports, which can suffer conflicts as described previously. Internal ISDN modems are treated by Windows as network adapters (see Chapter 7) rather than as modems.
As with any modem, make sure you have the latest drivers and, if applicable, the latest firmware.
Printers
Common printer problems involve bad cabling as well as bad drivers. Solving cabling problems is easy; just replace your parallel cable with a new one. IEEE 1284-compliant cables are the best, and are usually required for newer inkjet and laser printers. Some printers won't function if they're too far away from your computer, so try a shorter cable. See "Motherboards and CPUs" later in this chapter for more information on parallel ports.
As with most other peripherals, getting the right drivers is essential, though Windows can print plain text without fonts or graphics on nearly any printer without knowing what kind of printer you have. If you don't have a driver made especially for your model, you still may be able to use your printer with Windows by installing the Generic/Text Only driver included with Windows, although this will only enable very limited output. To use fonts or print graphics on your printer (only for printers that are capable of printing graphics, of course), you may be able to substitute another printer's driver. Try installing a driver for a similar printer made by the same manufacturer; look for a similar model number. For example, if you have a Hewlett-Packard 600 series inkjet printer, you might be able to get it to work with drivers for HP's 500 series.
Also, since many printers are compatible with Hewlett-Packard's PCL printer control language (PCL3, PCL5, etc.), you may be able to use the driver for the Hewlett-Packard Laserjet Series II for laser printers, or the Hewlett-Packard Deskjet driver for inkjet printers. If you have a Postscript laser printer, you should be able to use the driver for one of the Apple Laserwriter varieties.
Aside from drivers and cabling, common printer problems involve incorrect paper: use laser paper for laser printers, and inkjet paper for inkjet printers--none of this multipurpose junk.
Scanners and cameras
Scanners require not only the appropriate drivers to function in Windows, but special scanning software as well; as with tape drives, the two usually come together. If you can't find drivers or software that specifically supports your scanner, you're probably out of luck. However, since many companies simply repackage scanners made by other manufacturers, you may be able to obtain a driver from the original manufacturer (OEM) of the stuff under the hood.
As for cameras, since there are so many different kinds, probably the only productive discussion involves how they connect to your computer. Typically, this is through the parallel port.
Scanners commonly are connected through SCSI ports, but can also plug into parallel ports as well as proprietary controller cards. Getting scanners to work almost always involves getting the controller cards to work. See the previous discussion of SCSI cards for more information. See "Motherboards and CPUs," later in this chapter, for more information on ports.
Sound cards
Any sound card worth its weight is compatible with the Sound Blaster sound card by Creative Technology, originally called Creative Labs. If you can't find a driver for your sound card, try a driver for one of Creative's models.
If your sound card is older and doesn't support digitized sound (prerecorded sound effects and speech), it may still support MIDI synthesis (cheesy synthesizer music), and should be compatible with the driver for the Ad Lib card.
Windows should be able to detect your sound card, as well as the resources it uses. If your system crashes while trying to play sound on your sound card, and you know the correct driver is installed, try changing the resources used by the card; see "Resolving Conflicts and Other General Problems" earlier in this chapter for more information. If you can't get your sound card to work at all, and can't contact the manufacturer for the latest 32-bit drivers, you're probably going to have to replace the card.
Network cards
Windows should be able to detect your network adapter and install the correct drivers for it automatically. However, there are so many different types and manufacturers of network cards, and so many of those are proprietary, that you may be out of luck if you can't obtain drivers made specifically for your version of Windows. Note that due to the nature of Windows, you will have several drivers installed for any given network adapter, including the dial-up adapter, which isn't a network card at all (see Chapter 7 for more information). Since there are no "generic" or "standard" network drivers, if you can't find a driver for your network adapter, your only hope, other than replacing the card, is to use a driver for another card.
Memory
Bad memory can manifest itself in anything from frequent error messages and crashes to your system simply not starting. Errors in your computer's memory (RAM) aren't always consistent, either; they can be intermittent, and can get worse over time.
Nearly all newer computers, at least those capable of running Windows, use single inline memory modules (SIMMs), which are easy to install and remove. Most motherboards have about four SIMM slots, which should be clearly numbered on the motherboard. You'll either see something like slot 0, slot 1, slot 2, and slot 3, or bank 0 and bank 1. A bank is simply a pair of slots; some types of SIMMs must be installed in groups of two. All SIMMs must be installed in the lowest numbered slots first. You can't have a SIMM in slot 3 without having SIMMs in slots 0, 1, and 2.
The first thing you should do is pull out each SIMM, and make sure there isn't any dust or other obstruction between the pins and your motherboard. Use a dry tissue or lens cleaning paper; don't use any liquids or solvents. Look for broken or bent pins, broken SIMM holders, metal filings or other obstructions, and of course, any burn marks. Make sure all your SIMMs are seated properly; they should snap into place, and should be level and firm (don't break them while testing their firmness).
If all that is in order, there are three ways to determine if your RAM is actually faulty. The first way is to use a software testing program that is capable of checking RAM, such as CheckIt for DOS, a commercial package. Use the program to run a continual test of your RAM; have it repeat the test many times, perhaps overnight. The downside of testing your RAM by using any type of software is that this testing is not 100% reliable. Also, once you've found a problem, you need to follow the next method anyway to find and replace the faulty SIMM.
The second method requires a friendly, patient, helpful person at a small computer store--a rare commodity these days, especially with the popularity of large, faceless mega-super computer marts filled with inexperienced technicians. Look for a local mom-and-pop type store, and see if they have a memory testing device. These devices are too expensive for the average user, but almost anyone who sells RAM will have one. Take all your SIMMs in and ask the salesperson to check them for you. Not only is this test very reliable, but the person doing the testing will instantly be able to match whatever memory you need. Don't let them charge you for this service--they have enough to gain by selling you replacement memory if any of your SIMMs turn out to be faulty.
The third method of finding and replacing bad RAM is to go to your local computer store and just buy more RAM--it's only necessary to buy one SIMM, as most likely only one SIMM in your system is faulty. Make sure you get the right kind (32-pin, 70-pin, 168-pin, etc.), and the right capacity.
Systematically replace each SIMM in your computer with the one you've just acquired, and test the system by turning it on. If the problem seems to be resolved, you've found the culprit--throw it out immediately. If the system still crashes, try replacing the next SIMM with the new one, and repeat the process. If you replace all the memory in your system and the problem persists, it may be a bad CPU or motherboard. To eliminate the possibility that the problem is caused by a device other than the RAM, remove all unnecessary devices from your system before testing your RAM in this way.
You can, of course, also take this opportunity to add more memory to your system; at the time of this writing, memory costs a little over a dollar per megabyte, down from about $45 per megabyte at the release of Windows 95.
Motherboards and CPUs
There's really nothing you can do to diagnose a bad CPU chip (recognizable by frequent system crashes, or your machine not booting up at all), other than to simply replace it. Your best bet is to take your motherboard, complete with CPU and memory, into your local computer store, ask them to test it for you, and replace any components that need replacing.
Motherboards can also be finicky, but a problem may be caused only by a misconfiguration in the motherboard's BIOS setup. Consult the documentation for information on how to enter the BIOS setup; this usually involves pressing Del or Ctrl-Alt-Enter just after you turn the system on and memory is counted.
Also, check with the manufacturer of the motherboard to see if newer firmware (see "Firmware" earlier in this chapter) for your motherboard is available; newer motherboards allow you to update the BIOS by simply downloading and running a small program.
Lastly, even the newest motherboards come with jumpers (tsk, tsk). It's best to go through the entire manual and verify that each jumper is set correctly.
Error Messages
Error messages rarely describe a problem accurately, essentially because developers rarely put any thought into them. Usually, the computer will report that a program has crashed or isn't able to load, but the problem may be something completely unrelated to what the message is reporting. There are many different kinds of error messages, but only the more "popular" ones displayed by Windows are discussed here. If you're looking for a list of all possible error messages or codes, you're out of luck, since it doesn't exist.
If only one specific application displays a particular error message, your best bet is to contact the manufacturer of the application for technical support with their product. Many companies now have troubleshooting, updates and patches, and frequently asked questions (FAQs) on their web sites.
Common error messages usually tell you that a file is missing or corrupted, an error has occurred, or a specific device isn't working or turned on. Error messages telling you that you've done something wrong, such as trying to drag-drop a file onto your CD drive or trying to use quotes in a filename, don't really apply here, for obvious reasons.
Errors during startup
You may have seen a strange message when Windows is loading, either during the display of the Windows logo screen or after the Taskbar appears. This can be caused by many different things, but there are a few common culprits:
- A driver won't load.
- When Windows is starting up, it loads all of the installed drivers into memory. A driver may refuse to load if the device for which it's designed isn't functioning or turned on, or if the driver itself isn't installed properly. If you remove a device, make sure to take out the driver files as well--even if it isn't generating an error message, it could be taking up memory. Driver errors may also be caused by hardware problems. Refer to the "Hardware Configuration" section later in this chapter for more information.
- A program can't be found.
- After Windows loads itself and all of the drivers, it loads any programs configured to load at startup. These include screen savers, scheduling utilities, all the icons that appear in your tray, and any other programs you may have placed in your Startup folder or that may be configured in the system Registry to load automatically. If you remove an application, for example, and Windows continues to attempt to load it at startup, you'll have to remove the reference manually, as discussed under "Please wait while Windows updates your configuration files," later in this list.
- A corrupt Registry.
- See Chapter 3, The Registry, for any errors regarding your Registry.
- A corrupt or missing file.
- If one of Windows' own files won't load, and you're sure it isn't because of a third-party driver or application, you may actually have to reinstall Windows to alleviate the problem. I'll take this opportunity to remind you to back up frequently.
- An error message of this sort usually includes a filename. To isolate the problem, you must search your hard disk for the reported file, and look in several places in Windows for the reference to the file. If you don't know what exactly the error means, you should definitely do both; a lot can be learned by finding how and where Windows is trying to load a program. However, if you know that the file or files are no longer on your system, you can proceed to simply remove the reference. Conversely, if you know the file is still on your system, and you want to get it working again, you'll probably need to reinstall whichever component or application it came with to fix the problem.
- Please wait while Windows updates your configuration files.
- While this isn't an error, it is a message you may see occasionally when Windows is starting. It simply means that Windows is copying files, such as those required by software installed the last time you used Windows, that it wasn't able to copy while running. For example, if a program you install needs to replace an old .dll in your \Windows\System folder with a newer version, but the .dll is in use and can't be overwritten, the program's setup utility will simply instruct Windows to overwrite it automatically the next time it's restarted.
The following locations are places that files or drivers can be specified to load when Windows starts. Often simply removing the reference to the file solves the problem. At the very least, locating the driver will help determine the culprit:
- Search your hard disk for the file (right-click on the My Computer icon, select Find, and type the filename). If you find a .dll, .exe, or .vxd file, but you're still not sure what it is or how it got there, right-click on it and select Properties. Click on the Version tab to view the various information presented. Software developers will often place the name of the application or manufacturer associated with the file here. If the version tab isn't there, neither is the information. You can also QuickView the file, if you have QuickView installed, to see other, more cryptic information that may be of help.
- Look in your Startup folder (usually found in your Start Menu folder--\Windows\Start Menu\Startup, by default) for outdated or unwanted shortcuts. A previously installed application may have placed a shortcut here for some reason. If you have moved or deleted the application, the shortcut may still be there, irritating you every time you turn on the system. Right-click on any shortcut and select Properties to learn more about it.
- Older programs might still install themselves in your Win.ini file (anywhere on the line that starts with
LOAD=orRUN=). Use a text editor such as Notepad to edit this file. You may want to back it up before proceeding. Also, some older drivers are specified throughout your System.ini file, but can be hard to isolate. See Chapter 3 for more information on .ini files.
- Search your Registry for the filename in the message. If you don't know the name of the file, try looking in any of the following registry keys for other programs and drivers loaded at Windows startup:
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\ CurrentVersion\Run
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\ CurrentVersion\RunOnce
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\ CurrentVersion\RunOnceEx
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\ CurrentVersion\RunServices
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\ CurrentVersion\RunServicesOnce
HKEY_CURRENT_USER\SOFTWARE\Microsoft\Windows\ CurrentVersion\Run
HKEY_CURRENT_USER\SOFTWARE\Microsoft\Windows\ CurrentVersion\RunOnce
Note that in any of these keys, you may see a reference to a file without any description, such as Wfxctl32.exe. Obviously, the filename alone is not self-evident. However, a quick search on the hard drive will reveal that the file is located in the \Program Files\WinFax folder, evidence that the file belongs to the WinFax application.
Programs notorious for putting things in these places include backup utilities that automatically load their useless scheduler programs, antivirus utilities, fax programs, and the software that comes with older versions of Microsoft mice and keyboards. If in doubt, throw it out.
Page fault, Illegal operation, and Fatal exception
These errors are usually caused by a bug in software, where an application or driver tries to use part of your memory that's currently being used by another program. You should expect this to happen occasionally, due to the complexity of today's software, but if it happens more frequently than once a day, it could be the sign of a more serious problem. When software crashes, it can either cause one of these error messages to be displayed, or can cause your system to hang or even restart.
Often these error messages are accompanied by lists of numbers (accessible by clicking Details). Now, one would expect the Details option to present pertinent, useful information: what happened, why it happened, and how to keep it from happening in the future. Unfortunately, Microsoft generally doesn't like to bother with descriptive or helpful error messages, so we're stuck with this useless box. The information in the Details view is of absolutely no use to end users,[7] unless, for example, you're specifically asked to relay this information to someone in technical support. Otherwise, don't even bother with the Details; swallow your pride and click the OK button, knowing full well that it's not okay, and it never will be.
The first step is to see if you can reliably reproduce the problem. If it seems to be application- or device-specific, where the same action in a program or the repeated use of a certain device causes the crash, then you've found the culprit. This is the most common cause of crashes, especially today with companies shortening application testing periods to get their products to market faster. In the same way, buggy drivers can often cause crashes. For example, if your system crashes every time you try to use your scanner, the first thing to check is the scanner driver. It's then only a matter of fixing the problem. See "Drivers and Installing Hardware" earlier in this chapter for more information.
If the occurrences appear to be random and are not associated with any piece of hardware or software, there are some remaining possibilities. Errors in your system's memory and on your hard disk can cause these problems as well. To diagnose and repair any problems on your hard disk, use the Scandisk utility included with Windows, or one of the more-powerful third-party utilities available. While there are programs that can test system memory, a quicker and more reliable method is to replace your computer's memory, one SIMM at a time, until the problem is solved--see the section "Memory" earlier in this chapter for more information.
Crash and burn
If an application crashes and doesn't display an error message, it usually has just frozen. Depending on the severity of the crash, the application may not be responding to the mouse or keyboard, may not be updating its display, or may have turned the screen completely black. In most situations, you can press Ctrl-Alt-Del to display the Close Program box, as shown in Figure 6-4.
Figure 6-4. Pressing Ctrl-Alt-Del displays the Close Program box, allowing you to close a hung application
![]()
You are now fortunate to have the option of closing a misbehaving program, or shutting down Windows completely. You can close any running program here by selecting it from the list and clicking End Task, whether it has crashed or not. In fact, this is a good way to close programs that don't have windows, such as screen savers and other background programs. (Note that 16-bit applications closed in this way may behave unpredictably.) If an application has stopped responding, you'll notice that "not responding" appears next to the application's name. A program doesn't have to be misbehaving in order for you to have access to this window, however.
Applications that aren't responding don't always allow themselves to be shut down. Conversely, it's possible that an application reported as not responding may be doing so only temporarily; some programs, such as those that access certain devices, may appear to hang during normal operation. For this reason, it's best to have patience and give all hung applications a few minutes to correct themselves. Additionally, once you've used End Task on a hung application, it may take a little while for Windows to close it.
Another thing to be aware of is that many applications are made up of several components. It's possible for one of those components to crash and leave the rest of the program in operation. It's also possible for a program to crash and leave one or more of its components in memory. This can cause problems, although it certainly varies with the program. Try pressing Ctrl-Alt-Del again to see if the program is still running after it's been closed.
Since any crashed program can cause systemwide instability in Windows, it's good practice, though aggravating, to restart Windows every time a crash occurs. The following programs are part of Windows, and commonly appear in the Close Program window. Closing them with the method just discussed can solve some problems temporarily, but may not stop many other problems from reoccurring.
- Msgsrv32 (not responding)
- This program is one of Windows' components that runs invisibly in the background. It performs several necessary functions, including the following (this information is taken from Microsoft's knowledge base):
Mediates Plug-and-Play messages among various parts of the operating system.
Coordinates automatic responses to setup programs. This includes checking whether a setup program has improperly overwritten Windows files, and optionally restoring the Windows versions of those files--see "Your Sysbckup folder" later in this chapter.
Displays the initial logon dialog box if multiple users are configured, or in some circumstances, if networking is enabled.
Plays the system startup and shutdown sounds.
Loads installable Windows drivers at startup and unloads them at shutdown.
Runs the shell program (usually Explorer.exe) and reruns the shell if it closes or fails to respond.
Because of the varied nature of this program, almost anything can cause it to crash. If it does, you can end its task. If this happens more than once, it's best to restart Windows immediately.
Rundll32.dll This program is used by some applications to run parts of other applications, and therefore doesn't necessarily correspond with any particular error. However, the following two circumstances relate to this file:
If you see the message "Rundll32 - This program has caused an illegal operation and will be shut down," it could be caused by a missing entry in your System.ini file. Using a text editor such as Notepad to edit the file, look for the line that starts with drivers=in the[boot]section. If mmsystem.dll isn't specified on the right side of the equals sign, add it. If you continue getting this error, try removing all other drivers from the line until mmsystem.dll is alone. Make sure to make a backup of System.ini before editing it.
If the Close Program window reports "Rundll32.dll (not responding)," end the task, and try whatever it was you were doing again. This has been known to happen when a modem fails while trying to connect to another computer using Dial-Up Networking (see Chapter 7 for more information).
Explorer Explorer is the default Windows shell; it handles the desktop, the Taskbar, the Start Menu, the My Computer window, the Control Panel, Dial-Up Networking, and all single-folder and exploring windows. Unless a different application is configured as the shell (listed in the shell=line in System.ini), Explorer.exe is loaded automatically when you log in.
Explorer can crash for no apparent reason, and at any time. In most situations, if Explorer crashes or is terminated, it will be automatically restarted by Msgsrv32.exe (explained previously). However, if Explorer doesn't automatically restart--for example, following the "Abnormal Program Termination" error--you'll have to restart it manually. Since Explorer isn't running, pressing Ctrl-Esc or double-clicking on the desktop will launch the Task Manager (Taskman.exe). Select Run Application from the File menu, type explorer, and press Enter. This should restart Explorer, which will load the desktop and Taskbar. If it doesn't, press Ctrl-Alt-Del to display the Close Program box, and then click Shut Down.
Under most circumstances, Explorer crashing and restarting shouldn't adversely affect any other running applications, although you may find that the mouse isn't behaving or has even disappeared, drag-drop no longer works properly, the tray has vanished, or other minor and not so minor symptoms appear. As with other crashes, you should save any open documents and restart Windows at your earliest convenience.
Missing files Files can simply vanish from your hard disk, because of user error or file corruption, or because another program has removed or renamed the file. There's not much you can do about this, other than to keep backups of files that tend to disappear, making it easy to replace them when necessary. If you discover that a file that came with Windows or another application is missing, you usually need to reinstall the application to restore the file. Note that if a file is missing, it's likely that other files are missing as well, and reinstalling insures not only that all necessary files are present, but that any associations between the files are correct.
If you know that a single file is missing, you can usually retrieve it from the distribution disks. To save distribution costs, companies often compress their files, thereby reducing the number of disks required to store the application as well as the amount of time required to install it. If you're trying to retrieve a file from a newer Microsoft application (including Windows), the files are stored in cabinet files (see "Using the Windows Registry Checker" in Chapter 3 for more information). Otherwise, the company may be using the industry standard ZIP format (see Appendix E), or Microsoft's old Compress/Expand utility; if the names of the distribution files look like Ctl3d.dl_, you should be able to restore them to their usable state by typing expandctl3d.dl_ ctl3d.dllin DOS.
Shutdown problems If you've ever tried to shut down Windows, and have watched it simply hang at the "Please Wait" screen, or even reboot rather than shutting down, you've encountered a common problem with Windows 98.
This problem was even worse in Windows 95, which tried to unload all device drivers from memory before shutting down. If a particular driver did not allow itself to be unloaded, the system hung, and you were forced to turn it off or reboot manually.
Windows 98 theoretically solves the problem by no longer trying to unload drivers, although it still may hang at the shutdown screen for no apparent reason. If Windows 98 often hangs at the shutdown screen, you should try disabling Windows' "fast shutdown" feature. Run the System Configuration Utility (msconfig.exe) and choose the General tab. Click Advanced, turn on the Disable fast shutdown option, and click OK twice when you're done.
Another cause of shutdown hanging may be corrupted shutdown bitmaps (see "Replace the Startup and Shutdown Screens" in Chapter 2, Customizing the Interface).
Regardless of the cause, there should be no adverse effects if you manually shut off your computer whenever it hangs at the shutdown screen, but wait for the hard drive light to go out, as the drive may be saving important data to your disk.
Version control
Here's a tough one: many different programs share files, called Dynamic Link Libraries (DLLs), with one another. These files provide functionality that many applications share, such as the "File Open" box. In fact, Windows is primarily a collection of .dll files used by the various applications that run in the operating system.
The problem arises when an errant application overwrites a newer version of a .dll with an older or different version. This problem has been addressed with something called Version Control, where each file contains specific version information. For example, when you install a new application, the setup program will test every file it copies to your hard disk against any files it's replacing. If it detects that a file on your hard disk is newer than the one that came with the application, it will skip the file. Using a file's version information is more reliable than simply counting on file dates, which can be easily changed. For example, all files released with Windows 98 have the same 5-11-98 date, regardless of their version.
In theory, this works quite well. However, older applications don't always follow the rules, and newer applications sometimes come with shared files that introduce new bugs. Since the .dll files that come with Windows are used by the majority of applications, as opposed to .dll files used by only a single program, Microsoft DLLs are under the most scrutiny. Microsoft is notorious for replacing files shared by many different programs with new, buggy versions that either cause problems, or change some basic interface components to suit their own products. While this practice of Microsoft's is frowned upon, their monopolistic position in the industry pretty much insures that they can do what they want, and let other manufacturers clean up the mess.
To determine the version of any file, right-click on it in Explorer, and click Properties. You should see a Version tab; if not, the file you've chosen doesn't contain any version information. This tab displays the version of the file, some copyright information, usually the name of the manufacturer, and a short description of the file. Shuffle through the items in the Item name list to see the various clues.
You may find that a more effective solution is to use VersionTracker, part of O'Reilly Utilities--Quick Solutions for Windows 98 Annoyances (see Appendix F, Software to Solve Annoyances), which allows you to simply type in the name of a file to see the latest version. Not only is this faster than having to find the file, but if there's more than one copy of the file on your system, VersionTracker will tell you in no uncertain terms which copy is being used.
File types that usually contain version information include .dll files, .exe files, .drv files, .vxd files, and .ocx files.
Newer versions of .dll files usually serve the same purpose as the original version, but add more functionality, include bug fixes, or improve performance. In some isolated situations, a certain .dll file can be replaced with a completely different file, with which it shares only the filename. This is rare, with the possible exceptions of Winsock.dll and Mapi32.dll.
Occasionally, a program will complain that a .dll or other shared file cannot be found. For some reason, this happens quite often; files that were initially installed with an application simply vanish. While reinstalling the program usually solves the problem, you may be able to resolve the problem more quickly by looking on the application's distribution disk or CD for the file. If you find it, try copying it to the application's installation folder, or if that doesn't work, to your \Windows\System directory.
Now, there are more .dll files that can be listed in any one place, let alone in this book. Many come with Windows, and many more are installed on your system by the thousands of applications and drivers currently available. What follows is a list of a few common troublemakers, and how to cope with them. This should in no way be considered a comprehensive guide to resolving conflicts between all applications, but rather should be viewed as an exercise in dealing with several isolated problems. Essentially, the solution to any problem is to make sure you have the correct file for your system, not necessarily the newest one.
- Wsock32.dll
- This file and its 16-bit counterpart, Winsock.dll, are used by most, if not all, Internet applications, and the version on your hard disk depends on which Internet dialer you're using. If you're using Dial-Up Networking, which comes with Windows 98 (see Chapter 7), the file should be dated 5-11-98, although it may have been replaced by Microsoft with a more recent file. Updates to Internet Explorer may come with a newer version of Wsock32.dll.
- The problem occurs when another Internet dialer, such as Trumpet Winsock, AOL, or an older version of Netscape Navigator, replaces the file on your hard disk with its own version. Since the various creators of this file have included entirely different functionality in each version, it is not necessarily advantageous to have the latest version of the file. If your Internet applications stop working, check this file. You may need to replace it with the one appropriate to your dialer. See "The path less traveled," later in this chapter, for an important consideration.
- Mapi32.dll
- Mail Application Programming Interface (MAPI) is the specification that allows any application to email a file or block of text using the installed email program. However, since Microsoft neglected to include the necessary customization hooks in the file, each email program you install must overwrite the file in your \Windows\System folder with its own version.
- This causes substantial problems. For example, Eudora (a very popular email program and, in my opinion, the best available) includes its own version of the file. If you install Eudora and enable its MAPI feature, it will rename any existing Mapi32.dll file to Mapi32.000 and put its own version in its place. If you happen to have Microsoft Office 97 installed, this action will entirely disable Outlook 97,[8] which relies heavily on Microsoft's version of the Mapi32.dll file. To resolve the problem in this case, you'll need to first disable the MAPI feature in Eudora, and then restore Microsoft's version of the file.
- Ctl3d32.dll
- This file and its cousins, Ctl3d.dll and Ctl3dv2.dll, seem to cause lots of problems. They're used by older applications to display certain 3-D effects, wherein the controls in some dialog boxes have the "carved" look that has become the Windows standard. Software designed especially for Windows 95/98 already contains this functionality, but older applications rely on this file for basic operation. Although the newer versions usually are preferred, some applications will display an error if you use anything but the specific version that came with the product. You guessed it: bad programming. As of this writing, the most benign version of each of these files is as follows: for Ctl3d32.dll, Version 2.31; for Ctl3d.dll, Version 2.05; and for Ctl3dv2.dll, Version 2.31.
- If you don't have these versions, don't panic, and don't waste time trying to obtain them. This is just a guideline if you are having trouble with any of these files, or find that you have several versions floating around. Make a habit of backing up these files so that you can easily restore them if they become overwritten.
- Threed.vbx
- This is a custom control used by some applications written in the Visual Basic programming language. Custom controls like 16-bit .vbx files and 16- and 32-bit .ocx files are special .dll files that add modular functionality to programs. Threed.vbx is just one of these, and is mentioned because of its popularity. The problem of incompatible versions actually exists with all .vbx and .ocx files, now called ActiveX controls, just as with all .dll files. Be aware that these files have version information as well, and that newer versions of these files can be overwritten with older versions.
The path less traveled
While it isn't emphasized now as much as it was in the heyday of DOS and Windows 3.x, the system path is still an important setting in Windows 98. It can be helpful or detrimental, depending on how it's used.
The system path, a listing of folder names that is initialized when Windows boots, is an environment variable that is kept in memory until you shut down. If a folder name is listed in your system path, you'll be able to run a program contained in that folder without having to specify its location. For example, the MOVE utility, located in your \Windows\Command folder, can be executed from any directory in DOS, without having to specify the path, and without having to change the directory to the containing folder. (See Appendix B, DOS Lives, for more information on DOS commands.) Likewise, in Windows, if you select Run from the Start Menu, type
Notepad, and press Enter, Notepad will be run, even though you didn't specify the full path of the executable (as inc:\windows\ notepad.exe). Both of these are possible because both the \Windows\command and \Windows folders by default are included in the system path.You may not think this applies to you, but it may. The same rules that apply to program executables also apply to shared files, such as .dll and .vbx files. If you have multiple versions of a file floating around in different directories specified in your system path, any of the available versions of the file may be in use--not necessarily the latest one. For example, say you have the appropriate version of the file Winsock.dll in your \Windows\System directory, but you have another, older copy of Winsock.dll in your c:\AOL directory. It's entirely possible that some programs might mistakenly use the older version, just because they found it in a directory in the path.
How do you escape this trap? First of all, remove any unnecessary directories from your path, which is configured in your Autoexec.bat file. If you have a custom path configured, it will look something like this:
PATH=c:\norton;c:\;c:\aol;c:\progra~1\micros~1Each entry is separated by a semicolon, and any long filenames are shortened to their short filename counterparts: c:\Program Files\Microsoft Office 95\ will appear as c:\progra~1\micros~1. If you don't have a custom path configured, there will be no
PATH=line present, and you can move on. By default, \Windows, \Windows\Command, and \Windows\System are all in your path, although they are built into the operating system, and won't show up here.If the path ends up leading nowhere, there's another scenario that might be causing a file conflict. As an example, take the Ctl3d32.dll file discussed earlier in this chapter. Say you have the latest version of this file in your \Windows\System folder, where it should be. All applications that use this file will look for it there first, unless there happens to be a copy of the file in the application's own directory. If the file is not already in memory (it's loaded only once, no matter how many programs are using it), the application will load the first version it finds. If there's a copy of the file in the application directory, it will be loaded, even if it's older.
What's worse is that any subsequent programs that also use this file will simply use the one in memory, even if it's not the one in the \Windows\System folder.
Luckily, this is easy to solve, albeit somewhat difficult to isolate. Simply search your entire hard disk for the file; select Find and then Files or Folders from the Start Menu, and choose My Computer in the Look in list. If you see more than one copy of the file in the search results window, it could be causing a potential conflict.
You might want to make backups of all the found files before continuing.
Widen the In Folder column in the search results window so you can see where each file is located. If one of them is in \Windows\System (or in any other Windows subdirectories, for that matter), then it most likely belongs there. Compare the versions of the files by right-clicking, selecting Properties, and clicking on the Version tabs (see "Version control," earlier in this chapter). Now, you want to end up with only the newest file on your system, so what you can do at this point is simply delete all but the newest file, and move it to the \Windows\System folder if it's not already there.
Note that this solution by no means applies to all .dll files, which is why it's smart to back up any files before continuing. Some files have identical names only by coincidence, although this is rare. Of course, deleting a file just because there's another around by the same name is not a good idea unless you know that the files serve the same purpose. One way to make sure is to look through all the information in the Version tab; if the Company Name and Product Name are the same, you can be pretty sure that the files are duplicates. On the other hand, if the files have vastly different sizes, odds are that one is not a suitable replacement for the other.
Note also that if you've configured Windows to display hidden files, you may also find duplicates of some .dll and .vxd files in a folder called Sys-bckup (explained in the following section). It's important that you don't delete any files from this folder unless you update them with newer versions.
Your Sysbckup folder
In Windows, you'll find a hidden folder called Sysbckup under your \Windows directory. This folder contains copies of some of the more important files used in Windows, kept around in case the originals get overwritten or corrupted. If an application replaces a file in your \Windows\System directory that's also in your Sysbckup folder, Windows will automatically replace the new file with what it believes is a reliable copy from the Sysbckup folder.
You can use this to your advantage by placing copies of any files you wish to protect in this way into the Sysbckup folder. Be careful when doing this, however, not to replace any vital files with older versions, and make sure not to delete any files in this directory. For more information, see the reference to the "Msgsrv32 (not responding)" error in the section "Crash and burn" earlier in this chapter.
Preventative Maintenance
and Data RecoveryThere's no substitute for backing up, and there's no better method of disaster recovery than having a good copy of all your data. Any stolen or damaged hardware is easily replaced, but the data stored on your hard disk is not. Unfortunately, hindsight is 20/20, and if you didn't back up, there's not much you can do about it after the fact. So, we'll begin our discussion with some preventative maintenance before covering any disaster recovery techniques.
Back Up Your Entire System
There are more ways to back up your data than by storing it. The sole purpose of a backup is to have a duplicate of every single piece of data on your hard disk that can be easily retrieved in the event of data loss. Imagine if your computer were stolen, and you had to restore a backup to a brand-new computer. Could you do it? If the answer is no, you're not backed up.
You need to be able to complete a backup easily and often, store the backup in a safe place away from the computer, and retrieve all your data at any time without incident. If it's too difficult or time-consuming, odds are you won't do it--so make it easy for yourself.
The bare minimum backup should entail one or more floppy disks containing your recent, most important documents and data. Floppies are cheap, and all computers have floppy drives; the downside is that floppies are slow, too small to back up large amounts of data, and very unreliable. While it's most important to back up your recent documents, it's best to have all your documents as well as Windows and all your applications backed up as well. This will save you time and aggravation when you need to restore your system to full working order. If you only back up some of your files, you'll have to reinstall all your applications, reconfigure their toolbars and other preferences, reconfigure your Dial-Up Networking connections, and handle a myriad of other overwhelming tasks.
Ideally, you should be able to back up your entire hard disk on a single piece of media. Since floppies aren't suitable for this task, you should invest in a backup solution if your data is at all important to you. The hardware you use should be fully supported by Windows; you can't use 16-bit software to back up long filenames, so make sure you have 32-bit backup software that supports your backup hardware. The backup media (tapes, cartridges, or disks) should be cheap and reliable, and you should be able to use them over and over again.
Which backup solution is appropriate for you depends on your work habits. Tape drives, optical drives, removable cartridges, and recordable CDs are selling like hotcakes these days, and for good reason. While removable cartridge drives (such as Syquest drives, or IOmega Zip and Jaz drives) and recordable CDs are great for quickly archiving data, they still aren't as appropriate as tape drives for backing up entire systems and restoring them in the event of a disaster.
Removable drives and CDs offer random access, meaning that you can open Explorer and read or write to any file immediately. This may be convenient in the short run, but this convenience comes at a price; the media used for these types of backups can be quite expensive, and the backup procedures for random-access drives can be more labor-intensive than for tape drives.
Tape drives are still the most cost-effective, reliable, and appropriate method for backing up and recovering your system after a disaster. The caveat is that tape drives require special backup software,[9] and tend to be slower than comparably priced removables, especially when restoring single files. However, remember what's important here: you need to easily and painlessly duplicate the contents of your entire system on one cartridge, and be able to restore some or all of that data just as easily.
While tape backup software may seem awkward on the surface, it's designed to allow you to perform a backup in a single step, and without user intervention. Good backup software will also make restoring easy; the best programs keep catalogs of your backups, allowing you to find a backed-up file easily and to get it back with the least amount of hassle possible.
You need to find the system that works best for you and fits in your budget. Do some research before investing in any one technology, and make sure it truly suits your backup needs. Try this: add the cost of the drive you're considering to the cost of the media required to store the entire contents of your hard drive twice, and compare it with other solutions. Table 6-1 shows six example technologies, and the cost for each to back up a 4-gigabyte (4300-megabyte) hard drive.
Table 6-1: These Average Prices for Various Backup Solutions Show That Initial Bargains Are Rarely Good Deals Technology
Drive Cost
Cartridge Cost
Cartridge Capacity
Cartridges per Backup
Total Cost for 2 Backups
DAT tape
$400
$10
4 GB
1 = $10
$420
800MB tape
$150
$29
800 MB
5 = $145
$440
Rewritable CD
$350
$25
650 MB
7 = $175
$700
1GB removable
$300
$65
1 GB
4 = $260
$820
Zip drive
$150
$10
100 MB
43 = $430
$1,010
Floppies
$30
$0.35
1.4 MB
3,000 = $1,050
$2,130
Naturally, the prices and capabilities of the various technologies will change as quickly as the weather, but the methodology is always the same. Note also that the reference to the use of a recordable CD drive specifies rewritable CDs; if you were to do a backup with standard gold CDs, you'd have to buy new disks for each backup. See the next section for the reasoning behind having enough media for two backups.
Doing your research will save you time and money in the long run, not to mention the extra peace of mind.
Tips for a better backup
The following tips should help you ensure that you will never be without adequate data protection, whether you've already invested in a backup solution or not:
- The problem with backups is that most people don't do them. A few minutes every two weeks is all it takes, and this can save many, many hours in the future. A good time to do a backup is just before lunch, just before you go home (if the computer is at work), or just before you go to bed (if the computer is at home). You can also schedule your backup to automatically and regularly occur at any time.
- Don't do a backup while you're working on the computer. Your backup program not be able to reliably back up any files that are in use. Your system will also be slower and more likely to crash if you are doing too many things at once.
- Most backup utilities designed especially for Windows 95/98 give you the option of backing up the Registry. You should always take advantage of this feature, as it is a good safeguard, offering better assurance that you'll be able to restore your Registry, should the need arise. Naturally, backing up the Registry doesn't mean you must restore it along with everything else, but it's nice to have that luxury. Note that without a valid registry backup, all those backed-up applications won't do you any good.
- Run Scandisk, or Norton Disk Doctor, if you have it, before each backup. If your disk or your files are corrupted, so will be your backup.
- Don't back up to floppies if you can avoid it. Floppies are much more likely to fail than your hard disk, although it's better than no backup at all.
- Maintain at least two sets of backups, alternating media each time you back up. If you back up to tape, for example, use one tape for the first backup, the other tape for the second backup, then use the first tape again. That way, if there's a problem with one tape or the backup is fouled up somehow, you'll still have a fairly recent backup. Just imagine if you overwrote a good backup with a bad one and the power went out in the middle; you would then have no backup at all.
- Your backups should not be kept near your computer, especially not inside the computer. If your computer is stolen, or if there's a fire, your backups will go with it. Keeping one of the backups in your car or somewhere else off the premises is a good idea. And if you make your living off a computer, you might consider keeping a backup in a safe deposit box.
- Most backup programs allow you to specify a name for the media the first time you use them (or whenever you initialize), which allows the cataloging feature to tell you on which tape a certain file resides. Make sure each of your tapes or cartridges has a unique name that matches the tape's handwritten label, which will ensure that your software identifies each tape the same way you do. Call your tapes something like "Backup A" and "Backup B," or "Larry," "Moe," and "Curly." You'll be glad you did when your backup software asks you to insert Moe, for example.
- Make a copy of your backup program on floppies, and keep it somewhere safe. If you can't install your backup software, you won't be able to access your backups. And if the backup software is on a CD, you won't be able to get at it without installing a CD driver in DOS or reinstalling Windows completely.
- Configure your system for unattended backups. Ideally, you should only have to insert a single cartridge and click "Go" to complete a backup. Don't put up with lower-capacity backup devices that require you to swap cartridges in order to do a single backup. Additionally, most backup software has options to bypass any confirmation screens; by taking advantage of these, you eliminate the possibility of starting a backup before you go home, and coming to work the next day only to see the message, "Overwrite the data on tape?"
- Don't bother with incremental backups. Most backup software allows you to do a full-system backup, and then supplement it with incremental backups which only store the files that have changed since the last backup. While this means that you can back up in less time, it also means that you'll have to restore each of those backups when recovering from a disaster--one full backup and ten incremental backups adds up to 11 restores. More importantly, incremental backups require that the original full backup is intact. If something happens to that one backup, all subsequent incremental backups will be rendered completely useless.
- Lastly, test your system; don't wait until it's too late to find that the restore process doesn't work. Just do a simple trial backup of a single folder or branch. Then, try to retrieve the backup to a different drive or folder. Not until you've successfully and completely retrieved a backup can you consider your data safe.
Better floppy formats
Diskettes are still the standard, despite the fact that they're unreliable, slow, and small. However, everyone has a floppy drive, the disks are cheap, and it's a great way to transport small amounts of data. To avoid a headache, however, always format every floppy diskette before you use it. It'll take an extra couple of minutes, but it may save you hours in the long run.
Floppies are very unreliable. They are highly susceptible to dust, damage, and heat, and can turn on you in an instant.
Use DOS to format your floppies. It's faster and more reliable than using Windows' format feature, it gives you more free space, and it yields better multitasking. Just type
Formata:/uat the MS-DOS prompt, substitutinga:with the drive letter you wish to format. The/uparameter specifies an unconditional format, meaning that it won't use up part of the floppy with the unnecessary "unformat" information that the Windows format includes.If any errors such as bad sectors or sectors not found are reported during the disk format process, throw the disk out immediately. Disks cost around 30 cents apiece; if in doubt, throw it out.
Lastly, never use floppies to store any information for more than a few hours; that's what your hard disk is for. A floppy disk backup, for example, is much more likely to die than your hard disk. Floppies should only be used to install software and transport data from one computer to another.
See "Resurrecting the floppy diskette" in Chapter 5 for additional floppy tips.
Make a startup disk
You'll never need a boot disk until your system doesn't start, and then you'll wonder why you never took the three minutes required to make one. A boot disk is just a floppy with a few special files on it, enabling you to start your system if something goes wrong with your hard disk. It's easy, quick, and very useful.
Windows even has a built-in method for doing this; double-click the Add/Remove Programs icon in Control Panel, choose the Startup Disk tab, and then click Create Disk. If you're a do-it-yourselfer, you can just as easily do this through DO