Most of the time, installation completes normally and the Windows operating system starts without any problems. Some of the time, however, installation won't complete or, after installation, the server won't start up, and you must troubleshoot to figure out what's happening. The good news is that installation problems are usually the result of something simple. The bad news is that simple problems are sometimes the hardest to find.
Note
For more information about troubleshooting and recovery, see Chapter 41. Beyond that, you'll also find troubleshooters in the Help And Support console and in the Microsoft Knowledge Base, which is available online at http://support.microsoft.com/. Both are good resources for troubleshooting.
Setup can fail for a variety of reasons, but more often than not it's because of incompatible hardware components or failure of the system to meet the minimum requirements for Windows Server 2003 installation. With this in mind, start troubleshooting by looking at the potential points of failure and how these failure points can be resolved.
If a hardware component is incompatible with Windows Server 2003, this could cause failure of the installation or failure to start up after installation. Make sure that Windows Server 2003 is detecting the system hardware and that the hardware is in the Windows Server Catalog or on the Hardware Compatibility List (HCL). As discussed previously, you can perform a compatibility check prior to installing Windows Server 2003.
Once you've started installation, however, it's too late. At this point, you have several choices. You can reboot to a working operating system and then restart the installation using Winnt or Winnt32 and the /debug Setup option. This puts Setup in debug mode and helps identify what is going wrong. If Setup determines you have hardware conflicts, you can try to configure the hardware and server BIOS to eliminate the conflicts. Troubleshooting BIOS involves booting the server to the firmware and then completing the following steps:
Examine the boot order of disk devices You might want to configure the system so that it boots first from CD-ROM. Watch out, though; after installation, don't keep booting to CD-ROM thinking you are booting to the operating system—hey, we all get tired and sometimes we just have to stop and think for a moment. If the installation problem is that you keep going back to the installation screen after installing the operating system, you are probably inadvertently booting from CD-ROM—and you're probably way too tired by now to realize it.
Check Plug and Play device configuration and interrupt reservations If a system has older components or components that aren't Plug and Play compatible, you might have a device conflict for a hard-coded interrupt. For example, a non–Plug and Play sound card could be hard-coded to use interrupt 13, which is already in use by a Plug and Play device. To work around this, you must configure interrupt 13 under your Plug and Play BIOS settings to be reserved for a non–Plug and Play device. This ensures that Plug and Play does not attempt to use that interrupt and resolves the issue in most cases.
Tip
The only sure way to avoid problems with non–Plug and Play devices is to avoid using them altogether.
Rather than spending time—which could run into several hours—trying to troubleshoot a hardware conflict, you might consider removing the hardware component if it's nonessential— and you might be surprised at what I consider nonessential at this stage. By nonessential, I mean most anything that isn't needed to start up and give you a display for logon. You probably don't need a network card, a sound card, a multimedia controller card, a video coder/decoder (codec), or a removable media drive. If these items are incompatible, you might resolve the problem simply by removing them. You can always try to install the components again after installation is complete.
When you install directly from the Windows Server 2003 CD-ROM or perform a network install from a distribution share, you might encounter a media error that causes Setup to fail. With an actual CD-ROM, you might need to clean the CD-ROM so that it can be read or use a different CD. If the CD-ROM drive is the problem, you must replace the drive or install from a distribution share as discussed in Chapter 3. If you are working with a distribution share, the share might not have all the necessary files, or you might encounter problems connecting to the share. Try using an actual CD-ROM.
Windows Server 2003 requires a minimum of 128 megabytes (MB) of random access memory (RAM) and about 1.5 GB of disk space. If the system doesn't have enough memory, Setup won't start. If Setup starts and detects that there isn't enough space, it might not continue or you might need to create a new partition or delete an existing partition to get enough free space to install the operating system.
Sometimes installation fails because you've configured the server to join a domain and Setup is unable to access the domain. Problems accessing domains might result from a network card not working properly or because you removed a network card that was incompatible.
If a network card isn't working, you should see an orange or red traffic indication light (rather than a green one) or no connection indication light on the card. Check the network cable connection between the server and the router.
If you had to remove an incompatible network card, configure the server to join a workgroup temporarily. Then, after you complete installation, you can install a new network card and get TCP/IP services working, which should allow you to join the domain.
Setup can also fail to connect to a domain controller if you've improperly configured DNS or provided the wrong account information.
If DNS is configured incorrectly, the server won't be able to connect to the domain. Check the IP address, network mask, gateway, and preferred DNS server settings. If you are using DHCP to dynamically assign the IP settings, you still might need to enter the DNS server settings.
If you provided the wrong account information, logon to the domain will fail. You must use an account with Administrator or Account Operator permissions. If the account used has these permissions, ensure that you've entered the user name and password correctly. Remember, if you are using an account from a trusted domain of the domain you are joining, you must enter the full domain and account name in DomainUsername format, such as AdatumWRStanek.
If you can get past the potential points of failure, you still might find that the installation locks up or freezes. In this case, you might get a Stop error; then again, you might not.
Most Stop errors have cryptic error codes rather than clear messages telling you what's wrong. If you get a Stop error, write down the error number or code, then refer to the Microsoft Knowledge Base (available online at http://support.microsoft.com/) for help troubleshooting the problem. To break out of the stop, you most likely will have to press Ctrl+Alt+Del (sometimes several times) to get the server to restart. If this doesn't break out of the stop, press the Reset button on the server. Alternatively, turn off the system power, wait a few seconds, and then turn it back on.
The Windows operating system should start up and go directly back to Setup. In some cases, you will see a boot menu. If so, choose the Windows Setup option to allow the Setup program to attempt to continue the installation. Setup could freeze again. If it does, stay with it, and repeat this process—sometimes it takes several tries to get completely through the installation process.
Troubleshooting: Setup freezes at a specific point
Occasionally, Setup will continually freeze at a specific point in the installation. For example, one time I was installing a server and Setup froze every time it got to the network installation. The workaround was to change the network configuration options—I had to clear the Internet Protocol (TCP/IP) networking component installation. You might find that changing or simplifying the installation process works for you as well.
RAM and central processing units (CPUs) can also be the source of problems. Issues to watch out for include the following:
Incompatible RAM Not all RAM is compatible, and you can't mix and match RAM of different speeds and types. Ensure that all RAM modules are the same type and speed. Further, RAM modules from different manufacturers could in some cases perform differently (read incompatibly), and in such a case, try changing the RAM so that all modules have the same manufacturer.
Malfunctioning RAM Static discharges can ruin RAM faster than anything else. If you didn't ground yourself and use a static discharge wire before working with the RAM modules, you could have inadvertently fried the RAM so that the modules don't work at all or are malfunctioning. RAM could have also arrived in this condition from the manufacturer or distributor. There are several troubleshooting techniques for determining this. You could update BIOS to add a wait state to the RAM so that if the RAM is partially faulty the system will still boot (but you still must replace the RAM eventually). You can also try removing some RAM modules or changing their order.
Incompatible processors Not all processors are created equal, and I'm not just talking about their speed in megahertz (which you generally want to be the same for all processors on a server). Some processors might have a cache or configuration that is incompatible with the server hardware or other processors. Check the processor speed and type to ensure that it is compatible with the server. In some cases, you might need to change hardware jumpers, depending on the speed and type of your processors.
Misconfigured processors Adding additional processors to a server isn't a simple matter of inserting them. Often, you must change jumpers on the hardware, remove several terminators (one for a power subcomponent and one for the processor—save them because, trust me, you might find that you need them), and then insert the new components. Check the hardware jumpers (even if you think there aren't any), and ensure the processors and the power subcomponents you've added are seated properly. If you can't get the installation to continue or the server to start up, you might need to remove the components you've added. Watch out, though; you probably don't want to continue the installation until the processor issue is resolved—single-processor systems have different threading and default configuration than multiprocessor systems, meaning this situation might not be a simple matter of adding the processor after installation and making it all work properly.
System processor cache problems Sometimes there can be an issue with the system processor cache and compatibility with Windows Server 2003. Consult the server documentation to read the correct configuration settings available and how the cache can be disabled. If you suspect a problem with this, boot to BIOS and temporarily disable the system processor cache, following the server documentation. Once the installation is complete, you should be able to enable the cache to avoid a performance hit. Be sure to check both the hardware vendor support site and the Microsoft Knowledge Base to see whether any known issues with your server's processor cache exist.
Troubleshooting: RAM and CPUs are incompatible
You might be surprised at how common it is for incompatible RAM or CPUs to present problems, especially when installing enterprise-class servers. We had a problem once when we ordered all the components from a single hardware vendor that had verified the compatibility of every element down to the last detail only to find that the wrong processors and RAM were shipped for the systems ordered. The result was that every time we added the additional processors and RAM modules, the server wouldn't start up. The only recourse was to continue installation with the minimum processor and RAM configurations shipped or wait until replacements arrived. Electing to wait for replacements added time to the project but ultimately proved to be the right decision. You can bet that we were glad that we padded the project schedule to allow for the unexpected—because the unexpected usually happens.
Most of the time the installation or setup problem is caused by a compatibility issue with the Windows operating system, and that problem can be fixed by making changes to BIOS settings. Sometimes, however, the problem is the BIOS, and you'll find that you must upgrade the BIOS to resolve the problem. One problem in particular has to do with Advanced Configuration and Power Interface (ACPI)–compatible BIOS dated January 1, 2000, or earlier, which might not function properly in ACPI mode under Windows Server 2003. The result is that the Windows operating system freezes during setup and cannot get past it.
One way to fix this problem (and other BIOS problems) is to upgrade the BIOS. Check with the hardware vendor to see whether a BIOS upgrade is available. If so, install it as the hardware vendor directs. If a new BIOS version isn't available, you might be able to disable the incompatible option during setup. During setup, press F5 or other designated key at the beginning of the text-mode phase (just after it prompts you to press F6 to install third-party Small Computer System Interface [SCSI] drivers). If this doesn't work, the option you changed wasn't the source of the problem and you should reenable it before continuing.
Note
Reenabling the option might be necessary because some hardware-specific BIOS settings cannot be changed after the installation. Thus, the only way to enable the option would be to reinstall the operating system.
You can also manually enable or disable advanced BIOS options. You do this by editing Txtsetup.sif in a distribution folder. This file is read after the text-mode phase of the installation and before the Windows operating system restarts to the Setup program. With ACPI, from the command line on another computer, you type attrib –r –s –h c: xtsetup.sif and then type edit c: xtsetup.sif so that you can edit the .sif file. In the .sif file, search for ACPIEnable=, change the associated line to read ACPIEnable=0 to disable ACPI support (ACPIEnable=1 enables ACPI support), and then save the changes. After you save the changes, restart into the Setup program.
Tip
Skip editing the Txtsetup.sif file
You can press F7 during the portion of Setup that displays the message to press F6 for adding SCSI drivers. This configures Setup so that it won't try ACPI machine types. It also means you aren't required to edit the Txtsetup.sif file.
Finally, hard disk drive settings could also cause lockups or freezes, particularly if you are using Integrated Device Electronics (IDE) drives. When using IDE drives and controllers, you want to ensure that the system recognizes both the drives and the controllers and that both are enabled and configured properly in BIOS. You might have to check jumper settings on the drives and the cables that connect the drives. As discussed previously, check for conflicts between the drives, controllers, and other system components. You might need to temporarily remove unnecessary components, such as the sound card, to see whether this resolves a conflict. If a CD-ROM or DVD drive is on the same channel as the disk drives, try moving it to the secondary channel and configuring it as a master device. You can also try lowering the data transfer rate for the IDE drives.