Troubleshooting hardware and boot problems

Hardware faults and start-up failures are among the most stressful tickets for users, because the computer seems dead and the work on it seems lost. They are also among the most satisfying for technicians, because careful reasoning about the start-up sequence and the components usually leads straight to the cause. This lesson applies the troubleshooting method from lesson 8 to the hardware and boot problems you will meet most often: computers that will not power on, will not pass POST, will not load the operating system, crash with a stop error, overheat, or slowly fail.

Retracing the start-up sequence

Lesson 1 described what happens when you press the power button: power, POST, the firmware choosing a boot device, the boot loader, the operating system kernel and drivers, and finally the sign-in screen. When a computer fails to start, the first question is how far it gets. Each stage that completes rules out a set of causes, and the stage where it stops points at the next set to investigate.

When nothing happens

A computer that shows no sign of life is usually a power problem, and the checks go from the outlet inward. Is the outlet live? Try a lamp or a phone charger in it. Is the power strip switched on and working, or has its overload switch tripped? Is the switch on the back of the power supply on? Is the cable firmly in? On a laptop, does the charger's light come on, and does the laptop start from its charger with a known-good charger? On a desktop, a power supply tester or a known-good spare power supply settles whether the supply has failed. Less commonly, the case's power button or its connection to the motherboard header has failed; technicians can briefly short the two power-button pins on the header with a screwdriver to test, a trick that is safe because those pins carry only a tiny signal.

A related symptom is a computer that starts for a moment and then switches off. That can be a power supply that cannot hold its voltages under load, a short circuit, for example from a loose screw behind the motherboard, or a processor cooler that is not seated, causing the processor to overheat within seconds and the board to cut power to protect it.

POST failures

If the fans spin but nothing appears on screen, the computer is probably failing POST. Because the display may not work at this stage, motherboards report problems in other ways: patterns of beeps through a small speaker, coloured diagnostic lights marked CPU, DRAM, VGA and BOOT, or a two-digit code on a small display on higher-end boards. The meanings vary by firmware maker, so look them up in the motherboard or computer manual rather than guessing. In practice, the most common POST failure after any work inside the case is memory that is not fully seated. Reseating the memory, and if necessary testing with one module at a time in the recommended slot, finds faulty modules quickly. Also check the obvious: is the monitor plugged into the graphics card rather than the motherboard, switched on and set to the right input?

Boot failures

Once POST passes, the firmware looks for a boot device. A message such as No bootable device or Boot device not found means it found nothing it could start from. The causes are a boot order that puts an empty USB drive or the network first, a drive that has failed or come loose, or missing or damaged boot files. Enter the firmware setup and check whether the drive is detected at all. If it is not, the problem is hardware: the drive, its cable or its M.2 slot. If it is detected but will not boot, the problem is the boot files or the boot order.

Windows has a set of recovery tools for start-up problems, called the Windows Recovery Environment, stored on the recovery partition from lesson 5. Windows opens it automatically after start-up fails twice in a row, and it can also be reached from a Windows installation USB drive. Its options include Startup Repair, which diagnoses and fixes common boot problems automatically; uninstalling the latest quality or feature update; System Restore, which rolls system files and settings back to an earlier restore point without touching personal files; a Command Prompt for advanced repairs; Startup Settings, which can start Windows in Safe Mode; and Reset this PC, which reinstalls Windows as a last resort.

Safe Mode starts Windows with a minimal set of drivers and services. If a problem disappears in Safe Mode, it is caused by something that normally loads at start-up, such as a driver, a service or a startup program, which narrows the search considerably. From Safe Mode, a technician can uninstall a problem driver or update, run malware scans and check logs.

Stop codes and crashes

When Windows meets an error it cannot safely recover from, it stops and shows a stop error screen, often called the blue screen of death, with a stop code such as MEMORY_MANAGEMENT, CRITICAL_PROCESS_DIED, INACCESSIBLE_BOOT_DEVICE or DRIVER_IRQL_NOT_LESS_OR_EQUAL, and sometimes the name of the file that caused it. Write the stop code down or photograph it; it is the single most useful clue. A one-off crash may never recur, but repeated crashes need investigating. Stop codes that name a driver point at that driver, often after a recent update. Memory-related codes point at faulty memory, which the Windows Memory Diagnostic tool or a more thorough tool such as MemTest86 can test. INACCESSIBLE_BOOT_DEVICE often follows a change in the storage controller setting in the firmware or a failing drive.

Windows records much more detail than the stop screen shows. Event Viewer holds the system and application logs, where errors and warnings are recorded with the time and the component that reported them. Reliability Monitor presents the same information as a timeline, showing crashes, failed updates and installations day by day, which makes it easy to see whether a problem started on the same day as a change. Lesson 10 shows how to read logs for patterns.

Overheating and throttling

Heat causes a range of symptoms: sudden shutdowns under load, crashes, and computers that get steadily slower as they work, because the processor throttles itself to stay within a safe temperature. Desktop processors typically run at 30 to 50 degrees Celsius when idle and may reach 80 to 90 degrees under heavy load; most begin throttling at around 95 to 100 degrees. Causes include dust clogging heat sinks and fans, failed fans, dried-out thermal paste after many years, blocked vents, and hot surroundings, such as a computer in a cupboard or a laptop on a sofa. ROAST-LT lives in a hot room next to a coffee roaster, which is exactly the kind of setting that shortens hardware life.

Monitoring tools show temperatures, fan speeds and clock speeds. Jordan uses a simple script to read a temperature log exported from one of them and flag readings above a threshold, which makes it easy to show Sam when and how often a computer runs hot.

log = """10:00,52
10:05,61
10:10,78
10:15,96
10:20,98
10:25,84
10:30,71"""

LIMIT = 95
hot = []
for line in log.splitlines():
    time, celsius = line.split(",")
    if int(celsius) >= LIMIT:
        hot.append(time)
print("Readings at or above", LIMIT, "degrees:", ", ".join(hot) or "none")

Readings at or above 95 degrees: 10:15, 10:20

readings = [52, 61, 96, 98, 84, 95, 97, 99, 71, 60]

def longest_hot_run(readings, limit):
    return 0

print(longest_hot_run(readings, 95))

Failing drives

Storage failures deserve special attention because they threaten data. Hard drives often give warning signs: clicking or grinding noises, very slow access to some files, files that will not open, and errors during start-up. SSDs more often fail suddenly, although they may first become read-only or slow. Both kinds report their health through SMART, Self-Monitoring, Analysis and Reporting Technology, a set of counters kept by the drive. Important counters include reallocated sectors, which a hard drive has replaced because they went bad; pending sectors, waiting to be checked; and, for SSDs, the share of its rated life already used. A rising count of reallocated or pending sectors is a strong warning that the drive will fail soon. The moment a drive shows warning signs, the priority changes: copy the user's data off it first, then diagnose.

report = """Reallocated_Sector_Ct 0
Current_Pending_Sector 0
Power_On_Hours 18240
Temperature_Celsius 38
Reallocated_Event_Count 0"""

values = {}
for line in report.splitlines():
    name, value = line.split()
    values[name] = int(value)
warning = values["Reallocated_Sector_Ct"] > 0 or values["Current_Pending_Sector"] > 0
print("Power-on time:", values["Power_On_Hours"] // 24, "days")
print("Status:", "failing: back up and replace" if warning else "healthy")

Power-on time: 760 days
Status: healthy

report = "Reallocated_Sector_Ct 0\nCurrent_Pending_Sector 0\nTemperature_Celsius 38"

def drive_status(report):
    return "healthy"

print(drive_status(report))

Other common hardware faults

Throughout hardware work, keep safety in mind. Always shut down and unplug a desktop before opening it, and remove a laptop's battery connection before working inside it. Never open a power supply or a monitor: both contain capacitors that can hold a dangerous charge even after unplugging. Lesson 11 covers electrical safety in detail.