NewAI Content Creation is now live in Early Access
Turning Point Training Instituteby Training Center
Skip to content
0%
Module 4 — Identifying and Defining the ProblemLesson 2 of 4
Video lesson

Evidence, Reproduction and Intermittent Faults

BeginnerVideo lesson
9 min readBeginner

Where you are: Module 4 — Identifying and Defining the Problem · Lesson 4.2 · Based on Chapter 4 (§4.3–4.4) of Dr. Tahseen Al Doori's Troubleshooting Essentials.

What you'll be able to do

  • Collect and record the five kinds of evidence the book identifies.
  • Record an error message exactly, including its code, and explain why "word for word" matters.
  • Reproduce a fault using the same steps, device and account, and explain why reproduction speeds up testing.
  • Investigate an intermittent fault with an occurrence log and recognise patterns in it.

Watch: Evidence, Reproduction and Intermittent Faults

The video shows how to capture an error message properly, why a repeatable fault is a testable fault, and how a fictional user's occurrence log exposes the pattern behind video calls that freeze "for no reason". Keep these questions in mind:

  • Why is "some network error" almost useless as evidence?
  • What did the technician's own test call have in common with none of the logged freezes?
  • Does the pattern in the log prove the cause? If not, what does it give you?

Collecting evidence

In Lesson 4.1 the evidence came from people. Here it comes from the system and its surroundings. The book lists five kinds of evidence to collect during Step 1.

1

Exact error messages

Recorded word for word, or captured as a screenshot or photograph, including any error code.

2

Physical symptoms

Indicator lights, beep codes, unusual sounds, smells, heat or vibration.

3

Logs and history

Operating system event logs, application logs, update history and recent installations.

4

Scope

One user or many, one device or all, one site or every site.

5

Recent changes

Updates, new hardware, moved equipment, password changes, power interruptions, new staff.

Why "word for word" matters

Compare two reports of the same fault:

  • "It gave me some network error."
  • "Windows cannot access \\FILES01\Shared — Error code: 0x80070035 — The network path was not found." (The server name is fictional; the dialog wording is the standard Windows one — see the Beyond the book note below.)

The first tells you almost nothing. The second names the resource that could not be reached and gives an error code that can be searched in vendor documentation and knowledge bases, exactly as Step 2 asks you to do. Record the time the message appeared too: an event log — the system's record of errors, warnings and significant events — can then be checked for entries at the same moment (Module 7 covers event logs in detail).

Beyond the book (2026 note): capturing a screenshot is quick on current systems. On Windows, pressing the Windows logo key + Shift + S opens the Snipping Tool overlay to capture part or all of the screen. On a Mac, Shift-Command-3 captures the entire screen, Shift-Command-4 captures a selected area, and Shift-Command-5 opens the Screenshot app with more options. If the device itself is not working, a photograph taken with a phone is just as useful. The shared-folder dialog quoted above ("Windows cannot access \\server\share", "Error code: 0x80070035", "The network path was not found.") is shown in a Microsoft Community Hub article. Sources: Can't connect to Windows Share - Error code: 0x80070035 — Microsoft Community Hub, Use Snipping Tool to capture screenshots — Microsoft Support and Take a screenshot on Mac — Apple Support.

Evidence can be sensitive. Screenshots and photographs may show personal data, customer records or messages. Capture only what you need, store it in the ticket rather than on a personal phone where policy requires, and never capture a password. If physical evidence includes a burning smell, smoke, sparks or a swollen battery, stop: that is a safety issue to escalate, not evidence to investigate further.

Reproducing the problem

To reproduce a problem is to make it happen on purpose so that it can be tested. The book gives two reasons it matters: a problem that can be reproduced on demand can be tested quickly, and it is obvious when it has been fixed, because the steps that used to cause it no longer do.

How to reproduce a fault:

  • Repeat the exact steps that caused it, in the same order.
  • Use the same device and the same account. If you test with your own administrator account, you may not see a problem caused by the user's permissions or settings.

If the problem cannot be reproduced, do not conclude that it does not exist. The book explains that it may depend on conditions such as:

  • the time of day;
  • network load;
  • temperature;
  • a particular user's settings.

Handling intermittent problems

An intermittent fault is a problem that comes and goes. The book calls these the hardest to diagnose, because they disappear when examined. The technique for them is an occurrence log: every time the fault happens, record

  • the date,
  • the time,
  • the activity — what the user was doing,
  • the conditions — anything that might matter: location, power source, which applications were open, what else was happening nearby.

Over several occurrences, the book notes, a pattern usually appears — for example:

  • every afternoon;
  • only when a particular application is open;
  • only on battery power;
  • only when a neighbour's equipment is running.

A pattern is not yet a confirmed cause. It is something better than a guess: a theory you can take into Steps 2 and 3 and test.

Worked example

(Fictional scenario.) Sara, a project manager, reports that her video calls freeze for about 20 seconds and then recover, a few times a week. The technician runs a 30-minute test call at Sara's desk: it is perfect. The fault cannot be reproduced on demand.

Sara agrees to keep an occurrence log for a week:

Date and timeActivityConditions
Mon 10:20Video callMeeting room, on battery
Tue 15:05Video callKitchen, on battery
Wed 11:40Video callHer desk, on battery
Thu 14:30Video callMeeting room, on battery
Fri 09:50Video callLibrary, on battery

The places and times vary, but one condition never does: every freeze happened on battery power. The technician's test call had been on the charger. Now there is a testable theory — that something about running on battery causes the freezes — and a safe test: repeat a call on battery power, after recording the current power settings so they can be restored. Whatever that test shows, the occurrence log turned a fault that "disappears when examined" into one that can be investigated with the seven steps.

Try it

Choose something in your own life that goes wrong only occasionally — for example a phone that sometimes loses signal at home, a website that is sometimes slow, or a printer that occasionally jams. Do not change anything for this activity; you are only observing.

  1. Create an occurrence log with the columns Date, Time, Activity and Conditions.
  2. Record at least four occurrences over a week (or invent realistic ones, labelled fictional, if the fault does not recur).
  3. Underline any condition that appears in every entry.
  4. Write one sentence stating a theory suggested by the pattern, and one small, safe test for it.

Completion criteria: at least four entries, each with all four columns filled; the theory is phrased as something to test, not as a confirmed cause; the proposed test changes only one thing and is safe.

Knowledge check

Which of these is NOT one of the five kinds of evidence listed in the book?

Why should an error message be recorded word for word, including its code?

You cannot reproduce a user's fault on your own administrator account. What is the best next step?

An occurrence log shows a fault only between 14:00 and 16:00 each day. What is the best way to use this?

Answers explained

  1. C — the five are exact error messages, physical symptoms, logs and history, scope and recent changes. A first impression is a theory, not evidence.
  2. A — exact text and codes can be researched in vendor documentation and matched to log entries by time. Codes do not always give the solution (C), and appearance (B, D) is not the reason.
  3. D — the book says reproduce with the same device and the same account; the user typing their own password keeps credentials private. A ignores a real fault, B is an untested fix, and C breaks the rule never to ask for passwords.
  4. B — a pattern points to theories (heat, load, scheduled tasks) for Steps 2 and 3. It is not proof (A), time of day is one of the book's named conditions (C), and further logging may refine it (D).

Key takeaways

  • Collect five kinds of evidence: exact error messages, physical symptoms, logs and history, scope, and recent changes.
  • Record errors word for word with their codes and the time they appeared, or capture them.
  • A fault you can reproduce — same steps, same device, same account — is quick to test and obviously fixed when it stops.
  • Faults that will not reproduce may depend on conditions such as time of day, load, temperature or user settings.
  • For intermittent faults, log every occurrence with date, time, activity and conditions until a pattern appears; treat the pattern as a theory to test.

Sources