ServiceDesk Simulator
← All articles

Event Viewer: How to Read a Machine's Diary Without Drowning

August 18, 2026 · ServiceDesk Simulator · 5 min read

Every tech opens Event Viewer for the first time the same way. Someone senior says “check the logs,” you double click it, and twenty thousand rows of Information scroll past from services you have never heard of. You scan for something red, find nine, none of them mention the actual problem, and quietly close the window. That first impression is why a genuinely useful tool has a reputation for being useless.

The reputation is undeserved. Event Viewer is the machine’s diary, written honestly and without the user’s memory in the way. It just has to be read in the right order.

Pick the log before you filter anything

The left tree is not decoration. Windows keeps separate logs, and choosing the right one throws away most of the noise before you touch a filter.

System is where the operating system talks about itself: drivers loading or failing, services starting and stopping, hardware appearing and disappearing, unexpected shutdowns. Application is where installed programs report their own crashes, including the office suite that “just closes itself.” Security holds the audit trail, sign-ins, sign-outs, lockouts, privilege use, which makes it the log for account questions rather than hardware ones. Setup covers updates and servicing, the log to open when a machine started misbehaving right after patch night.

Match the log to the complaint first. A printer that vanished is a System story. An application that dies on save is an Application story. “Someone logged into my account last night” is Security. Opening the wrong log and concluding there is nothing there is the most common way this tool gets written off.

A remote session connected to a user workstation in ServiceDesk Simulator
Logs get read over the shoulder of a remote session, usually while the user is still on the phone describing what they saw.

Red is not the signal you think it is

New techs hunt for red icons. Experienced ones ignore most of them, because a perfectly healthy machine produces errors and warnings continuously. Services time out and retry. A component polls something that is not there yet during boot. Certificates get checked against a network that has not come up. None of it means anything, and none of it is what the user called about.

The signal is not the level. The signal is the timestamp.

So the real first step happens before the tool opens, in the conversation: when did this start, and what were you doing? “It has been slow for weeks” gives you nothing to align against. “It crashed twice this morning, both times right after I opened the report” gives you a two hour window and a trigger. Filter to that window, in the right log, and the row count drops from twenty thousand to forty. Now the red icons mean something, because the ones that survived the filter happened when the problem did.

That ordering is the whole skill. The log is where you confirm a theory, not where you go looking for one. Techs who arrive without a theory find only noise, and they find it every single time.

The handful of IDs worth carrying around

You do not need to memorize event IDs. You do end up recognizing a few, the way you recognize phone numbers you dial often, and these earn their space.

Service Control Manager 7045 says a new service was installed. On a managed machine that is usually a legitimate agent arriving through software deployment, boring and expected. It becomes the most interesting line in the log when the service file path points into a user AppData folder, the start type is automatic, and the account is LocalSystem. Legitimate software installs into Program Files. Something that installed itself into a user profile and arranged to start before anyone signs in has told you what it is, in writing, and this is the single best reason to teach a new tech where 7045 lives.

Application Error 1000 is the plain crash record: faulting application name, faulting module, an exception code. When exception code 0xc0000005 keeps appearing against the same module, you have stopped guessing about which application is unhappy and started reading its confession.

In System, driver and device events tell you whether hardware was ever really there. A device that never loaded a driver logs a failure and then stays absent, which matters because Windows only lists a device once its driver has loaded. If the user can see it and use it, a driver failure event is the wrong story, and treating the two as interchangeable produces confident wrong answers.

In Security, the sign-in family carries the account work. Successful logon 4624 with the logon type tells you whether someone sat at the machine or arrived over the network. Its 4634 sign-out partner brackets the session. Special privileges assigned at logon, 4672, marks the sign-ins that came in with administrative rights, which is exactly what you want to review after a suspected account compromise rather than the day after.

What it is good for, honestly

Event Viewer is at its best on problems the user cannot reproduce for you. Intermittent crashes, an overnight reboot nobody admits to, a device that works until it does not, a machine that got slow on a specific day. The log remembers precisely, which beats a user recollection that has been through three retellings by the time it reaches your queue.

It is worst as a fishing expedition. Reading logs without a symptom in hand produces a list of scary looking errors, a theory built to fit them, and a fix aimed at something that was never broken. Everyone does this once. The cure is the six-step method, where the log lives in the verification step, not the discovery step.

Practicing on a machine you cannot break

Reading logs is a habit you cannot build on your own laptop, because nothing interesting has ever happened on it. That is why ServiceDesk Simulator ships an Event Viewer inside the remote session, with real log names, real sources, real event IDs, and evidence that agrees with the ticket the user filed. It is read only, so poking around is safe, and no ticket forces you to open it. It is there for the tech who wants to know whether the story adds up before they touch anything, which turns out to be the tech everyone wants to hire.

Common questions

What is Event Viewer used for?

It is the log reader built into Windows. Every service start, driver load, crash, sign-in and update leaves a row there, so it is where a tech goes to confirm what happened on a machine and when, especially for problems the user cannot reproduce on demand.

Which Event Viewer log should I look at first?

System for drivers, services and hardware, Application for a program that crashes, Security for sign-ins and lockouts, Setup for updates and servicing. Picking the right log first removes most of the noise before you filter anything.

Are Error events in Event Viewer bad?

Usually not. A healthy machine logs errors and warnings all day from services that time out and retry. An event only means something when it lines up with the time the user says the problem happened.

What event ID means malware?

No single ID proves it, but Service Control Manager 7045, a new service installed with its executable sitting in a user AppData folder and set to auto start, is one of the most suspicious things a log can show you.

Built by Rena, who broke into IT with no degree. Read her story →