Why I Still Believe in Fixing the Actual Problem
One of my least favorite approaches to technical support is:
“Try this.”
Didn't work?
“Okay, try this.”
Still didn't work?
“Try these seventeen other things somebody posted online.”
Eventually something changes, nobody knows why, and everyone agrees not to touch the computer ever again.
That's not troubleshooting.
Start With the Symptoms
Good troubleshooting begins by understanding what is actually happening.
Not what somebody assumes is happening.
If a website is slow, saying “the server is slow” isn't a diagnosis.
Maybe it's:
- An enormous image
- A slow database query
- A third-party script
- DNS
- A failing disk
- A plugin
- Network latency
- Malware
- Insufficient resources
Each problem may look similar to the person using the website.
They require very different solutions.
Change One Thing at a Time
Randomly changing five settings makes troubleshooting harder.
If the problem disappears, which change fixed it?
If something else breaks, which change caused it?
I prefer making deliberate changes and testing the result.
Logs Are Your Friends
Error messages and logs frequently look intimidating, but they're often the system trying very hard to tell you what's wrong.
Web server logs, Windows Event Viewer, Linux system logs, browser developer tools, application logs, SMART data, and network diagnostics can turn “it doesn't work” into something much more specific.
Specific problems are easier to solve.
Fix Causes, Not Just Symptoms
Restarting a service can restore it.
That's useful.
But if the service crashes every Thursday because a disk fills with temporary files, restarting it every Friday morning isn't the solution.
Find out why.
Fix that.
That's My Approach to Technology
Whether I'm troubleshooting a Windows server, fixing a website, working on DNS, tracking down an email problem, or figuring out why a contact form stopped sending messages, my approach is essentially the same:
- Understand the system.
- Reproduce the problem.
- Gather evidence.
- Find the cause.
- Fix it.
- Document what happened.
Technology is complicated enough without making it mysterious.