Tips for effective troubleshooting

“My printing isn’t working, why isn’t it printing?” a user asks. The answer might sound simple, but under the hood hides a number of places where an error could have occurred. Wasting two hours installing printer drivers when it was just a loose network cable in the printer (or the paper tray was empty) is a very frustrating experience that in my opinion every “IT guy” had to go through at some point.

Here is my guide on how not to go crazy while hunting for bugs. It’s called: Follow the OSI model.

The Law of the OSI Model: Bottom-up

Before you start messing around in the settings of complex programs at the Application layer (Layer 7), make sure basic physics and networking are working.

graph TD
    Layer1((1. Physical Layer)) --> Layer2((2. Data Link Layer))
    Layer2 --> Layer3((3. Network Layer))
    Layer3 --> Layer7((7. Application Layer))
    
    Layer1 -. "Is it plugged in? Is there power? Is the cable clicked in?" .-> Layer1
    Layer2 -. "Is the light on the Switch on? Can you see the MAC address?" .-> Layer2
    Layer3 -. "Did it get assigned an IP address via DHCP? Does it ping?" .-> Layer3
    Layer7 -. "Is the program running? Do we have a license for it?" .-> Layer7

A specific example with a broken phone:

• Layer 1 (Cabling): Is the phone even plugged in? No? (Troubleshooting would have ended 20 minutes earlier here, rather than installing a new SIP protocol firmware at Layer 7).
• Layer 3 (Network): Does the phone have an IP address? If not, restart the port, or check DHCP.
• Layer 7 (Application): After clicking in the software on the PC, it doesn’t dial the client’s number on the desk. Only now do we investigate if the PBX has been renamed.

Divide and conquer

When I finally know which layer to search in, I try to isolate the problem as much as possible.

Splitting it in half: “It works for the user from a mobile phone on that Wi-fi. It doesn’t work from the laptop.” => The error logically doesn’t lie in the router’s transmitter nor in the company’s internet, but with 99% certainty in the Wi-Fi driver on the laptop or an incorrect password (or certificate). You have just cut the scope of the search in half!

Never change multiple things at once

You replace the cable, restart the Switch and set a static IP address. Suddenly it starts working! You’re a genius! But do you even know WHY it works? No. That means next time you won’t be able to resolve it in time or write documentation into the Knowledge Base.

Always test with a single changed variable.

1. Change cable -> Test -> Fails.
2. Restart switch -> Test -> Fails.
3. Delete certificates in Windows profile -> Test -> Works!

Problem found. Time for good coffee.