Entry-level troubleshooting

Ars Technica offers a very good, brief guide to troubleshooting computer hardware. Being two pages long, it doesn’t tell you everything, but includes some good tricks, including one I don’t always remember to tell people. To fully discharge a device, unplug it from the wall, remove the battery if it has one, then press and hold down the power button for 10-15 seconds. This discharges any power that could be lingering in the capacitors inside.

That’s not something you need to know every day, and on some hardware, you never need to do it. Then again, I’ve worked on some machines that need exactly that treatment. Early in my career, I worked in a shop that had a lot of IBM PS/2 Model 56s. Sometimes they would just refuse to power up, but if you unplugged them, threw the power button, waited a few seconds, turned off the power, and plugged them back in, they returned to normal.

In modern times, that problem seems to crop up most frequently after a power failure. I’ve brought many machines back from the dead after a power failure by giving them that treatment. The power button portion isn’t always necessary, but it never hurts. It’s easier to make a habit of doing it than it is to try to remember it in the heat of battle.

A variant of that trick works when resetting a device to its factory defaults. Say you’re working on a switch or a router. Usually you can reset it to factory defaults by holding down the reset button for a few seconds. Doing that, then unplugging the device and repeating the procedure without power ensures a full reset.

I also strongly agree with one piece of advice in the article. When calling support, use the vendor’s support tools, whatever they are, and have any error messages handy when you pick up the phone. It can save you 30-60 minutes of first-level troubleshooting. You know first-level troubleshooting. Have you rebooted? Is your antivirus software up to date? Did you put on clean underwear this morning?

This is the reason I’ve always loved working on HP servers. These days, most servers have a health status light. But when an HP server’s health light illuminates, you can pop open the case, and there’ll be an LED next to the failing (or failed) component. Note the number next to that LED, call HP support, and it’s literally a matter of telling them the machine’s serial number and the number next to the LED. The whole process normally takes less than 15 minutes. That’s why HP servers cost more than Dell servers. But the ease of troubleshooting could be the difference between five-nines (99.999% availability, or 5 minutes downtime per year) and four-nines uptime (99.99% availability, or 52 minutes downtime per year.)

%d bloggers like this: