Current Events > One of the coolest things in network engineering, imo...

(+) Yeah! (+)

Is when you're troubleshooting a weird problem and your understanding of network theory is validated in the real world.

We have a branch location that has two portions of the building that used to be connected via fiber. The cable vendor, years ago, somehow filled the wrong pipe with concrete and broke the fiber cables between the two parts of the building. The copper ethernet cables that were pulled through that same pipe somehow still worked just fine. Until last night.

Our overnight support team was trying to help a field tech get the connection back up, but they couldn't figure it out. I'm on call, so I got paged to help.

In an ethernet cable, there are 8 copper wires twisted into 4 "pairs." Pairs required for speed/duplex levels:

• 10 Mbps half duplex: 1st pair only
• 100 Mbps full duplex: 1st two pairs.
• 1000 Mbps full duplex: all 4 pairs.

I started out by doing a cable test. Result:

• Pair 1: OK
• Pair 2: OK
• Pair 3: Short
• Pair 4: OK

Based on this, I figured the best we could do is 100Mbps full duplex. The said they tried multiple ports, but the link just wouldn't come up. I logged into the switch in the main building and hardcoded the port for 100 Mbps full. The link immediately came up. I could then get to the switch in the other building.

The switch in the other building saw the link as 10 Mbps half duplex, though. I did a cable test there and the result was exactly the same: A short on the 3rd pair. You normally have to have the same port speed/duplex settings on both sides of a link. There was a risk, though, that if it didn't work, I'd lose access to the switch again. The overnight support team was already connected to the field tech's laptop through the console cable and a cellular hotspot. I walked them through how to hardcode 100/full on the far-end switch.

Boom. 100 Mbps/full on both sides and connectivity is now normal to the 2nd building. The field tech thought he was gonna be out there for hours when the link wouldn't come up, but I figured it out pretty quickly.

Another feather in the cap for CableZL.

CableZL posted...

I started out by doing a cable test

"Yeah, you're CableZL, successful test"

enjoyed reading
thanks for sharing

I was speaking with one of my network engineers on the phone last year, a site in Alberta had either a heavy thunderstorm or a blackout and the power shorted. I think it was Rogers that went out and did work on one of the lines and didn't think to contact anyone from my company about the fact that none of the lines had a surge protector in place in case of power loss. Something that in our field should be common sense. Luckily we only lost 2 Cisco devices, but it shouldn't have happened in the first place.