The fiber no-link checklist I run before swapping parts

When a fiber link goes down in a data center, the temptation is to start swapping parts until something comes up. I have watched that approach burn an hour and teach you nothing. If you change three things at once and the link comes up, you don't know which change fixed it. If it stays down, you now have three new suspects.

So I work the physical layer in a fixed order, one variable at a time, and I write down the result of each step as I go. This is the checklist I run before I swap any part blindly.

1. Verify both ends

Before touching anything, confirm the source, the destination, and the port IDs against the ticket or run sheet. A surprising number of "no link" calls turn out to be a wrong port, a moved patch, or a ticket written against the wrong device. If the paperwork and the panel don't agree, stop and sort that out first.

2. Inspect and clean

Look at both connectors for dust, damage, or a bent ferrule, then clean them with the approved method and reseat until they click. Dust is the most common cause I see, and it is also the cheapest fix. Never look into a live fiber end face — the light is invisible and it can damage your eyes. Use a scope, not your eyes.

3. Check for physical damage

Before swapping anything, look at the run itself: crushed cable, a bend tighter than the cable allows, a door pinching a patch cord, a damaged panel port. If you find damage, that is your answer, and no amount of swapping optics will fix it.

4. Swap the patch cable

Use a known-good cable of the same type and length class. Same type matters — single-mode and multimode are not interchangeable, and mixing them will keep you chasing a problem that isn't there. One swap, one result, written down.

5. Swap the optic

Use a known-good, compatible optic, and only if you are authorized to do so. Check that the optic type matches on both ends and matches the cable type. Record the serial or port of the optic you removed so the next person isn't guessing.

6. Test the port

If you are authorized, test with a known-good device or a loopback test. This is the step that separates a cable problem from a port problem, and skipping it is how a bad port gets blamed on three good cables.

7. Check polarity

On duplex fiber, confirm transmit goes to receive on both sides. A polarity flip looks exactly like a dead link, and it survives every cable and optic swap because the swap doesn't change the pairing. Trace Tx to Rx end to end before you escalate.

8. Escalate with facts, not guesses

If the link is still down after all of that, escalate — but bring the list, not a shrug. What I send looks like this:

• Tested: both ends verified, connectors inspected and cleaned, known-good cable swapped, known-good optic swapped, port loopback tested, polarity traced.
• Ruled out: cable, optic, dirty connector, polarity.
• Still down: yes, since [time].
• What I need next: [for example, a second pair to test against, or a higher-level network check].

That note saves the next person from repeating every step you already did, and it gets you a faster answer than "link still down, tried everything."

The mistake that wastes the most time

Changing three things at once. One variable at a time, written down as you go — that is the whole discipline. It feels slower in the first ten minutes and it saves you the hour you would have spent un-swapping parts to figure out what actually fixed the link.

That is the checklist. It is not fancy, but it works, and it works the same on the first link of the shift as on the last one.