How to Escalate a Bad Optic Without Guessing
"Bad optic, please replace" is the escalation that bounces. Most no-link faults blame the transceiver, and most transceivers are innocent — the connector was dirty, the patch cable was kinked, or the port is dying. Engineers know this, so they approve fast only when the ticket proves the fault follows the optic. Here is the isolation order and the note that gets it done.
Rule out everything else first (one variable at a time)
- Verify the ticket. Device, rack/U, port, and the far end. Confirm you are testing the link that is actually down.
- Inspect, clean, reseat. Pull the fiber, inspect both faces, clean, reseat the optic and both connectors. Record link state and DOM readings after this alone — a large share of "bad optics" end here.
- Known-good patch cable. Swap in a tested cable of the same type and polarity. If link restores, the optic was never the problem.
- Suspect optic → known-good port. Move the suspect optic (with a known-good cable) to a port that is up. Still down or marginal DOM? The fault is following the optic.
- Known-good optic → original port. If a tested spare brings the original port up, the port and plant cabling are cleared and the case against the suspect optic is closed.
The numbers to capture before you escalate
- DOM readings: Tx and Rx power (dBm) on both ends, before and after each swap. A receiver at −28 dBm next to a healthy −6 dBm neighbor is evidence; "signal weak" is not.
- Error counters: CRC/FEC errors and link-flap count with timestamps — rising counters on a clean, known-good path point at the optic's laser or receiver.
- Identity: part number, serial number, vendor coding/EEPROM type, and the device/port it came out of. RMAs and spares audits live on the serial.
An escalation note that gets approved
15:10 — Link down: srv-1042 rack C12 U18 Eth1 ↔ sw-c12-04 port 1/1/18.
Both ends verified against ticket. Inspect/clean/reseat fiber + optic: no change.
DOM: Rx −27.4 dBm (neighbor links −5 to −7 dBm). CRC errors rising, 14 flaps/hr.
Test 1: known-good patch cable — no change.
Test 2: suspect optic (PN QSFP-100G-SR4, SN XXXXXXX) in known-good port sw-c12-04
1/1/22 with known-good cable — link down, Rx −26.9 dBm. Fault follows optic.
Test 3: known-good spare optic in original port — link up, Rx −6.1 dBm, 0 errors
in 20 min. Port + cabling cleared.
Ask: approve replacement optic from spares; failed optic SN XXXXXXX to RMA bin?
Next update 16:00 or on approval.
The mistakes that get escalations bounced
- Swapping optic and cable together. Link comes up and you learned nothing — the good optic gets blamed next time the cable fails.
- No DOM numbers. Without readings, "bad optic" is an opinion the engineer has to re-test from zero.
- No far-end check. Half the link lives in another rack (or another building). State both ends or the escalation is incomplete.
- Failed optic back in the spares drawer, unlabeled. Tag it with the ticket number or someone reinstalls it on the next night shift.
This plugs into the full floor workflow in RackReady Core ($19): the 8-step fiber no-link flow, escalation and handover formats, and ticket templates. The free sample includes the universal break-fix checklist. Fault-isolation order for down links:
fiber no-link checklist; escalation format in depth:
how to write an escalation ticket.
See RackReady Free sample Ticket note examples
Support: support@hasherventures.com
Free tool: turn your checks into a structured ticket note with the RackReady Ticket Builder.