Think before you tools
Random plugging wastes hours. Professionals follow a ladder: (1) define the symptom precisely — no light, low light, errors, intermittent; (2) gather facts — what changed, when, which wavelengths/services affected; (3) consult the baseline records; (4) split the problem in half with the least invasive test; (5) fix, verify, and update the records.
The first three checks
Ninety percent of tickets die here: inspect and clean every connector in the path (both sides); verify transmit power at the source with a meter; swap in known-good patch cords. Only then escalate to OTDR shots and closure openings.
Symptom signatures
Loss at 1550/1625 but fine at 1310 → macrobend (look at recent work: ties, closures, slack loops). High reflectance event growing over time → damaged or contaminated connector. Total loss at a known distance → break; the distance from *both* ends brackets the closure. Intermittent errors that follow temperature → strained splice or connector at its limit — the OTDR at the right moment catches it.
Intermittents and the honest write-up
For intermittents, monitor: a power meter with min/max logging or an OTDR in comparison mode across hours. When closed, the ticket states root cause, evidence, fix, and post-fix measurements — "cleaned and it works" without inspection photos is a return visit scheduled.