Client visibility

Diagnosing Intermittent CPE Drops ("It Drops Every Afternoon")

Every WISP support queue has this ticket: the connection is fine most of the day, then drops or crawls at a specific time. It's rarely random — the fix is finding the pattern.

Try it in the live demo →

Why this ticket is hard

Device-level monitoring tells you the AP is up. It doesn't tell you which of the 20–40 CPEs riding that AP is actually having a bad time, or when. By the time a tech rolls a truck, the bad window has often passed and everything looks normal — the classic “works when the tech is here” problem.

Step 1 — check the client roster, not just the AP

Open the sector AP's drawer and look at its client roster: every CPE it serves, with signal, CCQ and retry rate, sorted worst-first. If one client sits well below its neighbors on the same AP — same distance, same antenna, much worse CCQ — the problem is local to that client (alignment, cable, a failing radio), not the sector.

Step 2 — check the roaming/reconnect history

A per-client roaming/reconnect log tells you exactly when the client dropped and reconnected. Cluster the timestamps — all 4–6pm? All after a specific date? That timing is the single most useful diagnostic signal you have, because it rules entire categories of cause in or out.

Step 3 — cross-check against the spectrum view

Afternoon-specific drops often correlate with time-of-day interference: a neighbor's network ramping up after school or work hours, or thermal drift on a radio that's marginal to begin with. Pull up the spectrum tab for the site and see whether noise floor or channel occupancy on that AP's frequency rises around the same window.

Step 4 — rule out maintenance noise

If the site has had recent planned work, make sure a maintenance window was actually used to suppress alerts during it — otherwise a string of expected blips during a scheduled change can get mixed in with the real pattern you're trying to isolate.

Common root causes, by pattern

Same time every day, weekdays only — likely interference from a nearby network with a human usage pattern (home or office Wi-Fi).
Hottest part of the day — thermal throttling or a radio nearing end-of-life; check CPU/temperature trend if your platform retains it.
One client, all other CPEs fine — local issue: alignment drift, a damaged cable, a firmware mismatch versus the rest of the fleet.
Whole sector, all clients, sudden onset — a new co-channel conflict (another AP added nearby) or hardware fault on the AP itself.