Recurring T3 upstream ranging timeouts / packet loss since February 2026

Summary

I've had intermittent internet issues going back to around February 2026 and have been tracking them since summer with continuous network monitoring (OPNSense + modem-side DOCSIS telemetry) to try to figure out what's going on. Sharing what I've found in case someone here can help point this toward the right team.

What I'm seeing

• An original monitoring campaign in Feb–Mar 2026 showed packet loss to an external target (Quad9 DNS) on nearly every day observed.
• A follow-up 30-day campaign in Aug–Sep 2026 confirmed the same pattern is still happening, predominantly in a daytime/evening window (roughly 9am–9pm ET, though not exclusively).
• My modem's event log recorded at least 429 T3 upstream ranging time-outs (event 82000500) over that Aug–Sep window, in bursts on Aug 8, Aug 13, Aug 31, Sep 2, and Sep 5. There's a gap in my own logging around mid-to-late August, so the real number could be higher.
• Downstream signal itself looks healthy throughout. SNR roughly 39–43 dB, downstream power ~7 dBmV, upstream power roughly 44–47 dBmV, which points to the modem checking in and not getting a response rather than a noisy line or weak signal.
• I replaced my router's network card in May 2026, partly to see whether my own equipment was contributing, and the same pattern continued afterward, including a short outage this morning (Sep 21) that seemed to line up with a downstream profile reassignment logged by the modem at 11:01:56 AM ET, right before a ~2-minute full loss of connectivity.
• I've also tried to check whether it's local congestion (Wi-Fi, too many devices, etc.) — usage during the loss periods looks statistically indistinguishable from usage when there's no loss, so that doesn't seem to be it.

What I'm hoping for

• If possible, having someone check the ranging/registration history for my modem from the CMTS side. The signal readings on my end look clean to me, but I'd welcome a second opinion.
• A look at the tap/node for my address, since the timing pattern makes me wonder if it might be something upstream rather than local.

Happy to share full logs/timestamps if that's useful. Just trying to get this in front of someone who can look at it from the network side, since I'm not sure what else I can check from my end.