A lightweight cron-driven probe installed on telep-mainframe to characterize the suspected fixed-wireless peak-hour degradation on the Telephely uplink (the TP-Link NE200 outdoor unit behind the double-NAT). Installed the same night as the IPv6 fix, but tracks a different concern — actual link quality over time, not the v6 misconfiguration.

For Agents

What: /home/levander/net-monitor/probe.sh, run every 5 min by the user crontab, appends to /home/levander/net-monitor/netlog.csv. Pings 1.1.1.1 every run; runs a 20 MB throughput + bufferbloat sample only near the top of each hour (~480 MB/day). Review: column -s, -t netlog.csv | tail. Remove: crontab -l | grep -v net-monitor/probe.sh | crontab -.

Purpose

Fixed-wireless links commonly degrade at peak hours (shared sector congestion). The IPv6 investigation confirmed raw bandwidth was healthy at the moment of testing (300/38–48 Mbit, 18–20 ms, 0% loss), but the user reports intermittent slowness. This logger captures a continuous record so a real peak-hour dip can be seen (or ruled out) rather than guessed.

How it works

  • Every 5 min: ping 1.1.1.1, record RTT + loss. Cheap, so it runs continuously.
  • Near the top of each hour only: a 20 MB throughput sample + a loaded-latency (bufferbloat) measurement. Capped to ~24 samples/day → ~480 MB/day of test traffic, negligible on a 300 Mbit link.

CSV schema (netlog.csv)

timestamp_iso,rtt_avg_ms,rtt_max_ms,loss_pct,dl_mbps,loaded_rtt_ms
  • dl_mbps / loaded_rtt_ms are populated only on the hourly heavy sample; the 5-min ping rows leave them empty.

Operating it

column -s, -t /home/levander/net-monitor/netlog.csv | tail    # review recent rows
 
crontab -l | grep -v net-monitor/probe.sh | crontab -         # remove the cron job

This lives on telep-mainframe — it only records while the box is up

The mainframe has an unstable-mains history (see 2026-07-23-mains-power-shutdowns); gaps in netlog.csv may be power outages, not link outages. Cross-check against boot times if a gap looks suspicious.