mirror of
https://github.com/kuhyx/testsAndMisc.git
synced 2026-11-11 13:47:47 +01:00
Krzysztof kuhy Rudnicki
836f4ef2de
Diagnosing "Steam downloads are slow" kept pointing at the ISP because every averaged measurement showed a steady ~170 Mbps. Sampling at 1s instead of 5s showed that plateau does not exist: the link alternates between ~450 Mbps bursts and multi-second stalls. Effective rate is burst x duty_cycle, and only the second factor was wrong. steam-download-duty.sh reports those two factors separately, which is what the Steam UI cannot do. line-speed-probe.sh separates burst from sustained throughput, because a single stream tops out near 180 Mbps on this line and reads convincingly as ISP shaping when it is not — a many-connection swarm reaches 986 Mbps with zero retransmits. Arithmetic is extracted into pure helpers with main() behind a BASH_SOURCE guard so the bats suite can source them without starting a measurement. install.sh installs bats and links both scripts onto PATH. Note: no CI workflow runs .sh/.bats tests, including the four shell tests that predate this one, so the suite is manual-only for now.
127 lines
4.5 KiB
Bash
Executable File
127 lines
4.5 KiB
Bash
Executable File
#!/bin/bash
|
|
|
|
# ============================================================================
|
|
# steam-download-duty.sh — measure a download's DUTY CYCLE, not its average.
|
|
#
|
|
# Why this exists: on this machine the link is not bandwidth-limited, but the
|
|
# Steam client is. Sampling at 1 s shows Steam alternating between
|
|
# multi-hundred-Mbps bursts and multi-second stalls in which it uses ~1% CPU,
|
|
# no disk, and transmits only keepalives. Any averaged "Mbps" figure is really
|
|
# burst_rate x duty_cycle
|
|
# and hides which of the two factors is actually wrong.
|
|
#
|
|
# Measured 2026-07-19 (see the diagnosis writeup): Steam client 38% duty /
|
|
# ~170 Mbps effective, vs depotdownloader -max-downloads 32 at 97% duty /
|
|
# 696 Mbps effective on the same line, same CDN, same minute.
|
|
#
|
|
# Use this to compare download clients or regions objectively — the Steam UI's
|
|
# averaged number cannot distinguish "slow link" from "idle client".
|
|
#
|
|
# Usage: steam-download-duty.sh [seconds] (default 60)
|
|
# ============================================================================
|
|
|
|
set -euo pipefail
|
|
|
|
IFACE="${IFACE:-enp6s0}"
|
|
IDLE_THRESHOLD_MBPS="${IDLE_THRESHOLD_MBPS:-20}" # below this = stalled second
|
|
|
|
# --- pure helpers (unit-tested by linux_configuration/tests) ----------------
|
|
|
|
# Convert a byte delta observed over N seconds into Mbps (integer).
|
|
bytes_to_mbps() {
|
|
local delta="$1" secs="$2"
|
|
(( secs > 0 )) || { printf '0'; return; }
|
|
printf '%d' $(( delta * 8 / secs / 1000000 ))
|
|
}
|
|
|
|
# Percentage of sampled seconds that carried traffic.
|
|
duty_pct() {
|
|
local active="$1" total="$2"
|
|
(( total > 0 )) || { printf '0'; return; }
|
|
printf '%d' $(( active * 100 / total ))
|
|
}
|
|
|
|
# Integer mean of the numeric arguments (0 when given none).
|
|
mean_of() {
|
|
(( $# > 0 )) || { printf '0'; return; }
|
|
local sum=0 v
|
|
for v in "$@"; do sum=$(( sum + v )); done
|
|
printf '%d' $(( sum / $# ))
|
|
}
|
|
|
|
# Classify a run. Kept separate from printing so it can be asserted on.
|
|
# LINK_LIMITED - transferring almost always; the rate is the rate
|
|
# CLIENT_IDLE - fast when active but idle a lot => client's fault
|
|
# MIXED - neither clean bursts nor high duty
|
|
classify() {
|
|
local duty="$1" burst="$2"
|
|
if (( duty >= 80 )); then
|
|
printf 'LINK_LIMITED'
|
|
elif (( burst >= 400 )); then
|
|
printf 'CLIENT_IDLE'
|
|
else
|
|
printf 'MIXED'
|
|
fi
|
|
}
|
|
|
|
# --- measurement -----------------------------------------------------------
|
|
|
|
rx_bytes() { cat "/sys/class/net/${IFACE}/statistics/rx_bytes"; }
|
|
|
|
main() {
|
|
local total="${1:-60}"
|
|
local prev cur mbps active=0 peak=0
|
|
local -a samples=() active_samples=() trace=()
|
|
|
|
printf 'Sampling %ss at 1s resolution on %s ...\n\n' "$total" "$IFACE"
|
|
prev=$(rx_bytes)
|
|
|
|
local i
|
|
for (( i = 1; i <= total; i++ )); do
|
|
sleep 1
|
|
cur=$(rx_bytes)
|
|
mbps=$(bytes_to_mbps $(( cur - prev )) 1)
|
|
prev=$cur
|
|
samples+=("$mbps")
|
|
if (( mbps > peak )); then peak=$mbps; fi
|
|
|
|
if (( mbps >= IDLE_THRESHOLD_MBPS )); then
|
|
active=$(( active + 1 )); active_samples+=("$mbps"); trace+=("#")
|
|
else
|
|
trace+=(".")
|
|
fi
|
|
done
|
|
|
|
printf 'pattern (# = transferring, . = stalled):\n '
|
|
printf '%s' "${trace[@]}"
|
|
printf '\n\n'
|
|
|
|
local duty burst overall
|
|
duty=$(duty_pct "$active" "$total")
|
|
burst=$(mean_of ${active_samples[@]+"${active_samples[@]}"})
|
|
overall=$(mean_of ${samples[@]+"${samples[@]}"})
|
|
|
|
printf 'duty cycle : %d%% (%ds transferring / %ds stalled)\n' \
|
|
"$duty" "$active" "$(( total - active ))"
|
|
printf 'rate when active: %d Mbps (peak %d)\n' "$burst" "$peak"
|
|
printf 'effective avg : %d Mbps\n\n' "$overall"
|
|
|
|
case "$(classify "$duty" "$burst")" in
|
|
LINK_LIMITED)
|
|
printf 'VERDICT: link is the limit — %d Mbps is close to what you get.\n' "$burst" ;;
|
|
CLIENT_IDLE)
|
|
printf 'VERDICT: NOT bandwidth-limited. Bursts reach %d Mbps but the\n' "$burst"
|
|
printf ' transfer is idle %d%% of the time. At 100%% duty this\n' "$(( 100 - duty ))"
|
|
printf ' link would give ~%d Mbps. Compare clients/regions by\n' "$burst"
|
|
printf ' duty cycle, not by the averaged UI number.\n' ;;
|
|
*)
|
|
printf 'VERDICT: mixed — bursts only reach %d Mbps and duty is %d%%.\n' \
|
|
"$burst" "$duty" ;;
|
|
esac
|
|
}
|
|
|
|
# Only run when executed directly, so tests can source the helpers above.
|
|
if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
|
|
main "$@"
|
|
fi
|