Home › Documentation › Configuration
Configuration
GNULTE has no system-wide configuration. Every tool ships with working defaults;
what you can tune is saved per user, under ~/.config/gnulte-go/.
The settings editor
Most tools accept --settings, which opens a small full-screen editor
for that tool's saved defaults (alarm thresholds, update intervals, history depth,
sounds, report behaviour) and exits. Settings are stored in
~/.config/gnulte-go/config.json and apply to every future run of that
tool; command-line flags always override them for a single run.
Editors inside the tools: press o in gnulte-lan and in
the gnulte-scan -T browser.
Impairment profiles
The shaping engine ships with twelve built-in profiles. List them with:
gnulte --list-profiles
| Profile | Latency | Jitter | Loss | Dup | Reorder | Cap |
|---|---|---|---|---|---|---|
gaming | 1500 ms | 300 ms | 2 % | 0 % | 0 % | — |
streaming | 2500 ms | 500 ms | 5 % | 0 % | 0 % | — |
voip | 3000 ms | 200 ms | 0 % | 0 % | 5 % | — |
web | 2000 ms | 400 ms | 3 % | 0 % | 0 % | — |
extreme | 5000 ms | 1000 ms | 10 % | 5 % | 5 % | — |
throttle | 500 ms | 200 ms | 0 % | 0 % | 0 % | 512 kbps |
satellite | 600 ms | 40 ms | 1 % | 0 % | 0 % | — |
cellular | 150 ms | 90 ms | 2 % | 1 % | 0 % | — |
dialup | 200 ms | 120 ms | 0 % | 0 % | 0 % | 56 kbps |
congested | 400 ms | 350 ms | 5 % | 3 % | 2 % | — |
bufferbloat | 60 ms | 200 ms | 0 % | 0 % | 0 % | — |
nightmare | 8000 ms | 2000 ms | 25 % | 10 % | 10 % | — |
Apply one with --profile NAME. Save your own from the current
command line with --profile save:NAME; user profiles live in
~/.config/gnulte-go/profiles.json and appear in
--list-profiles alongside the built-ins.
Scenario scripts
A scenario turns one run into a sequence of timed phases — the shape of a real
bad day rather than a constant. Pass it with --scenario FILE; the
dashboard names the phase currently in force.
A scenario file is JSON: a name, an optional loop flag, and an ordered list of
phases. Each phase has a duration in seconds and any of the usual impairment keys
(latency, jitter, loss, dup,
reorder, bandwidth), plus an optional
wobble or burst generator:
{
"name": "storm-drill",
"loop": false,
"phases": [
{ "name": "baseline", "duration": 60 },
{ "name": "congested", "duration": 120, "latency": 200, "jitter": 100, "loss": 3,
"wobble": { "type": "sine", "amplitude_ms": 80, "period_s": 20 } },
{ "name": "lossy", "duration": 60, "latency": 120, "loss": 1,
"burst": { "loss_pct": 20, "on_s": 2, "off_s": 5 } }
]
}
wobble modulates the phase's base latency — "sine"
swings it by amplitude_ms either side of the base,
"sawtooth" rises from the base to base + amplitude and snaps back.
burst overrides the phase's loss with loss_pct for
on_s seconds every on_s + off_s. Unknown JSON keys are
rejected, so a typo like "latancy" is an error rather than a silently
ignored field.
Per-target overrides
With several targets, one set of conditions is rarely what you want.
--per-target FILE maps a target IP to a partial impairment
override; every key you leave out is inherited from the command line:
{
"192.168.1.50": { "latency": 400, "loss": 10 },
"192.168.1.62": { "bandwidth": 256 }
}
Each override becomes its own tc/netem class, so the targets really
do get different networks. Keys: latency, jitter,
loss, dup, reorder, bandwidth
(milliseconds, percent and kbps as on the command line).
Other files
| File | Written by / read by |
|---|---|
devices.json | the shared device store — written by
gnulte-lan (x to save a watch) and
gnulte-scan, read by gnulte's target picker
(w) and by the scanner for cross-reference. |
watch-history.jsonl | the LAN-watch up/down timeline
(v16.13+); one JSON object per debounced transition. Controlled with
--history-file and --no-history. |
acceptance.json | the authorised-use acceptance record;
clear it with --reset-safety on any tool that asks for consent. |
Reports are written to ~/GNULTE Reports/ unless you name a file
(--report FILE) or suppress them (--no-report).