Repository navigation
System
The SYSTEM tab contains maintenance tools, component updates, access control, and backup management.
The tab is split into two columns: left — actions and maintenance (Quick Commands, Maintenance, System), right — settings (Chat Settings, Access Control, Telegram, Backup & Restore, Versions).
Run commands without SSH:
| Command | Description |
|---|---|
| PORTS | List connected USB devices (lsusb, ls /dev/ttyUSB*) |
| RNSTATUS | Status of all Reticulum interfaces |
| RESTART RNSD | Restart Reticulum |
| RESTART LXMF | Restart LXMF Bridge |
| MEMORY | RAM usage |
| UPTIME | System uptime |
| CLEAN LOGS | Clear system logs |
| ⬆ UPDATE NOEMA | Update the gateway from GitHub and restart services |
The ⬆ UPDATE NOEMA button does:
-
git fetch+git reset --hard origin/main— brings the local copy exactly in line with the repository - Updates Python dependencies
- Restarts the dashboard, noema_lxmf_bridge, and rnsd services
If an orange ⬆ UPDATE badge appears in the dashboard header, a new version is available. Click it to trigger the update.
⚠️ Any unsaved local changes on the device will be overwritten by the repository version.
If the version is old and the ⬆ Update NOEMA button isn't there yet (or the dashboard won't open):
cd /root/NOEMA-RNSGate-Lite && git fetch origin && git reset --hard origin/main
systemctl restart dashboard noema_lxmf_bridge rnsdThis is exactly what the Update NOEMA button does in the dashboard — brings the local copy in line with the repository and restarts the core services.
⚠️ Don't edit project files manually over SSH (install.sh,app.py,index.html, etc.) — git doesn't track such edits as a commit, and the next auto-update will fail withYour local changes... would be overwritten by merge, aborting the update. If you edited something manually for testing, either revert it (git checkout -- filename) or accept that it'll be overwritten by the command above on the next update.
| Setting | Description |
|---|---|
| Restart rnsd every night at 04:00 | Automatic nightly Reticulum restart via cron — frees up accumulated memory (the routing table grows over time on active networks) |
| TCP watchdog | Checks TCP interface status (rnstatus) every 5 minutes and automatically restarts rnsd if a dropped TCP interface is detected — useful on unstable internet or during power outages |
Both toggles control the system cron directly. State is pulled from the real crontab every time you open the tab.
TCP watchdog has a 10-minute cooldown between restarts — it won't restart-storm during a long outage, it just waits for the next check. Log: /var/log/noema_tcp_watchdog.log.
ℹ️ The nightly restart and the TCP watchdog solve different problems: the first frees memory from a bloated routing table on a schedule, the second reacts to an actual dropped connection. Recommended to enable both.
| Action | Description |
|---|---|
| Restart Dashboard | Restarts only the web app (with a background delay — the button responds before the process restarts) |
| Reboot Device | Reboots the whole device |
| Reset All Identity | Resets all cryptographic keys |
Deletes all of the gateway's identifiers. Use this before selling or handing off the device.
What gets deleted:
- LXMF keys (bridge and chat)
- Reticulum identity
- I2P keys (router + tunnel)
- Nomadnet identity
- Chat history and contacts
What's kept:
- MQTT and Reticulum configs
- Nomadnet pages (.mu files)
- Scripts
⚠️ Irreversible. After a reset, the device gets new addresses. Take a Backup beforehand.
ℹ️ Since v1.4.5, Reset All Identity correctly resets all addresses, including LXMF Bridge (previously there was a bug: the bridge identity is stored in
lxmf-tools/lxmfy_data/lxmfy_config, not the old path, so the Bridge address wasn't changing). Update your dashboard to the latest version.
| Setting | Description |
|---|---|
| Max messages per contact | Maximum messages kept in history per contact |
| Max files MB | Maximum total size of attachments |
| Notification sound | New message sound (Default, Double beep, High, Low, Bell, Off) |
| Volume | Notification volume 0–100% |
Buttons:
- Save limits — save the limits
- Clean files now — delete files over the limit
- Clear all history — clear all chat history and files
Protection against unwanted messages to the gateway's addresses. Two independent whitelists — for the LXMF Bridge and for the built-in NOEMA Chat. Both paths are additionally protected against duplicate messages.
Restricts who can send messages to the LXMF Bridge address (the one that forwards to MQTT → Home Assistant).
| Field | Description |
|---|---|
| Whitelist toggle | When on — only messages from listed addresses are accepted, everything else is silently dropped |
| Allowed LXMF addresses | A list of hex addresses (32 characters), add/remove via buttons |
| Rate limit | Max messages per minute from a single sender (default 5) |
| Cooldown | Sender blocked when the limit is exceeded, in seconds (default 60) |
Changes are saved to /etc/noema/bridge.cfg (section [access]) and automatically restart noema_lxmf_bridge to apply.
⚠️ Don't add the gateway's own address to the whitelist — it's shown in the dashboard header and in the bridge logs at startup ([NOEMA] Address: ...). The whitelist should contain sender addresses (your phone, trusted contacts), not the recipient's.
Same principle, but for incoming messages to the dashboard's NOEMA Chat (doesn't affect Bridge/MQTT).
| Field | Description |
|---|---|
| Whitelist toggle | When on — only messages from listed addresses get saved to chat |
| Allowed LXMF addresses | Separate list from the Bridge whitelist |
Saved to /etc/noema/bridge.cfg (section [chat_access]), applies immediately, no dashboard restart needed.
Works automatically, no separate settings in the dashboard. With multiple TCP interfaces to the same contact (rmap.world, relay nodes, etc.), an LXMF message can arrive more than once over different routes at the same time. Both Bridge and Chat remember each message's unique ID for 1 hour and silently drop repeats — prevents double-triggering Home Assistant automations and duplicate entries in chat history.
- CrossTalk: Settings → LXMF Address (bottom of the chat list)
- Sideband: identity icon in the app header
-
NomadNet:
!icommand in the console, or in identity settings
The LXMF Bridge can answer questions using a local LLM (llama.cpp or an Ollama-compatible server) right in the chat — fully offline, no internet required.
Add a [llm] section to /etc/noema/bridge.cfg:
[llm]
enabled = 1
url = http://YOUR_LLM_SERVER:8090/completion
max_tokens = 256
timeout = 60
system_prompt = You are a helpful assistant reachable over a Reticulum mesh. Keep answers short.| Parameter | Description |
|---|---|
enabled |
1 enables the /ask command, defaults to 0 (off) |
url |
Your llama.cpp server's (llama-server) /completion endpoint address |
max_tokens |
Maximum length of the model's reply |
timeout |
Reply wait timeout, in seconds |
system_prompt |
System prompt that shapes the model's behavior |
After editing the config, restart the bridge:
systemctl restart noema_lxmf_bridgeSend /ask <question> to the LXMF Bridge address from any LXMF client (Sideband, CrossTalk). The bot immediately replies "🤖 Thinking…", then sends the model's answer as a separate message titled "AI" a few seconds later.
The query to the model runs in a separate thread — it doesn't block processing of other LXMF messages while the answer is being generated.
ℹ️ Requires an LLM server (llama.cpp/llama-server with an OpenAI-compatible
/completionAPI) already running and reachable on the network. The gateway itself doesn't run or store the model.
The gateway can send Telegram notifications when it receives incoming messages.
- Create a bot via @BotFather → get a Bot Token
- Find your Chat ID via @userinfobot
- Enter the token and chat_id in the fields → click Save
- Click Test — you should receive a test message
| Checkbox | Description |
|---|---|
| LXMF Bridge messages | Notifications for messages to the LXMF Bridge (bot) address |
| Chat messages | Notifications for messages in the gateway's personal chat (NOEMA Chat) |
| Category | Files |
|---|---|
| configs | reticulum.cfg, noema_bridge.cfg (including access/chat_access/llm settings), nomadnet.cfg, i2p tunnels |
| identity | LXMF Bridge identity (lxmfy_data/lxmfy_config), reticulum identity, i2p router keys, tunnel keys, rns destinations, chat lxmf identity |
| pages | All Nomadnet .mu files |
| scripts | All .py and .sh scripts |
| data | chat_history, chat_contacts, chat_files, monitored_nodes |
Click DOWNLOAD BACKUP — downloads a noema_backup_YYYYMMDD_HHMMSS.zip archive.
- Click RESTORE FROM BACKUP
- Select the
.ziparchive - Click UPLOAD
- Reboot the device
After restoring, all gateway addresses (including LXMF Bridge), chat history, whitelists, pages, and scripts return to their original state.
✅ Backup preserves all cryptographic keys, including the LXMF Bridge identity — gateway addresses stay the same after restoring.
ℹ️ Since v1.4.5, Backup correctly saves and Restore correctly restores the LXMF Bridge identity (previously there was a bug: the backup pointed to a nonexistent old path, so the Bridge address wasn't being restored). Backups made before this fix don't contain the bridge identity — take a new backup after updating.
Table of installed and latest component versions:
| Component | Description |
|---|---|
| Python | Interpreter version |
| RNS | Reticulum Network Stack |
| LXMF | Messaging protocol |
| lxmfy | LXMF bot framework |
| nomadnet | Nomadnet node |
| paho-mqtt | MQTT client |
| Flask | Dashboard web framework |
If an update is available — an UPDATE button appears. If a component is missing — INSTALL. Updates run without SSH.
| Problem | Solution |
|---|---|
| LoRa not detected | Quick Commands → PORTS, check the USB cable |
| MQTT BROKER: ERR | Check the broker IP in Configs, login, password |
| Service crashed | Logs → pick the service, or journalctl -u service_name -n 50
|
| No route to contact | Quick Commands → RESTART RNSD, wait 30–60 sec |
| Dashboard won't open | systemctl restart dashboard |
| I2P unstable | Normal for the first 5–10 minutes after startup |
| Forgot the gateway's IP | Check your router's client list, or ping noema.local
|
| Telegram not sending | Check the token and chat_id, click Test in settings |
| No update notification | The gateway checks GitHub every 5 minutes |
| Whitelist enabled, messages don't get through | Check that the list has your sender address, not the gateway's; verify the hex character by character |
| Memory usage grows over time | Enable Maintenance → nightly rnsd restart, or check journalctl -u rnsd for the routing table |
| TCP interface stuck after an internet outage | Enable Maintenance → TCP watchdog; check manually with rnstatus, log: /var/log/noema_tcp_watchdog.log
|
| Access Control settings not showing in the UI | Refresh the page (Ctrl+Shift+R), usually a browser cache from an older dashboard version |
| LXMF Bridge address didn't change after Reset Identity | Update the dashboard to v1.4.5+ — this known bug was fixed |
| LXMF Bridge address doesn't match the backup after Restore | Make sure the backup was made on v1.4.5+; older backups don't contain the bridge identity |
/ask doesn't respond / LLM error |
Check that enabled=1 under [llm], and the LLM server is reachable at the configured url: curl SERVER_URL/health
|
Update fails with local changes would be overwritten by merge
|
A file on the device was edited manually over SSH and no longer matches the repository. Fix: cd /root/NOEMA-RNSGate-Lite && git fetch origin && git reset --hard origin/main, then systemctl restart dashboard noema_lxmf_bridge rnsd. Manual edits will be lost — save them elsewhere first if they matter |