Troubleshooting

Start with these three commands:

lesysbot             # status: is the model reachable? is the service running?
lesysbot doctor      # what's missing on this machine, and how to fix it
lesysbot chat -v     # chat with the log on screen

Chatting#

"LLM unavailable"#

The model isn't reachable.

curl localhost:11434      # should print "Ollama is running"
ollama list               # is your model downloaded?

• Ollama not running → sudo systemctl start ollama, or ollama serve.
• Model missing → ollama pull <name>, or change llm.model in your config.
• Remote backend → check that llm.base_url ends in /v1 and the key is right.

/ commands keep working while the model is down.

It picks the wrong tool, or none#

Use a bigger model — see Choosing a model. Being specific helps too ("check the CPU temperature" beats "how is it"). Or run the tool yourself with /tool_name.

The first reply is slow#

The model is loading into memory. Later replies are faster.

My second message got no answer#

It's waiting for the first to finish — you'll get both, in order. If a confirmation is waiting for your answer, answer it first. / commands never wait.

It forgot what we were talking about#

It remembers the last 50 messages (agent.max_history), and /clear wipes them.

Tools#

A tool is missing from /help#

• Check it's in ~/.lesysbot/tools/ — lesysbot shows the folder in use.
• Files and folders starting with _ are skipped.
• An error in the file stops it loading. Check the log: tail -n 50 ~/.lesysbot/logs/lesysbot.log

"unavailable on this machine"#

'traceroute' is unavailable on this machine — requires 'traceroute' on PATH (not found).

Install the missing program (here, your distribution's traceroute package). It works straight away, no restart.

"is disabled"#

Turn it back on: lesysbot enable traceroute.

lesysbot install fails#

See Install tools.

Installing#

lesysbot: command not found#

Open a new terminal. If that doesn't help:

export PATH="$HOME/.local/bin:$PATH"

The service#

It starts, then stops#

See why:

journalctl --user -u lesysbot -n 50

Usual causes: a wrong Telegram/Discord token, or a missing dependency.

"Another LeSysBot instance … is already running"#

The service is already running — only one copy can run at a time. Use lesysbot chat to chat alongside it, or stop it first with systemctl --user stop lesysbot.

Config changes do nothing#

Restart the service: systemctl --user restart lesysbot. Turning tools on and off is the only change that applies without a restart.

The control panel says offline#

The service isn't running: systemctl --user start lesysbot. Or run the panel in your terminal with lesysbot manage.

"Control panel not started — port … already in use"#

Something else is using port 8700. Set another management.port in your config and restart the service. The bot keeps running either way.

I can't open the control panel from another computer#

That's on purpose — it only listens on 127.0.0.1. Use an SSH tunnel: ssh -L 8700:127.0.0.1:8700 you@server, then open http://127.0.0.1:8700 locally.

Telegram and Discord#

Dashboards#

Logs#

tail -f ~/.lesysbot/logs/lesysbot.log     # what the program did
tail -f ~/.lesysbot/logs/traces.jsonl     # what the model decided, per message

Tokens and keys are removed from both, so they're safe to share in a bug report. traces.jsonl does include what your tools returned — read it before sharing.

Start over#

mv ~/.lesysbot/config.yaml ~/.lesysbot/config.yaml.bak && lesysbot setup  # reset settings
rm ~/.lesysbot/tool_state.json                                             # turn every tool back on

To remove everything, see Uninstall.

Still stuck? Open an issue with the output of lesysbot and the last lines of ~/.lesysbot/logs/lesysbot.log.