LeSysBot docs

Use it

Dashboards

Graphs of your machine over time in Grafana — and how to share them.

Graphs of your machine over time — CPU, memory, disk, network, temperatures and GPU — in Grafana at:

http://localhost:3000

The installer set this up. Log in with the username and password you chose during setup (admin by default; an unattended install generated one and saved it in ~/.lesysbot/grafana.env). Your dashboards are in the LeSysBot folder.

Start and stop#

bash
lesysbot dashboard start
lesysbot dashboard stop

It needs Docker. Everything listens on 127.0.0.1 only, and nothing needs sudo.

Don't have Docker yet?

Install Docker Engine for your distribution, then let your user run it without sudo:

bash
sudo usermod -aG docker $USER      # then log out and back in
docker compose version             # should print v2.x
lesysbot dashboard start

What's included#

DashboardShows
System OverviewCPU, memory, disk, network per interface, temperatures, NVIDIA GPU
Network TrafficThroughput per network interface
GPU DetailNVIDIA utilization, memory, temperature and power

Panels your hardware can't fill are left out, so an empty panel means something is actually wrong. A dashboard whose data isn't available — GPU Detail on a machine without an NVIDIA driver, say — isn't shown at all. lesysbot dashboard list tells you why.

Share a snapshot#

Ask the bot:

You: share me the dashboard for a day
Bot: 📊 Dashboard shared — expires in 1d:
     https://snapshots.raintank.io/dashboard/snapshot/…

Anyone with the link can see a copy of your graphs. Links can last from an hour to a month. Say "list my shared dashboards" to see them, and "delete snapshot 1" to take one down (a cached copy may linger for up to an hour).

Add more dashboards#

bash
lesysbot search --kind dashboard     # see what's available
lesysbot install owner/repo          # install one from GitHub
lesysbot dashboard render            # show it in Grafana

Grafana picks it up within 30 seconds. The Dashboards tab in the control panel does the same.

Change or write one#

Installed dashboards live in ~/.lesysbot/dashboard/installed/<name>/. Edit the files there, then run lesysbot dashboard render. Changes made in Grafana's own editor are lost on the next render, so edit the files instead.

The simplest dashboard is a Grafana export: click Export in Grafana, save the JSON as dashboard.json in a folder with a README.md, and push it to GitHub. Anyone can then install it with lesysbot install.

Declare what a dashboard needs

Put the requirements in the README frontmatter. LeSysBot checks them against Prometheus before showing the dashboard:

yaml
---
name: postgres
kind: dashboard
description: PostgreSQL connections, cache hit rate, replication lag
version: "1.0.0"
prerequisites:
  - service: prometheus
  - metric: pg_up
---

For panels that depend on the hardware, use dashboard.py instead of JSON:

python
def build(host, caps, ctx):
    panels = [cpu_panel(), memory_panel()]
    if "nvidia" in caps:
        panels.append(nvidia_panel())
    return {"title": "My dashboard", "panels": panels}

Ports, .env settings and the stack's internals are in dashboard/README.md. Empty panels? See Troubleshooting.

Edit this page on GitHub