ArgusSyS — lightweight system stats dashboard

📸 Screenshot

ArgusSyS is a small Node.js service that:

• serves a JSON stats endpoint at /stats
• serves a static web UI from /
• optionally runs and schedules speed tests
• stores a small server-side history ring buffer (shared by all clients)

GPU support: the container image is intended for NVIDIA GPUs via nvidia-smi. If NVIDIA drivers / NVIDIA Container Toolkit are not available, the API will still run, but GPU fields will be null / n/a.

---

Requirements

• Docker Engine + Docker Compose
• (Optional, for GPU metrics) NVIDIA GPU + NVIDIA drivers on the host
• (Optional, for GPU metrics inside containers) NVIDIA Container Toolkit (so Docker can pass GPUs through)

If you run without NVIDIA support, everything except the GPU section should still work.

---

Quick start

Install (from GitHub)

TAG=$(curl -fsSL https://api.github.com/repos/G-grbz/argusSyS/releases/latest | grep -m1 '"tag_name"' | cut -d '"' -f4) && curl -fsSL -L "https://github.com/G-grbz/argusSyS/archive/refs/tags/${TAG}.zip" -o argus.zip && unzip -q argus.zip && cd argusSyS-*

docker compose up -d --build

Open

• UI: http://<host>:3012/
• API: http://<host>:3012/stats
• Health: http://<host>:3012/health

---

Docker Compose example

This is a generic template. By default it runs without NVIDIA. If you want NVIDIA GPU metrics, uncomment the gpus: line and the NVIDIA_* environment variables.

services:
  argussys:
    container_name: argussys
    build: .

    # If you need host-level visibility (e.g. /proc, /sys), you may enable these.
    # Remove them if you prefer a stricter container.
    privileged: true
    pid: host
    network_mode: host

    working_dir: /app
    command: ["node", "stats-api.js"]

    # NVIDIA (optional). If you have NVIDIA + NVIDIA Container Toolkit, uncomment these:
    # gpus: all

    environment:
      TZ: Europe/Istanbul
      PORT: "3012"

      # Comma-separated list of mountpoints the API should report (df).
      # IMPORTANT: if running in a container, each path here must be mounted into the container.
      DISK_PATHS: "/host"

      # UI
      UI_DIR: "/app/ui"
      UI_INDEX: "index.html"

      # NVIDIA runtime hints (optional)
      # NVIDIA_VISIBLE_DEVICES: "all"
      # NVIDIA_DRIVER_CAPABILITIES: "compute,utility,video"

      # Speedtest
      SPEEDTEST_TIMEOUT_MS: "120000"
      SPEEDTEST_INTERVAL_MIN: "0"          # 0 = disabled (manual only)
      SPEEDTEST_RUN_ON_START: "1"          # 1 = run once on startup
      SPEEDTEST_STATE_FILE: "/app/data/speedtest-state.json"

      # History
      HISTORY_MAX_MIN: "120"

    volumes:
      - ./:/app
      # Host visibility (optional)
      - /sys:/sys:ro
      - /:/host:ro

      # Persist app data
      - ./data:/app/data

      # Serve UI from repo
      - ./ui:/app/ui:ro

    restart: unless-stopped

Notes

• network_mode: host makes the UI/API available on the host network without port mapping. If you prefer bridge networking, remove network_mode: host and add ports: ["3012:3012"].
• If you disable privileged / pid: host, some host-level detection may be reduced.
• For DISK_PATHS, every listed path should be mounted in the container, otherwise the API may report it as “not mounted on host”.

---

Environment variables

Core

• PORT (default: 3012) — HTTP server port
• TZ — timezone used for formatting (recommended)

UI

• UI_DIR (default: ./ui) — directory containing static UI files
• UI_INDEX (default: index.html) — index file name

Disk reporting

• DISK_PATHS (default: /) — comma-separated mountpoints to report with df and I/O stats

GPU

• GPU_POLL_MS (default: 1000) — GPU polling period
• GPU_TIMEOUT_MS (default: 1000) — timeout for nvidia-smi calls

NVIDIA-only: GPU stats use nvidia-smi. On systems without NVIDIA, the GPU block will be missing / null.

History (server-side ring buffer)

• HISTORY_SAMPLE_MS (default: 1000) — sampling interval
• HISTORY_MAX_MIN (default: 120) — maximum history window
• HISTORY_DB_PATH (default: ./data/history_state.json) — persisted history file

Speedtest

• SPEEDTEST_TIMEOUT_MS (default: 120000) — per-run timeout
• SPEEDTEST_INTERVAL_MIN (default: 0) — schedule interval in minutes (0 disables scheduling)
• SPEEDTEST_RUN_ON_START (default: 0/1) — run once on boot
• SPEEDTEST_STATE_FILE (default: /app/data/speedtest-state.json) — persisted state file

---

API endpoints

• GET /stats — full stats payload (includes system, cpu, mem, net, gpu, disks, and history)
• GET /health — health check

Speedtest

• GET /stats/speedtest/last — last speedtest snapshot
• GET /stats/speedtest/run — trigger a speedtest run
• GET /stats/speedtest/config?interval=<minutes> — set interval (minutes)
• GET /stats/speedtest/history — 24h speedtest history

---

Troubleshooting

GPU shows n/a / missing

• Confirm nvidia-smi works on the host.

• If you want GPU metrics inside Docker:

  • install NVIDIA Container Toolkit on the host
  • ensure your Compose file includes gpus: all
  • keep NVIDIA_VISIBLE_DEVICES=all

If you are not using NVIDIA, remove the gpus: line and the NVIDIA_* environment variables.

Disks show “not mounted on host”

When running in a container, each path in DISK_PATHS must be a real mountpoint on the host and be mounted into the container at the same path.

---

Development

Run locally:

npm install
node stats-api.js

Then open http://localhost:3012/.

---

License

MIT