glances

A modern, feature-rich system monitor written in Rust. Provides both a TUI (terminal) and a web frontend. Inspired by glances with GPU support, Docker integration, port monitoring, and a built-in API tester.

Install

# Full installation (GPU + Docker + Web)
cargo install glances --features full

# Default installation (GPU + Docker, no web)
cargo install glances

# Minimal installation (no GPU/Docker/Web dependencies)
cargo install glances --no-default-features --features minimal

Feature Flags

System Dependencies

Linux (GPU monitoring):

# NVIDIA drivers required for NVML
ls /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1

Linux (building from source):

# Ubuntu/Debian
sudo apt install pkg-config libssl-dev

# Fedora
sudo dnf install pkg-config openssl-devel

macOS: Metal GPU support works out of the box on Apple Silicon and AMD GPUs.

Features

TUI Mode (default)

A full terminal UI with real-time monitoring, graphs, and interactive process management.

System Monitoring

• CPU -- global usage, per-core bars, user/sys/idle breakdown, load averages (1m/5m/15m)
• Memory -- RAM and swap with total/used/free, gradient bars
• Disk -- df-h style display: filesystem, size, used, avail, use%, mount points (grouped by device)
• Network -- all interfaces with rx/tx rates, throughput graphs
• Battery -- charge level and state in header
• Sensors -- temperature readings with color-coded thresholds
• Power -- Intel RAPL power draw per domain (package, core, uncore, dram)
• Processes -- sortable by CPU, memory, disk I/O, with kill support and filtering

GPU Monitoring

NVIDIA GPUs (Linux/Windows via NVML):

• Multi-GPU utilization, temperature, fan speed, power, clocks, P-states
• VRAM usage, encoder/decoder utilization, PCIe throughput
• Per-process GPU memory tracking

Apple Silicon GPUs (macOS via Metal):

• GPU utilization via IOKit, memory usage and allocation
• Metal API version detection
• GPU process listing via IOKit

Docker Integration

• Container list with CPU%, MEM/MAX, ports, uptime, compose labels
• Block I/O and network I/O per container
• Container logs viewer -- press l on Docker tab
• Built-in HTTP API tester -- press Enter on a container to open a Postman-like overlay (GET/POST/PUT/DELETE with headers, JSON body, pretty-printed responses)

Port Monitoring

• Lists all processes listening on TCP/TCP6 ports
• Shows port, protocol, PID, user, CPU%, memory, bind address, and full command
• Kill processes directly from the ports view

Alert System

• Auto-detects high CPU (>85%), memory (>75%), load (>1.0/core)
• Shows ongoing/resolved alerts with timestamps
• Displays top contributing processes for each alert

Graphs

• Braille sparkline charts for CPU, memory, network, and disk I/O
• Area-fill charts for CPU/memory and GPU utilization history
• 60-second rolling window

Web Mode

A browser-based dashboard with split-pane multi-server monitoring. Built with Vue 3 and ECharts.

# Start web server (requires 'full' or 'web' feature)
glances -w

# Custom port and bind address
glances -w --port 8080 --bind 127.0.0.1

# With custom TLS certificate
glances -w --tls-cert /path/to/cert.pem --tls-key /path/to/key.pem

Web Features

• Split-pane layout -- iTerm2-style recursive splitting for monitoring multiple servers
• Display modes -- full, normal, and minimal views per pane
• HTTPS -- auto-generated self-signed certificate or custom TLS cert/key
• PWA -- installable as a Chrome/Progressive Web App
• Real-time charts -- CPU, RAM, and swap history with server-side persistence
• Process management -- sort, filter, and kill processes from the browser
• REST API -- JSON endpoints for integration with other tools

API Endpoints

Usage

# Run TUI with defaults (1s refresh)
glances

# Custom refresh rate (ms)
glances -r 500

# Disable GPU/Docker monitoring
glances --no-gpu --no-docker

# Start with per-core CPU bars and graphs
glances --per-core

# Start in compact mode
glances -c

# Show all processes including idle
glances -a

# Run as web server
glances -w

# Web server with custom TLS
glances -w --tls-cert cert.pem --tls-key key.pem

# Debug GPU detection
glances --debug-gpu

CLI Options

Views

Keyboard Shortcuts

Docker View

Architecture

src/
├── main.rs              -- entry point, event loop, terminal setup
├── app.rs               -- app state, input handling, alerts, process kill
├── cli.rs               -- clap argument parser
├── types.rs             -- data structures (metrics, history, sort)
├── utils.rs             -- formatting helpers (colors, bars, duration)
├── web.rs               -- axum web server, REST API, HTTPS/TLS
├── frontend/
│   └── index.html       -- Vue 3 + ECharts web dashboard (single-file)
├── metrics/
│   ├── mod.rs           -- module exports, feature-gate stubs
│   ├── system.rs        -- CPU/mem/disk/net/battery/process collection
│   ├── gpu.rs           -- NVML (Linux/Win) + Metal (macOS) backends
│   ├── docker.rs        -- bollard Docker container stats
│   ├── ports.rs         -- /proc/net/tcp parser, port-to-process mapping
│   └── power.rs         -- Intel RAPL power monitoring
└── ui/
    ├── mod.rs           -- module exports
    ├── layout.rs        -- main layout coordinator, view routing
    ├── tabs.rs          -- tab bar renderer
    ├── header.rs        -- top bar (hostname, uptime, battery)
    ├── footer.rs        -- keybinding hints
    ├── system.rs        -- CPU/MEM/SWAP/LOAD bars, network compact, df-h disk
    ├── gpu.rs           -- GPU cards and details
    ├── graphs.rs        -- braille sparkline and area charts
    ├── processes.rs     -- CPU process table with scrollbar
    ├── ports.rs         -- port process table
    ├── docker.rs        -- Docker container table
    ├── temps.rs         -- temperature sensor panel
    ├── alerts.rs        -- alert events panel
    ├── http_dialog.rs   -- Postman-like API tester overlay
    ├── logs_dialog.rs   -- container logs viewer overlay
    └── dialogs.rs       -- help screen, kill confirm, status bar

Performance

• Differential refresh: fast-changing metrics (CPU, memory, processes, network) update every cycle; slow metrics (disks, temps, Docker, ports) update every 5th cycle
• Parallel collection: GPU, power, ports, and Docker metrics are collected in parallel using scoped threads
• Optimized release builds: LTO enabled, single codegen unit, panic=abort

Requirements

• Rust 1.70+
• Linux/Windows: NVIDIA drivers for GPU monitoring (optional)
• macOS: Metal-compatible GPU (Apple Silicon or AMD)
• Docker daemon running (optional, for container monitoring)

License

MIT -- see LICENSE for details.

Acknowledgments

• Originally forked from nvglances
• Inspired by glances
• Built with ratatui, sysinfo, bollard, axum