Self-Hosting Glances with Docker Compose

What Is Glances?

Glances is a cross-platform system monitoring tool that displays CPU, memory, disk, network, and process metrics in a single real-time dashboard. It replaces the need to juggle htop, iotop, nload, and docker stats by combining everything into one view. The web UI makes it accessible from any browser, and the built-in REST API enables integrations with Grafana, InfluxDB, and Prometheus.

Prerequisites

A Linux server (Ubuntu 22.04+ recommended)
Docker and Docker Compose installed (guide)
256 MB of free RAM (Glances is lightweight)
A domain name (optional, for remote access)

Docker Compose Configuration

Create a docker-compose.yml file:

services:
  glances:
    image: nicolargo/glances:4.5.2
    container_name: glances
    restart: unless-stopped
    pid: host
    environment:
      - GLANCES_OPT=-w
      - TZ=UTC
    ports:
      - "61208:61208"
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - /etc/os-release:/etc/os-release:ro

Key configuration choices:

Setting	Purpose
`pid: host`	Required — lets Glances see all host processes, not just container processes
`GLANCES_OPT=-w`	Starts the web UI server on port 61208
`docker.sock` mount	Enables Docker container monitoring (CPU, memory, network per container)
`/etc/os-release` mount	Shows the host OS name instead of Alpine Linux (the container’s OS)

Start the stack:

docker compose up -d

Initial Setup

Open http://your-server-ip:61208 in a browser
The dashboard loads immediately — no account creation or configuration wizard
The web UI auto-refreshes every 3 seconds by default

The dashboard shows:

System info — hostname, uptime, OS
CPU — per-core usage, load average, steal time
Memory — RAM and swap usage
Disk I/O — read/write rates per disk
Network — bandwidth per interface
File systems — mount point usage
Processes — sorted by CPU or memory, with PID and command
Docker containers — resource usage per container (if docker.sock is mounted)

Configuration

Enable Password Authentication

By default, the web UI has no authentication. Add password protection:

environment:
  - GLANCES_OPT=-w --password

On first start with --password, Glances generates a password and prints it to the logs:

docker compose logs glances | grep password

To set a specific password, create a password file:

# Create the password file
echo "your-password" > glances.pwd

# Mount it in the container

services:
  glances:
    image: nicolargo/glances:4.5.2
    container_name: glances
    restart: unless-stopped
    pid: host
    environment:
      - GLANCES_OPT=-w --password
      - TZ=UTC
    ports:
      - "61208:61208"
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - /etc/os-release:/etc/os-release:ro
      - ./glances.pwd:/root/.config/glances/glances.pwd

Full Network Monitoring

Bridge networking limits network interface visibility. For complete network stats:

services:
  glances:
    image: nicolargo/glances:4.5.2
    container_name: glances
    restart: unless-stopped
    pid: host
    network_mode: host
    environment:
      - GLANCES_OPT=-w
      - TZ=UTC
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - /etc/os-release:/etc/os-release:ro

With network_mode: host, remove the ports: section — port 61208 is exposed directly on the host.

Nvidia GPU Monitoring

services:
  glances:
    image: nicolargo/glances:4.5.2-full
    container_name: glances
    restart: unless-stopped
    pid: host
    environment:
      - GLANCES_OPT=-w
      - TZ=UTC
    ports:
      - "61208:61208"
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - /etc/os-release:/etc/os-release:ro
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

Use the -full image tag — the minimal image doesn’t include GPU monitoring dependencies.

Custom Configuration File

Create glances.conf for advanced settings:

[global]
refresh=5
check_update=false

[outputs]
curse_theme=black

[network]
hide=lo,docker.*

[diskio]
hide=loop.*

[containers]
show=true

[connections]
disable=true

Mount it in the container:

volumes:
  - ./glances.conf:/etc/glances/glances.conf:ro

Export to Prometheus

Glances can expose metrics in Prometheus format:

environment:
  - GLANCES_OPT=-w --export prometheus
ports:
  - "61208:61208"
  - "9091:9091"   # Prometheus metrics endpoint

Add to your prometheus.yml:

scrape_configs:
  - job_name: 'glances'
    static_configs:
      - targets: ['glances:9091']

REST API

Glances exposes a full REST API at http://your-server:61208/api/4/:

Endpoint	Returns
`/api/4/cpu`	CPU usage details
`/api/4/mem`	Memory statistics
`/api/4/diskio`	Disk I/O rates
`/api/4/network`	Network interface stats
`/api/4/docker`	Container resource usage
`/api/4/processlist`	All running processes
`/api/4/all`	Everything in one call

Reverse Proxy

For Nginx Proxy Manager, create a proxy host pointing to http://glances:61208.

For Caddy:

glances.example.com {
    reverse_proxy glances:61208
}

See Reverse Proxy Setup for detailed configuration.

Backup

Glances is stateless — it reads live system metrics and stores nothing persistently. Back up only your docker-compose.yml and glances.conf configuration files.

See Backup Strategy for a comprehensive approach.

Troubleshooting

Only Container Processes Visible

Symptom: Process list shows only Glances itself, not host processes. Fix: Add pid: host to your Docker Compose configuration. This shares the host’s PID namespace with the container.

Network Interfaces Missing

Symptom: Only eth0 shows in network stats, not host interfaces like enp0s3. Fix: Switch to network_mode: host. Bridge networking limits interface visibility.

Docker Containers Not Showing

Symptom: No container section in the dashboard. Fix: Mount the Docker socket: /var/run/docker.sock:/var/run/docker.sock:ro. Ensure the container user has permissions to read the socket.

High CPU Usage

Symptom: Glances itself uses 10-20% CPU. Fix: Increase the refresh interval in glances.conf: refresh=10 (default is 3 seconds). Disable unused plugins like connections.

Shows Alpine Linux Instead of Host OS

Symptom: System info shows “Alpine Linux” instead of your actual OS. Fix: Mount /etc/os-release:/etc/os-release:ro in the container.

Frequently Asked Questions

Does Glances store historical data or is it real-time only?

Glances is primarily a real-time monitoring tool. By default, it shows live system metrics with no history. However, you can enable data export to InfluxDB, Prometheus, Elasticsearch, or CSV files for historical storage. For built-in historical monitoring with graphs, look at Netdata or Beszel. Glances is best used as a live dashboard, not a time-series platform.

Can Glances monitor multiple servers from one dashboard?

Yes, using Glances’ client-server mode. Run Glances in server mode on each remote machine (glances -s), then connect from a central Glances client or use the web UI’s server browser. Alternatively, export metrics from each server to a shared InfluxDB + Grafana instance for a unified multi-server dashboard.

What is the difference between the minimal and full Docker images?

The minimal image (nicolargo/glances:latest) is ~150 MB and includes core system monitoring (CPU, RAM, disk, network, processes). The full image (nicolargo/glances:latest-full) is ~400 MB and adds plugins for Docker monitoring, GPU stats (NVIDIA), RAID, sensors, and export modules (InfluxDB, Prometheus, MQTT). Use the minimal image unless you need specific plugins from the full version.

Why does Glances show Alpine Linux instead of my host OS?

By default, the container reads its own /etc/os-release (Alpine Linux) instead of the host’s. Mount the host’s OS release file: -v /etc/os-release:/etc/os-release:ro in your Docker Compose volumes. This gives Glances accurate host OS information in the system info section.

Can Glances send alerts when CPU or memory exceeds a threshold?

Not natively in the Docker web UI. Glances has a built-in alert system for terminal mode (colored warnings based on thresholds), but the web UI does not send external notifications. For alerting, export Glances data to Prometheus and use Alertmanager, or pair Glances with Uptime Kuma for threshold-based notifications. Glances is best for visual monitoring, not automated alerting.

Resource Requirements

Resource	Usage
RAM	50-100 MB
CPU	1-3% (minimal image), 3-5% (full image)
Disk	~150 MB (image), no persistent storage

Glances is one of the lightest monitoring tools available. The minimal image is ~150 MB; the full image with all plugins is ~400 MB.

Verdict

Glances is the best single-container system monitor for quick deployment. It gives you everything htop, iotop, and docker stats provide, but in a web-accessible dashboard with an API. The trade-off is depth — it’s a real-time snapshot tool, not a time-series monitoring platform. There’s no historical data, no alerting, and no dashboarding beyond the built-in view.

Use Glances as your everyday “what’s happening on my server right now” tool. Pair it with Prometheus + Grafana for historical trends and alerting, or use Beszel if you want lightweight monitoring with history. For endpoint uptime checks, add Uptime Kuma.