gdu — Pretty Fast Disk Usage Analyzer Written in Go

What it is

gdu is a disk usage analyzer written in Go, inspired by ncdu. It scans filesystems using a pool of goroutines for parallel I/O, then presents results in an interactive terminal UI where you can drill into directories, sort by size, and delete files directly.

gdu is for system administrators, DevOps engineers, and developers who need to find what is consuming disk space. It works on servers (via SSH) and local machines, with a static binary that requires no dependencies.

How it saves time or tokens

Traditional du -sh requires multiple runs to drill into directories. ncdu scans sequentially, which is slow on SSDs that can handle parallel reads. gdu's goroutine-based scanning takes advantage of SSD parallelism, completing scans faster on modern storage.

The non-interactive mode (gdu -n) outputs results as plain text, suitable for CI pipelines or scripts that monitor disk usage. Combined with JSON output (--output-file report.json), you can feed results into monitoring dashboards.

How to use

1. Install gdu:

# macOS
brew install -q gdu

# Linux (static binary)
curl -L https://github.com/dundee/gdu/releases/latest/download/gdu_linux_amd64.tgz | tar xz
sudo mv gdu_linux_amd64 /usr/local/bin/gdu

1. Launch the interactive TUI:

gdu

1. Scan a specific directory with options:

# Non-interactive, top 10 largest dirs
gdu -n /var

# Ignore paths and export JSON
gdu -i /var/log,/tmp --output-file report.json

# Show apparent size instead of disk usage
gdu -a /home

Example

Using gdu in a CI pipeline to check disk usage before deployment:

#!/bin/bash
# Check if /var/lib/docker exceeds 50GB
SIZE=$(gdu -n -p /var/lib/docker 2>/dev/null | head -1 | awk '{print $1}')
if [ "$SIZE" -gt 53687091200 ]; then
  echo 'Docker storage exceeds 50GB, running cleanup'
  docker system prune -af
fi

Or parse JSON output for monitoring:

gdu -n --output-file /tmp/disk-report.json /
jq '.[] | select(.size > 1073741824) | {name: .name, size_gb: (.size / 1073741824)}' /tmp/disk-report.json

Related on TokRepo

• DevOps AI tools -- infrastructure monitoring and management
• Automation tools -- scripted automation for server maintenance

Common pitfalls

• gdu's parallel scanning is optimized for SSDs. On spinning disks (HDD), parallel reads cause seek thrashing and can be slower than sequential scanning. Use gdu --no-cross or reduce parallelism for HDDs.
• The interactive TUI requires a terminal. When running over SSH with limited terminal capabilities, ensure your TERM variable is set correctly (e.g., xterm-256color).
• gdu counts hard links separately by default. If your filesystem has many hard links (e.g., Nix store), apparent sizes may differ from actual disk usage. Use the -l flag to count hard links.