DiePi-Software | Dagu: self-contained, zero-dependency workflow engine and DAG (Directed Acyclic Graph) scheduler with Web UI, written in Go

[lukasMega]

Creating a software request

Formal software information

• Software title | Dagu
• Short description | Dagu is a self-contained, zero-dependency workflow engine and DAG (Directed Acyclic Graph) scheduler with a built-in Web UI, written in Go
• Official URL | https://daguit.dev
• GitHub URL | https://github.com/dagu-org/dagu
• (Official) install documentation | https://docs.daguit.dev/getting-started/installation

Are there similar/alternative software titles available with DietPi-Software?

• None directly, though cron could be considered a primitive alternative for simple scheduling

Dagu vs Cron: Key Differences

Feature	Cron	Dagu
Task Dependencies	❌ Not supported	✅ Full support
Logging	❌ Basic (email only)	✅ Detailed per-run logs
Retry Logic	❌ No	✅ Built-in
Monitoring	❌ No visibility	✅ Web UI + CLI
Parallel Tasks	❌ No	✅ Yes
Error Alerts	❌ Limited	✅ Full control
Workflow Visualization	❌ No	✅ Web UI shows DAG
Configuration	Bash scripts	✅ Clean YAML

What makes your requested software better than the above solutions, if available?

• Dagu is a single statically compiled Go binary (~16 MB) with absolutely zero runtime dependencies — no Python, no database, no message broker, no Redis. Everything is stored as plain files on the local filesystem.
• Unlike cron, Dagu supports complex multi-step pipelines with conditional branching, retries, parallel execution, and step-level dependencies defined in simple YAML.
• Comes with a built-in responsive Web UI for visualizing DAG execution graphs, monitoring run history, and triggering workflows manually — no separate installation needed.
• Designed with resource-constrained environments as a first-class target. Idle RAM footprint is ~20–50 MB. Pre-built linux_arm64, linux_armv7, and linux_armv6 binaries are published for every release, covering all Raspberry Pi variants including the Pi Zero.
• Excellent fit for homelab automation: backup pipelines, data sync jobs, IoT event processing, scheduled API calls, and more.

How can DietPi make the installation easier or compatible, than following the install instructions or do APT installation, if available?

• The official installer script auto-detects architecture and downloads the correct binary, but it doesn't integrate with systemd, set up a dedicated service user, configure persistent data paths aligned with DietPi conventions (/mnt/dietpi_userdata/), or register with dietpi-services for unified start/stop/backup control.
• DietPi integration would give users a properly sandboxed, automatically updating, backup-aware installation in a single dietpi-software install command.

Can you provide the installation steps that you would suggest DietPi-Software to do?

1. Detect CPU architecture (uname -m) and download the matching pre-built binary from the latest GitHub release (e.g. dagu_linux_arm64.tar.gz) to /usr/local/bin/dagu. Make it executable.
2. Create a dedicated system user and group dagu (no login shell, no home directory) for service isolation.
3. Create the data directory at /mnt/dietpi_userdata/dagu/ (following DietPi conventions) with subdirectories dags/, logs/, and data/. Set ownership to dagu:dagu.
4. Write a minimal default config to /etc/dagu/config.yaml pointing dags, logDir, and dataDir to the paths above, with host: 0.0.0.0, port: 18080, histRetentionDays: 30, and maxOutputSize: 2097152 (2 MB, conservative for low-RAM devices).
5. Create a systemd service unit /etc/systemd/system/dagu.service running dagu start-all as user dagu, with Restart=on-failure, NoNewPrivileges=true, and ProtectSystem=full for hardening.
6. Enable and start the service via systemctl enable --now dagu.
7. Register the service with dietpi-services so it appears in the DietPi services management menu.
8. Print the Web UI URL (http://<device-ip>:18080) to the post-install message.

Are you willing to help maintaining the software installation, e.g. in case of needed setup changes due to updates etc.?

• Yes. Dagu publishes regular releases on GitHub with consistent binary naming conventions, so version update logic should be straightforward to maintain. Happy to test new releases on Pi Zero 2W hardware and submit PRs if the installation script needs adjustments.

---

Screenshots / GIF

Dashboard	DAG Definitions
DAG Details	Execution Details
Execution History	Execution Log
DAG Editor	Workers

Video:

• https://www.youtube.com/watch?v=xHc8dgMQ0nE

---

Update:

do not forget to update, check:

• https://github.com/MichaIng/DietPi/wiki/DietPi-TCP-UDP-port-usage-list
• https://docs.daguit.dev/configurations/server

[lukasMega]

Dagu on a Raspberry Pi 2W (512 MB RAM) runs very well and is fast.
I also run Pi-hole & Unbound, Vaultwarden, Gitea, and Homer (lighttpd) — five services in total, each with its own web UI.

Why Dagu is Perfect for Homelab

Problem	Dagu Solution
Manual backups	Schedule automated, multi-step backups with verification
Forgotten updates	Auto-update systems and services safely
No visibility	Web UI shows what ran, what failed, when
Complex workflows	Chain multiple tasks with dependencies
Single points of failure	Retry logic, error alerts via Discord/email
Learning curve	Simple YAML, no complex code needed
Resource-heavy	Lightweight (written in Go), minimal CPU/RAM
Ad-hoc scripts	Centralized, scheduled, monitored execution