Improve your Airflow failure emails with simple templates and customizable callbacks—engineer-friendly and easy to extend. GitHub repo

---

🚩 Why Customize Airflow Alerts?

Default Airflow alert emails are basic and lack key context. With minimal setup, you can:

• Improve visual clarity
• Add metadata (DAG tags, retries, links, owners)
• Include recovery tips or failure history

---

🧰 Option 1: Simple Template-Based Alerts

This approach uses Airflow’s built-in support for custom subject and HTML email templates.

Files

• email_body_template.html – styled jinja-powered HTML alert body
• email_subject_template.html – dynamic subject line

Configuration

In airflow.cfg:

[email]
html_content_template = /path/to/email_body_template.html
subject_template = /path/to/email_subject_template.html

Or as environment variables:

AIRFLOW__EMAIL__HTML_CONTENT_TEMPLATE=/path/to/email_body_template.html  
AIRFLOW__EMAIL__SUBJECT_TEMPLATE=/path/to/email_subject_template.html

Benefits

• Clean visual layout
• Rich metadata (task ID, DAG ID, links, retries)
• No code changes

---

⚙️ Output (Basic Template)

On task failure, it uses this subject template:

{% if task_instance.try_number > task.retries %}
🚨 [Airflow Alert] FINAL FAILURE ❌ — {{ dag.dag_id }}.{{ task.task_id }} | Exec: {{ execution_date }}
{% else %}
⚠️ [Airflow Alert] RETRY {{ task_instance.try_number }}/{{ task.retries + 1 }} — {{ dag.dag_id }}.{{ task.task_id }} | Exec: {{ execution_date }}
{% endif %}

---

---

🧠 Option 2: Advanced Alerts with Python Callback

For more control and richer alerts (e.g., run history, recovery tips), use a custom on_failure_callback.

Files

• dags/utils/email_body_template_callback.html – advanced HTML template
• dags/utils/email_callback.py – Python callback that renders and sends the email
• dags/failure_test_dag_callback.py – sample DAG using the callback

Highlights

• Injects DAG/task metadata dynamically
• Includes last 5 DAG run statuses
• Optional recovery guidance
• Flexible to extend (Slack, PagerDuty, etc.)

---

🧪 Output

Alerts look like this when powered by the custom callback:

---

✅ Conclusion

Clear and contextual alerts aren’t just a “nice-to-have” — they’re essential for fast recovery, reliable systems, and a happier on-call experience.

If you’re setting up alerts for an engineering team:

• Start with the simple HTML templates for an instant visual upgrade
• Move to a custom callback when you need logic, historical insight, or extensibility
• Both methods are battle-tested and production-friendly

Want to try it out?

→ Explore the GitHub repo
→ Contributions and suggestions welcome!

---