AIP-89: Privacy-First Telemetry for Apache Airflow

Status: Draft
Authors: [Your Name]
Created: 2025-10-08
Target Airflow Version: 3.2.0
Related Issues: [Link to GitHub issue]

Abstract

This AIP proposes replacing Airflow's current telemetry implementation (Scarf) with a privacy-first, transparent, and community-governed telemetry system that uses Apache Software Foundation's Matomo instance as the default collection endpoint.

Motivation

Currently, Airflow uses Scarf for telemetry collection, which has raised several concerns:

Privacy and GDPR compliance issues, particularly around data collection consent
Limited transparency about what data is collected and how it's used
Lack of user control over what information is shared
Corporate adoption barriers - enterprise users face challenges enabling telemetry due to internal policies
Trust erosion - the opt-out nature of previous implementations damaged community trust

However, telemetry data is crucial for Airflow's development:

Feature prioritization - Understanding how features are used helps prioritize development efforts
Error identification - Error reports help identify and fix issues more quickly
API design decisions - Usage patterns inform API design and deprecation decisions
Performance optimization - Performance metrics guide optimization efforts
Provider maintenance - Understanding which providers are actively used informs maintenance priorities

The current situation has led to lower telemetry adoption, reducing the insights available to maintainers while also potentially creating privacy risks for users who do enable telemetry.

Goals

Privacy by design - Collect only minimal, non-personal data with explicit user consent
Transparency - Users can see exactly what data is collected and sent
User control - Easy opt-in/opt-out with clear visibility into telemetry status
Community governance - Changes to collected data require community approval
Trust rebuilding - Demonstrate Airflow's commitment to user privacy and autonomy

Non-Goals

Collecting personally identifiable information (PII)
Collecting DAG names, task names, or other deployment-specific identifiers
Collecting connection details, credentials, or sensitive configuration
Real-time monitoring or performance profiling of individual deployments
Commercial use of telemetry data

Proposal

Data Collection Endpoint

Default: Apache Software Foundation's Matomo instance (https://xxx)

Hosted and managed by the ASF
Complies with ASF privacy policies
Data is owned by the Apache Airflow project
Users may configure alternative endpoints if required by organizational policies

Data Collected

The following minimal data will be collected when telemetry is enabled:

Installation Metrics

Installation UUID: A randomly generated identifier (UUID4) created at first startup
Airflow version: Full semantic version (e.g., 2.10.0)
Python version: Major and minor version only (e.g., 3.11)
Deployment type: One of docker, kubernetes, systemd, standalone, unknown
Operating system: Generic OS type (e.g., linux, darwin, windows)
Architecture: System architecture (e.g., x86_64, arm64)

Usage Metrics

Active providers: List of installed provider packages and versions (e.g., apache-airflow-providers-amazon==8.0.0)
Executor type: Configured executor (e.g., LocalExecutor, CeleryExecutor, KubernetesExecutor)
Database backend: Database type only (e.g., postgres, mysql, sqlite)
Enabled features: Boolean flags for optional features (e.g., has_webserver_auth, has_dag_serialization)
Operator usage counts: Aggregated counts of operator types used (no task names or parameters)

Aggregated Statistics (collected weekly)

DAG count: Total number of DAGs (integer only)
Task count: Total number of tasks across all DAGs (integer only)
DAG run count: Total number of DAG runs in the past 7 days
Task instance count: Total number of task instances in the past 7 days

Technical Metadata

Timestamp: UTC timestamp of telemetry event
IP address: Used only for geolocation to country level, then immediately discarded (last octet zeroed before logging)

Data NOT Collected

The following data will explicitly not be collected:

DAG names, descriptions, or any DAG content
Task names, parameters, or configurations
Variable names or values
Connection names, URIs, or credentials
Log contents or error messages containing user data
User names, emails, or authentication information
Full IP addresses (only country-level geolocation)
Hostnames or deployment identifiers
Code or custom operator implementations
File paths or directory structures

Consent Mechanism

First-Time Installation

When Airflow is started for the first time (or after upgrade to a version with this AIP):

CLI Installation: A 10-second interactive prompt appears:

╔═══════════════════════════════════════════════════════════════╗
║                  Apache Airflow Telemetry                     ║
╠═══════════════════════════════════════════════════════════════╣
║                                                               ║
║  Airflow would like to collect anonymous usage data to help  ║
║  improve the project. This is entirely optional and can be   ║
║  disabled at any time.                                       ║
║                                                               ║
║  Data collected:                                             ║
║  • Airflow version and Python version                        ║
║  • Installed providers and operators used                    ║
║  • Deployment type and database backend                      ║
║  • Aggregate usage counts (no personal data)                 ║
║                                                               ║
║  Full details: https://airflow.apache.org/docs/telemetry     ║
║                                                               ║
║  Enable telemetry? [Y/n] (auto-decline in 10 seconds)       ║
╚═══════════════════════════════════════════════════════════════╝

User Options:
- Press Y or y: Enable telemetry
- Press N or n: Disable telemetry
- No input within 10 seconds: Default to disabled (not recorded as explicit choice)
Non-interactive Mode: If stdin is not a TTY or --non-interactive flag is present, no prompt appears and telemetry defaults to disabled

Web UI Admin Screen

If no explicit choice has been recorded, the Admin UI displays a banner:

╭─────────────────────────────────────────────────────────────────╮
│ ⓘ  Airflow Telemetry Not Configured                            │
│                                                                 │
│ Help improve Airflow by sharing anonymous usage data.          │
│ [Learn More] [Enable Telemetry] [Disable Telemetry]            │
╰─────────────────────────────────────────────────────────────────╯

Telemetry Settings Page (Admin > Telemetry Settings):

Current status (Enabled/Disabled/Not Configured)
Last transmission timestamp
Summary of last data sent (viewable as JSON)
Enable/Disable toggle
Link to full telemetry documentation
Export of all telemetry data sent (for GDPR compliance)

Configuration

Database Storage

Telemetry preference is stored in the airflow_settings table:

INSERT INTO airflow_settings (key, value) VALUES 
  ('telemetry.enabled', 'true'),
  ('telemetry.installation_uuid', 'a1b2c3d4-e5f6-7890-1234-567890abcdef'),
  ('telemetry.consent_timestamp', '2025-10-08T12:34:56Z'),
  ('telemetry.last_sent_timestamp', '2025-10-08T13:00:00Z');

Configuration File

Users can override database settings via airflow.cfg:

[telemetry]
# Options: enabled, disabled, unset
# unset = defer to database setting (default)
enabled = unset

# Optional: Override collection endpoint
# Default: https://matomo.apache.org/
endpoint = https://matomo.apache.org/

# Optional: Custom installation UUID (for testing)
# installation_uuid = custom-uuid-here

# Collection interval in seconds (default: 86400 = daily)
collection_interval = 86400

Configuration Precedence:

Environment variable AIRFLOW__TELEMETRY__ENABLED
Configuration file [telemetry].enabled
Database setting
Default (disabled)

Data Transmission

Transmission Schedule

Frequency: Once per day (configurable via collection_interval)
Time: Randomized within a 1-hour window to avoid thundering herd
Retry logic: Up to 3 retries with exponential backoff on failure
Timeout: 10-second timeout per request
Graceful degradation: Failures are logged but do not impact Airflow functionality

Transmission Method

Protocol: HTTPS POST to Matomo tracking API
Format: JSON payload
User-Agent: Apache-Airflow/{version} Telemetry/1.0
IP Anonymization: Last octet zeroed before Matomo processing
No cookies: No tracking cookies or persistent identifiers beyond installation UUID

Sample Payload

{
  "telemetry_version": "1.0",
  "timestamp": "2025-10-08T12:00:00Z",
  "installation_uuid": "a1b2c3d4-e5f6-7890-1234-567890abcdef",
  "airflow_version": "3.2.0",
  "python_version": "3.11",
  "deployment_type": "kubernetes",
  "os": "linux",
  "architecture": "x86_64",
  "database_backend": "postgres",
  "executor": "KubernetesExecutor",
  "providers": [
    {"name": "apache-airflow-providers-amazon", "version": "8.0.0"},
    {"name": "apache-airflow-providers-google", "version": "10.0.0"}
  ],
  "enabled_features": {
    "webserver_auth": true,
    "dag_serialization": true
  },
  "usage_stats": {
    "dag_count": 47,
    "task_count": 312,
    "dag_runs_7d": 1840,
    "task_instances_7d": 24576
  },
  "operator_usage": {
    "PythonOperator": 145,
    "BashOperator": 67,
    "S3ToRedshiftOperator": 23
  }
}

Governance and Changes

Process for Modifying Collected Data

Any changes to the data collected (additions or removals) require:

AIP or GitHub Discussion: Proposal explaining the change and justification
Dev List Vote: Lazy consensus vote on dev@airflow.apache.org (72-hour voting period)
Documentation Update: Update telemetry documentation with exact fields
Release Notes: Prominent mention in release notes under "Telemetry Changes" section
In-App Notification: Users with telemetry enabled see a one-time notification in Web UI about data collection changes with option to review and opt-out

Telemetry Schema Versioning

Telemetry payloads include a telemetry_version field
Breaking changes increment the major version
Additive changes increment the minor version
Older Airflow versions continue sending their schema version
Backend supports multiple schema versions simultaneously

Transparency and Data Access

Public Dashboard

A public dashboard will be created showing:

Aggregate statistics (total installations, version distribution)
Provider popularity
Executor type distribution
Database backend distribution
Deployment type breakdown
Geographic distribution (country level only)

Dashboard URL: https://telemetry.airflow.apache.org/ (to be created)

Raw Data Access

Aggregated, anonymized data will be made available as quarterly CSV exports
Individual installation data will never be published
Large deployment users (AWS, Google, Astronomer, etc.) may request access to aggregate insights for comparison purposes

Security Considerations

No Authentication Required: Telemetry endpoint is unauthenticated (prevents tracking via auth tokens)
Rate Limiting: Backend implements rate limiting per installation UUID to prevent abuse
Schema Validation: All payloads are validated against JSON schema before processing
Data Retention: Raw telemetry data retained for 2 years, then deleted; aggregates retained indefinitely
ASF Infrastructure: Hosted on ASF infrastructure with ASF security policies
HTTPS Only: All transmissions over TLS 1.2+
No External Dependencies: Telemetry collection uses only Python standard library (except HTTP client)

Implementation Plan

Phase 1: Core Infrastructure (Airflow 3.2.0-alpha)

[ ] Implement telemetry data collection module
[ ] Add database schema for telemetry settings
[ ] Create CLI prompt for first-time setup
[ ] Implement configuration file parsing
[ ] Add basic transmission logic with Matomo integration
[ ] Create admin UI for telemetry management

Phase 2: Documentation and Transparency (Airflow 3.2.0-beta)

[ ] Complete telemetry documentation page
[ ] Set up public dashboard infrastructure
[ ] Create data export functionality for GDPR compliance
[ ] Add release notes and upgrade guide
[ ] Implement in-app changelog for data collection changes

Phase 3: Testing and Refinement (Airflow 3.2.0-rc)

[ ] Community review period
[ ] Security audit of telemetry implementation
[ ] Performance testing (ensure no impact on Airflow operations)
[ ] Privacy review
[ ] Integration testing with Matomo

Phase 4: Launch (Airflow 3.2.0 GA)

[ ] Enable telemetry system in release
[ ] Launch public dashboard
[ ] Announcement blog post
[ ] Monitor adoption and feedback

Backward Compatibility

Removal of Scarf

Scarf telemetry was removed in Airflow 2.10.x following community feedback
No migration of previous telemetry data to new system
Users who had any previous telemetry enabled will need to explicitly opt-in to new telemetry

Upgrade Experience

Users upgrading from Airflow 3.1.x or earlier to 3.2.0:

On first 3.2.0 startup, see the telemetry consent prompt
Release notes prominently explain new telemetry system
Admin UI shows telemetry banner until explicit choice is made

Testing Strategy

Unit Tests: All telemetry collection, transmission, and configuration logic
Integration Tests: End-to-end telemetry flow with mock Matomo endpoint
Privacy Tests: Verify no PII is collected or transmitted
Security Tests: Attempt to inject malicious data or cause DoS
Performance Tests: Ensure telemetry adds < 1ms to startup time and no runtime impact
User Acceptance Testing: Community beta testing period

Documentation

New documentation will be added:

Main Documentation Page: /docs/telemetry.rst
- What data is collected (comprehensive list)
- How to enable/disable telemetry
- Where data is sent
- How data is used
- Privacy policy
Admin Guide: /docs/administration-and-deployment/telemetry.rst
- Configuration options
- Enterprise deployment considerations
- Troubleshooting
Contributing Guide: /docs/contributing/telemetry-changes.rst
- Process for proposing data collection changes
- Schema versioning guidelines
Release Notes: Dedicated section in every release with telemetry changes

Metrics for Success

Success of this AIP will be measured by:

Adoption Rate: Target 15-20% opt-in rate within 6 months (up from essentially 0% post-2.10 removal)
Community Trust: Reduction in GitHub issues/discussions complaining about telemetry
Data Quality: Sufficient data to answer development prioritization questions
Transparency: Public dashboard launch within 3 months of 3.2.0 release
Enterprise Adoption: At least 2 major cloud providers participate (even if via aggregate data sharing)

Alternatives Considered

Alternative 1: No Telemetry

Rationale for rejection: Development prioritization would rely solely on GitHub issues and surveys, which don't represent actual usage patterns.

Alternative 2: Opt-Out by Default

Rationale for rejection: Violates privacy-first principles and would erode community trust further given the history with Airflow 2.10.

Alternative 3: Self-Hosted Telemetry Endpoint

Rationale for rejection: Users could configure this, but default should be ASF-hosted for trust and convenience.

Alternative 4: Third-Party Service (e.g., PostHog, Segment)

Rationale for rejection: Using ASF infrastructure keeps data under project control and avoids third-party dependencies.

References

Apache Matomo Documentation
GDPR Compliance Guidelines
ASF Privacy Policy
Telemetry Best Practices for Open Source
Previous discussion:

Open Questions

Should telemetry be enabled for dev/test environments by default, or only production?
What is the process for enterprises to share aggregate data without raw telemetry?
Should we offer a "telemetry lite" mode with even less data for privacy-sensitive users?
How do we handle telemetry in CI/CD environments where Airflow is started hundreds of times?

Risks and Mitigations

Risk	Mitigation
Low adoption rate	Provide clear value proposition; show public dashboard early
Performance impact	Thorough testing; async transmission; graceful degradation
Privacy concerns	Privacy-first design; ASF hosting; full transparency
Community backlash	Early community engagement; clear communication; opt-in default
Data not useful	Start with minimal set; iterate based on actual needs
ASF infrastructure unavailable	Configuration allows alternative endpoints

Conclusion

This AIP proposes a privacy-first, transparent, and community-governed telemetry system that rebuilds trust while providing the data Airflow maintainers need to make informed development decisions. By defaulting to opt-in, using ASF infrastructure, and maintaining full transparency, we can achieve the right balance between user privacy and project sustainability.

Discussion:
Vote Thread: TBD after discussion period

Space shortcuts

Page tree