GitHub user javierherrer added a comment to the discussion: Support for 
authenticating to airflow metadata database like postgres sql with cloud 
specific authentication mechanism

Correct me if I'm wrong. The `_CMD` approach is helpful but has a major 
limitation: the command is executed only once per Python process, and the 
resulting connection string is cached. For long-lived components like the 
scheduler or webserver, this means the token will expire (typically 2–4 hours) 
and DB operations will start failing.

#### Proposal

Introduce native support for dynamic token refresh in the SQLAlchemy engine 
layer or a pluggable “connection provider” interface. This would allow Airflow 
to re-fetch credentials on demand without restarting pods or relying on 
external hacks like PgBouncer.
This feature would enable secure, cloud-native authentication (e.g., Azure 
Managed Identity, AWS IAM) without static passwords and without lifecycle 
issues.

GitHub link: 
https://github.com/apache/airflow/discussions/43837#discussioncomment-14797873

----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]

Reply via email to