anyeole opened a new issue, #62325:
URL: https://github.com/apache/airflow/issues/62325
### Apache Airflow Provider(s)
fab
### Versions of Apache Airflow Providers
Apache Airflow version: 3.1.6 Provider version:
apache-airflow-providers-fab==3.3.0 Database backend: MySQL (RDS)
### Apache Airflow version
3.1.6
### Operating System
Debian GNU/Linux 11 (bullseye)
### Deployment
Other Docker-based deployment
### Deployment details
_No response_
### What happened
Description
When hitting the new FastAPI endpoint GET /auth/fab/v1/users to retrieve the
list of users, the server crashes with a 500 Internal Server Error.
The root cause is a mismatch between the legacy FAB database models and the
new FastAPI Pydantic response schemas regarding timezone awareness.
Logs / Exception Trace
log
`INFO: 127.0.0.1:49162 - "GET /auth/fab/v1/users HTTP/1.1" 500 Internal
Server Error
ERROR: Exception in ASGI application
...
File
"/usr/local/lib/python3.10/site-packages/airflow/providers/fab/auth_manager/api_fastapi/services/users.py",
line 91, in get_users
users=[UserResponse.model_validate(u) for u in users],
...
pydantic_core._pydantic_core.ValidationError: 3 validation errors for
UserResponse
last_login
Input should have timezone info [type=timezone_aware,
input_value=datetime.datetime(2026, 2, 22, 10, 13, 13), input_type=datetime]
For further information visit
https://errors.pydantic.dev/2.12/v/timezone_aware
created_on
Input should have timezone info [type=timezone_aware,
input_value=datetime.datetime(2026, 1, 21, 17, 0, 38), input_type=datetime]
For further information visit
https://errors.pydantic.dev/2.12/v/timezone_aware
changed_on
Input should have timezone info [type=timezone_aware,
input_value=datetime.datetime(2026, 1, 21, 17, 0, 38), input_type=datetime]
For further information visit
https://errors.pydantic.dev/2.12/v/timezone_aware`
Root Cause Analysis
1. The Pydantic Model (UserResponse) completely enforces UTC timezone-aware
datetimes. It defines last_login, created_on, and changed_on as UtcDateTime
(from airflow.api_fastapi.common.types).
2. The SQLAlchemy Model (User) defines these properties using standard
sqlalchemy.DateTime
The Clash: When using MySQL or Postgres, SQLAlchemy returns naive Python
datetime.datetime objects. When UserResponse.model_validate(u) executes,
Pydantic immediately rejects these naive datetimes and throws a 500 error.
### What you think should happen instead
_No response_
### How to reproduce
1. Run Airflow 3.1.6 with apache-airflow-providers-fab==3.3.0 backed by
MySQL.
2. Authenticate to obtain a valid JWT/Session.
3. Call GET /auth/fab/v1/users.
Observe the 500 Internal Server Error in the API server logs.
### Anything else
Suggested Fix:
1. Update the FAB UserResponse model to use standard datetime instead of
UtcDateTime
to allow for the naive dates returned by the legacy Flask-AppBuilder
SQLAlchemy models.
### Are you willing to submit PR?
- [x] Yes I am willing to submit a PR!
### Code of Conduct
- [x] I agree to follow this project's [Code of
Conduct](https://github.com/apache/airflow/blob/main/CODE_OF_CONDUCT.md)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
