I'm trying to enable MLFLOW user based authentication in Local Linux server using python
Able to start the server the mlflow server with the following command :
mlflow server --backend-store-uri sqlite:///mlflow.db --default-artifact-root ./artifacts --host 0.0.0.0
But wasn't able to find any kind of authentication setup available when starting a server
Expectation:
Need a server web interface with Username and Password based authentication
Suggestions and Appropriate solutions are much appreciated
You can use mlflow's basic authentication: Provide a basic_auth.ini
file similar to this
[mlflow]
default_permission = READ
database_uri = sqlite:///basic_auth.db
admin_username = admin
admin_password = SuperSecurePassword
and add --app-name basic-auth
to mlflow
mlflow server --backend-store-uri sqlite:///mlflow.db --default-artifact-root ./artifacts --host 0.0.0.0 --app-name basic-auth
basic_auth.ini
needs to be in mlflow/server/auth/
under the site-packages directory, so for example /usr/local/lib/python3.10/site-packages/mlflow/server/auth/basic_auth.ini
You can find the correct site-packages directory with python -c 'import mlflow;print(mlflow.__path__)'