Please https://pytest-django.readthedocs.io/en/latest/database.html#tests-requiring-multiple-databases. Does that work?

@pytest.mark.django_db cannot be used for fixtures, so it does not apply here. The only way to declare that a fixture requires a database is the db fixture, and that does not take any parameters.

I'm also running into this issue.

I was hoping not to auto mark all tests as using a DB and instead let the fixtures pull in a database as needed. This is not something that can be easily configured per fixture, but maybe returning the default set of DBs to pull in via a fixture would allow some customization.

We have another issues. Before our tests were running using test_<db> but now our main database is cleaned/overwritten as we are calling flush and loaddata to pre-populate some static data:

@pytest.fixture(scope='session')
def django_db_setup(django_db_setup: None, django_db_blocker: pytest_django.DjangoDbBlocker, django_db_createdb: bool) -> None:
    del django_db_setup
    if django_db_createdb:
        with django_db_blocker.unblock():
            call_command('flush', '--noinput')
            call_command('loaddata', *pathlib.Path().glob('tests/db_fixtures/**/*.yaml'))

And these commands are executed on <db> database.

Reverting to 4.10.0 fixes the issue.

Please note, that test_<db> is not even getting created...

Reverting to 4.10.0 fixes the issue.

Did you downgrade from 4.11.1?

What were you doing prior?