Quotas tests fail when run as a subset

[Julie Pichon]

Although the tests pass when run as part of the full openstack_dashboard test suite, the quota tests behave differently when run as a subset and fail.

$ ./run_tests.sh openstack_dashboard.test.tests.quotas
...
======================================================================
ERROR: test_tenant_quota_usages_no_instances_running (openstack_dashboard.test.tests.quotas.QuotaTests)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/jpichon/devel/horizon/openstack_dashboard/test/helpers.py", line 77, in instance_stub_out
    return fn(self)
  File "/home/jpichon/devel/horizon/openstack_dashboard/test/tests/quotas.py", line 136, in test_tenant_quota_usages_no_instances_running
    quota_usages = quotas.tenant_quota_usages(self.request)
  File "/home/jpichon/devel/horizon/horizon/utils/memoized.py", line 33, in __call__
    value = self.func(*args)
  File "/home/jpichon/devel/horizon/openstack_dashboard/usage/quotas.py", line 120, in tenant_quota_usages
    disabled_quotas=disabled_quotas):
  File "/home/jpichon/devel/horizon/openstack_dashboard/usage/quotas.py", line 103, in get_tenant_quota_data
    tenant_id=tenant_id)
  File "/home/jpichon/devel/horizon/openstack_dashboard/usage/quotas.py", line 79, in _get_quota_data
    tenant_id = request.user.tenant_id
  File "/home/jpichon/devel/horizon/.venv/lib/python2.7/site-packages/django/utils/functional.py", line 203, in inner
    return func(self._wrapped, *args)
AttributeError: 'AnonymousUser' object has no attribute 'tenant_id'
...

[Julie Pichon (jpichon)]

It happens for other api_tests as well (e.g. swift_tests, lbaas_tests), different errors relating to the AnonymousUser being returned.

It looks like when running this subset of the tests, openstack_auth.urls doesn't get loaded, therefore the middleware get_user monkey patching doesn't happen and we call directly to django.contrib.auth.middleware later on.

While waiting for bug 1207840 to be resolved, we could force the monkey patching to happen in the helpers.TestCase's setUp() method.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/40656

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/40656
Committed: http://github.com/openstack/horizon/commit/589092c5f21124c7025360b284ed1dbc2a125c8a
Submitter: Jenkins
Branch: master

commit 589092c5f21124c7025360b284ed1dbc2a125c8a
Author: Julie Pichon <email address hidden>
Date: Wed Aug 7 16:39:17 2013 +0100

    Ensure the user monkey patching is done for the API tests

    When running a subset of the tests, the get_user middleware monkey
    patching in openstack_auth.utils did not always happen, causing some
    of the tests to fail due to missing expected properties.

    Fixes bug #1209177

    Change-Id: Ic00b842e740954b0f36353381f511f7ba3eab70b