
GET to list /resource_providers/{uuid}/inventories POST to create a single new inventory via /resource_provider/{uuid}/inventories PUT to (re-)set all inventories at /resource_providers/{uuid}/inventories GET, DELETE or UPDATE one inventory of a particular resource class /resource_provider/{uuid}/inventories/{resource_class} Inventories are presented with a view marker named 'resource_provider_generation' which is used to enable consistent views of inventory across concurrent updates. They are also used in fashion similar to ETags to avoid concurrent updates. ETags were considered for this functionality but discussion amongst interested parties led to the conclusion that the marker was more explicit, especially in the case of taking an inventory representation "seen" in a collection representation, updating it, and sending it back to the server. Change-Id: Ib49c7cddd2f15869f01e8b1e09d8378c26fb7ddc Partially-Implements: blueprint generic-resource-pools
OpenStack Nova Testing Infrastructure
This README file attempts to provide current and prospective contributors with everything they need to know in order to start creating unit tests for nova.
Note: the content for the rest of this file will be added as the work items in the following blueprint are completed: https://blueprints.launchpad.net/nova/+spec/consolidate-testing-infrastructure
Test Types: Unit vs. Functional vs. Integration
TBD
Writing Unit Tests
TBD
Using Fakes
TBD
test.TestCase
The TestCase class from nova.test (generally imported as test) will automatically manage self.stubs using the stubout module and self.mox using the mox module during the setUp step. They will automatically verify and clean up during the tearDown step.
If using test.TestCase, calling the super class setUp is required and calling the super class tearDown is required to be last if tearDown is overridden.
Writing Functional Tests
TBD
Writing Integration Tests
TBD
Tests and Exceptions
A properly written test asserts that particular behavior occurs. This can be a success condition or a failure condition, including an exception. When asserting that a particular exception is raised, the most specific exception possible should be used.
In particular, testing for Exception being raised is almost always a mistake since it will match (almost) every exception, even those unrelated to the exception intended to be tested.
This applies to catching exceptions manually with a try/except block, or using assertRaises().
Example:
self.assertRaises(exception.InstanceNotFound, db.instance_get_by_uuid,
elevated, instance_uuid)
If a stubbed function/method needs a generic exception for testing purposes, test.TestingException is available.
Example:
def stubbed_method(self):
raise test.TestingException()
self.stubs.Set(cls, 'inner_method', stubbed_method)
obj = cls()
self.assertRaises(test.TestingException, obj.outer_method)
Stubbing and Mocking
Whenever possible, tests SHOULD NOT stub and mock out the same function.
If it's unavoidable, tests SHOULD define stubs before mocks since the TestCase cleanup routine will un-mock before un-stubbing. Doing otherwise results in a test that leaks stubbed functions, causing hard-to-debug interference between tests1.
If a mock must take place before a stub, any stubs after the mock call MUST be manually unset using self.cleanUp calls within the test.